Advanced search

Message boards : Graphics cards (GPUs) : New nvidia beta application

Author Message
Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14283 - Posted: 22 Jan 2010 | 18:14:02 UTC

Dear all,
we have developed a new Nvidia application which so far performs 60% faster than the current one. It is stable, so we will release it in the next few days (next week). Today we compiled it also for Windows.
I expect it to become the standard application quite quickly, but for now please accept beta work from us.

GDF

Profile Stoneageman
Avatar
Send message
Joined: 25 May 09
Posts: 224
Credit: 34,057,224,498
RAC: 190
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14284 - Posted: 22 Jan 2010 | 18:55:16 UTC - in response to Message 14283.
Last modified: 22 Jan 2010 | 19:22:26 UTC

Excellent news. That should blow the cobwebs off the server :)

What effect will this have on card temperatures?

Profile K1atOdessa
Send message
Joined: 25 Feb 08
Posts: 249
Credit: 370,320,941
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14292 - Posted: 23 Jan 2010 | 12:47:53 UTC - in response to Message 14283.

Sounds good. Is this 60% improvement estimate an average across numerous card types, or should only a subset of users experience this?

Just for those interested, can you give some detailed on what are the changes -- how you got this improvement, did you take advantage of cuda differently, etc.

Profile [AF>Libristes>Jip] Elgran...
Avatar
Send message
Joined: 16 Jul 08
Posts: 45
Credit: 78,618,001
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14293 - Posted: 23 Jan 2010 | 13:09:07 UTC

I hope it will be a fantastic step to reduce CPU occupation.
I am ready to test it on my Linux computers.

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14294 - Posted: 23 Jan 2010 | 13:42:41 UTC

Does this beta build address issues with cuda/fft bugs? That is should those of us with the 65nm GTX260 try them?
____________
BOINC blog

Profile JockMacMad TSBT
Send message
Joined: 26 Jan 09
Posts: 31
Credit: 3,877,912
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwat
Message 14296 - Posted: 23 Jan 2010 | 19:54:17 UTC
Last modified: 23 Jan 2010 | 19:55:34 UTC

How can recognise if its one of the new units when its downloaded?

And yes will it support GTX260-192 65nm GPUs? I have these sat idle these days when they could be useful so it would be great if they will work.
____________

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14297 - Posted: 23 Jan 2010 | 21:47:28 UTC - in response to Message 14296.
Last modified: 23 Jan 2010 | 21:56:37 UTC

From what was said in other threads,
It will be about 60% better for Compute Capable 1.3 and slightly lower for CC1.1 (but still very significant).

This should facilitate the return of more work units on time, for the lesser cards, and therefore in itself reduce the likelihood of failures over time – hopefully rendering some cards useful once again. You should note that there have also been some other advances that have recently been implemented that seem to improve reliability of the CC1.1 cards.
Only long term testing will determine that for sure, and how well the new application will work for the GTX 260 sp192 cards.
It was successfully compiled for Linux before Windows.

I suggest you enable Beta work units!
The Application should automatically download through Boinc.
I would expect some rise in temperatures for many cards, so keep an eye on that.

Although CUDA bugs are in CUDA, and this is not a re-write of CUDA, but rather the GPUGrid application, I think the bugs crept in due to using far too small time out settings (but I dont know this for sure), so perhaps there is scope for a work around.
Hopefully one of the Guru's will reply here come Monday.

If you think I am wrong about anything, shout up!

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14300 - Posted: 23 Jan 2010 | 22:31:50 UTC - in response to Message 14297.
Last modified: 23 Jan 2010 | 22:32:29 UTC

From what was said in other threads,
It will be about 60% better for Compute Capable 1.3 and slightly lower for CC1.1.


What was said was that the new app is about 60% faster when compiled for 1.3 (double precision) and somewhat less fast when compiled for 1.1 (single precision).

What was not said was which version or versions would actually be released.

I would expect some rise in temperatures for many cards, so keep an eye on that.


There's been nothing from the project saying this. If you have more information, please share it.

GPUGRID currently runs at 77% utilization on my GTX280. Of the projects I currently run, Milkyway@Home has the highest utilization; approximately 90%. The temperature difference between GPUGRID and Milkyway is only 2 or 3 degrees. That with the GPU running some 25 degrees Centigrade below its maximum temperature, with the fan running well below full speed. Even running at 100% utilization, it's unlikely the running temperature would change significantly.

On older architectures (which are less massively parallel than the G200/G200b) the application is more likely to be already be running closer to 100% utilization than it is on the G200s -- it's harder to keep a large number of parallel processors busy than it is to keep a small number of processors busy. So, most likely, the 77% utiization figure I'm seeing on the G200 is close to the lowest number you would see on any GPU. On older cards, there should be less room for improvement simply by increasing the GPU utilization.

The bottom line is that on the G200 based cards, an increase in utilization probably won't raise the temperature a lot, and they have a lot of headroom. On older cards, you're not likely to raise the utilization (since they're probably closer to 100% to start with), so there's unlikely to be any rise in temperature.

And all that, of course, is all making one huge assumption, that the increase in performance is due to increasing the efficiency of the parallelization to increase the GPU utilization. That's not necessarily true. The new application may use normal optimization techniques to be more efficient by performing fewer (or faster) calculations to achieve the same result. This would not increase the operating temperature.

In any event, we'll see soon enough. We can replace the speculation and assumptions with some real observations as soon as the beta apps are released.
____________
Want to find one of the largest known primes? Try PrimeGrid. Or help cure disease at WCG.

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14302 - Posted: 24 Jan 2010 | 0:59:57 UTC - in response to Message 14300.
Last modified: 24 Jan 2010 | 1:02:22 UTC

Michael, you're not seeing much higher temperatures at MW because it uses double precision and GT200 has only 1 dp unit for every 8 normal sp shader units. If GPU Grid went from 77 to 90% utilization the temperatures would surely increase quite a bit.

Edit: that's also why MW is so slow on current nVidia hardware. Only a tiny fraction of the chip can be used at all, whereas ATI can use 40% of their regular shader power.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14303 - Posted: 24 Jan 2010 | 2:16:11 UTC - in response to Message 14302.

Interesting. Thanks for the information.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14306 - Posted: 24 Jan 2010 | 20:29:50 UTC - in response to Message 14303.

Unless the GPUGrid team start writing different applications for different cards, there will only be one release that is capable of increasing performance by around 60% for CC1.3 cards, but which will also increase performance for CC1.1 cards, but by slightly less.

Running an IBUCH task my (40nm) GT 240:
GPU load average is 82%, memory controller utilisation is 42%, memory usage is 171MB and the average GPU temp is 43degrees C (readings taken from GPU-Z).

Compare this to my GTX 260 216sp (55nm):
GPU load average is 80%, memory controller is only 12%, RAM used is 273MB, and the average temperature is 72degrees C.

So a 60% increase in performance cannot come from simply upping the GPU load by 18 to 20%.
It is more likely that it comes primarily from coding optimisations, which may or may not result in a GPU load increase (we will see soon enough). But if it does rise at all there will be an increase in heat.

With CPU projects, the processors can be at 100% for 2 different projects, but the system power consumption can change by up to 10%, depending on the project, and higher Watts means more heat.

GPUGRID Role account
Send message
Joined: 15 Feb 07
Posts: 134
Credit: 1,349,535,983
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 14307 - Posted: 24 Jan 2010 | 21:15:50 UTC - in response to Message 14306.

Hi,

We've improved the application by making several algorithmic and implemention changes, which the effect that it now requires less computation to calculate the same results as before. Just to clarify GDF's statement about the performance: the new app will complete WUs in ~60% the time of the current version.

I'd expect the overall GPU utilisation to remain around the same level as before, maybe a little higher (though I don't know exactly what GPU-Z is measuring to derive that figure). Similarly, I'd not expect any significant change to GPU operating temperatures with the new app.

The difference between 1.1 and 1.3, incidentally, is not an issue of floating point precision (we hardly use d.p. at all) but rather one of compiler efficieny. When we compile for 1.1 -- which is necessary for producing a binary that will run on both G80 and G200 hardware -- the compiler produces slightly less efficient code than if it is targetting 1.3, persumably as there are some G200-specific optimisations it is unable to apply.

Lastly a word about the FFT bug on the 260s: For now we are still using CuFFT but, because of the extensive changes we've made to the code which uses the FFTs, it may be that the problem will no longer occur. Unfortunately we won't know for sure until we have some beta WUs out because we don't have the equipment to reproduce the bug in our lab.

Matt

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14308 - Posted: 25 Jan 2010 | 0:53:34 UTC - in response to Message 14307.

Thanks for clearing that up.

I suspect CPU-Z just looks at wheather the GPU core is busy or not, working or waiting, so it does not reflect how much of the card is being used just if it is being used (but there is no literature on it). It may actually reflect that 18% of the time the CPU is being used! Which going by task manager would be about right.

PS. I thought CC1.0 applied to the now obsolete G80, CC1.1 is G92, 1.3 is G200, and the CC1.2 cards use G215 and G216 cores?

GPUGRID Role account
Send message
Joined: 15 Feb 07
Posts: 134
Credit: 1,349,535,983
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 14312 - Posted: 25 Jan 2010 | 17:39:32 UTC - in response to Message 14308.

I thought CC1.0 applied to the now obsolete G80, CC1.1 is G92, 1.3 is G200, and the CC1.2 cards use G215 and G216 cores


Yes, I was playing fast and loose with terms there. Here's the deal: With the exception of the very first G80 cards (8800GTX 768MB, IIRC), all of the G8x and G9x are compute 1.1 cards. As far as the CUDA programmer's concerned, all of the 1.1 silicon has pretty much the same performance characteristics and capabilities.

The second generation G2xx silicon is rather more capable and has more features that are very useful to us[1], which is why we care to make the distinction for our app. Initially all of those devices described themselves as compute 1.3 but recently some low-mid GPUs have been released that call appear as compute 1.2. In practice, these seem to have the same performance characteristics as the original 1.3 devices, minus double-precision support (we've not had any of these in the lab to test, though so don't quote me).


Matt


[1] more on-chip registers so kernels can be more complex, relaxed memory access rules, better atomic memory ops and double precision, to name the highlights.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14340 - Posted: 26 Jan 2010 | 11:54:45 UTC - in response to Message 14312.

We are now testing the Windows build on our local server. Possibly in the afternoon we will upload some WUs with the new application.

gdf

lorenzo palloni
Send message
Joined: 17 Dec 08
Posts: 3
Credit: 333,482
RAC: 0
Level

Scientific publications
watwatwatwatwat
Message 14351 - Posted: 26 Jan 2010 | 15:21:08 UTC - in response to Message 14340.

Hi, i have nvidia 9800gt 512mb, will it be able to crunch new wu?
(is it cc1.1, cc1.3...?).

One more question: in "Maximum CPU % for graphics 0 ... 100", which parameter i should write? And what does it change by changing this %, in computing the wu?

thank you

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14355 - Posted: 26 Jan 2010 | 16:10:46 UTC - in response to Message 14351.

One more question: in "Maximum CPU % for graphics 0 ... 100", which parameter i should write? And what does it change by changing this %, in computing the wu?


That parameter applies only to projects that have a screensaver (SETI, Einstein, CPDN, Rosetta, Docking, etc.). Also, as far as I know, only CPU-based tasks have screen savers and there are no GPU applications with screensavers.

The parameter says how much of the CPU should be dedicated to running the screensaver graphics (and taking away from the number crunching).

For GPUGRID, this setting has no effect.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14360 - Posted: 26 Jan 2010 | 17:53:52 UTC - in response to Message 14351.

Maximum CPU for graphics should be set to 100%

9800gt 512mb is Compute Capable 1.1
It will take plenty of time to run a task on that card, perhaps a couple of days, but it should work.
Remember to disable use GPU when computer is in use, as this can cause problems with that card.
When the new application is out and any fine tuning performed, your card should operate faster than at present.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14545 - Posted: 28 Jan 2010 | 19:42:55 UTC - in response to Message 14360.

On windows it does not work well. We will come out with Linux at first.
gdf

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14790 - Posted: 29 Jan 2010 | 14:34:22 UTC - in response to Message 14545.

We have uploaded the new application for Windows and Linux with some 50 workunits to test. First come first served.

gdf

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14798 - Posted: 29 Jan 2010 | 17:15:10 UTC - in response to Message 14790.

Does that mean the Windows version now works well?

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14800 - Posted: 29 Jan 2010 | 17:26:56 UTC - in response to Message 14798.

What are these workunits called?

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14801 - Posted: 29 Jan 2010 | 17:29:10 UTC - in response to Message 14800.

We actually uploaded only the Windows one now.
It works well, 50 more WUs uploaded.

gdf

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 14803 - Posted: 29 Jan 2010 | 17:44:04 UTC - in response to Message 14801.

They are called something like L*-TEST.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14805 - Posted: 29 Jan 2010 | 19:11:22 UTC - in response to Message 14803.
Last modified: 29 Jan 2010 | 19:17:40 UTC

Sorry guys. I let my ION try to pick up tasks and it downloaded and spat out 13 tests. Its now set to not pick up any more tasks. Not sure why I could not pick up any tests on my GTS 250 or GTX 260, perhaps just timing.

My GT 240 (G215 core) picked up one task and it worked a bit better. Lasted 5min 34sec.

1792896, 1127796, 29 Jan 2010 17:35:16 UTC, 29 Jan 2010 18:57:03 UTC Completed and validated 334.26 331.80 36.03 48.64

PS the Test Application is called,

ACEMD beta version v6.05 (cuda)

Details:
Name B40-TONI_TEST2901-1-5-RND7016_0
Workunit 1127796
Created 29 Jan 2010 16:28:36 UTC
Sent 29 Jan 2010 17:35:16 UTC
Received 29 Jan 2010 18:57:03 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 55914
Report deadline 3 Feb 2010 17:35:16 UTC
Run time 334.2612
CPU time 331.7985
stderr out

<core_client_version>6.10.18</core_client_version>
<![CDATA[
<stderr_txt>
# There is 1 device supporting CUDA
# Device 0: "GeForce GT 240"
# Clock rate: 1.46 GHz
# Total amount of global memory: 1073741824 bytes
# Number of multiprocessors: 12
# Number of cores: 96
MDIO ERROR: cannot open file "restart.coor"
called boinc_finish

</stderr_txt>
]]>

Validate state Valid
Claimed credit 36.0299189814815
Granted credit 48.640390625
application version ACEMD beta version v6.05 (cuda)


The Ions were all swanMalloc failed.

One of the failed Ion efforts:
Name Lhp2-TONI_TEST2901-0-10-RND3354_0
Workunit 1127831
Created 29 Jan 2010 16:53:30 UTC
Sent 29 Jan 2010 18:56:55 UTC
Received 29 Jan 2010 18:58:22 UTC
Server state Over
Outcome Client error
Client state Compute error
Exit status -40 (0xffffffffffffffd8)
Computer ID 55951
Report deadline 3 Feb 2010 18:56:55 UTC
Run time 9.484798
CPU time 8.704856
stderr out

<core_client_version>6.10.18</core_client_version>
<![CDATA[
<message>
- exit code -40 (0xffffffd8)
</message>
<stderr_txt>
# There is 1 device supporting CUDA
# Device 0: "ION"
# Clock rate: 1.25 GHz
# Total amount of global memory: 268435456 bytes
# Number of multiprocessors: 2
# Number of cores: 16
SWAN: FATAL : swanMalloc failed


</stderr_txt>
]]>

Validate state Invalid
Claimed credit 0.0147670414732421
Granted credit 0
application version ACEMD beta version v6.05 (cuda)

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14807 - Posted: 29 Jan 2010 | 21:12:02 UTC
Last modified: 29 Jan 2010 | 21:40:44 UTC

L15-TONI_TEST2901-0-10-RND2988 (state: In progress)

i7-860 / GTX295 / Vista 64



Elapsed Time (wall clock time: 3:43:58)

CPU Time: 3:42:32 !

CPU Eff.: 99,365% !



EDIT: completed

L15-TONI_TEST2901-0-10-RND2988_0
Workunit 1127719
Created 29 Jan 2010 16:04:07 UTC
Sent 29 Jan 2010 16:08:46 UTC
Received 29 Jan 2010 21:32:36 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 53295
Report deadline 3 Feb 2010 16:08:46 UTC
Run time 16040.183401
CPU time 15938.39
stderr out

<core_client_version>6.10.17</core_client_version>
<![CDATA[
<stderr_txt>
# There are 2 devices supporting CUDA
# Device 0: "GeForce GTX 295"
# Clock rate: 1.51 GHz
# Total amount of global memory: 939524096 bytes
# Number of multiprocessors: 30
# Number of cores: 240
# Device 1: "GeForce GTX 295"
# Clock rate: 1.51 GHz
# Total amount of global memory: 939524096 bytes
# Number of multiprocessors: 30
# Number of cores: 240
MDIO ERROR: cannot open file "restart.coor"
# Time per step: 25.659 ms
# Approximate elapsed time for entire WU: 16036.845 s
called boinc_finish

</stderr_txt>
]]>

Validate state Valid
Claimed credit 4503.73958333333
Granted credit 6080.0484375
application version ACEMD beta version v6.05 (cuda)

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14810 - Posted: 29 Jan 2010 | 22:46:08 UTC - in response to Message 14807.

4 more tasks completed and validated:

1. 1793522 B32-TONI_TEST2901-1-5-RND4868_0 203.16 199.92 36.03 48.64
2. 1793556 B31-TONI_TEST2901-1-5-RND1267_0 326.62 320.47 36.03 48.64
3. 1793228 B39-TONI_TEST2901-1-5-RND7704_0 311.53 301.89 36.03 48.64
4. 1792896 B40-TONI_TEST2901-1-5-RND7016_0 334.26 331.80 36.03 48.64

Systems:
1. GTX 260 sp216 (55nm) driver 19038 VistaU64bit Phenom II 940 4GB
2&3. GeForce 8800 GTS 512 Driver 19038 W7Pro 64bit Q6600 4GB
4. GTS 250 (Factory OC) driver 19562 W7Pro 64bit Q9400(OC) 4GB


One Beta task still Running after 43min on Q9400 system,

L6-TONI_TEST2901-0-10-RND2222

Profile Stoneageman
Avatar
Send message
Joined: 25 May 09
Posts: 224
Credit: 34,057,224,498
RAC: 190
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14811 - Posted: 29 Jan 2010 | 23:12:43 UTC - in response to Message 14807.

gtx260 OC 696/1500/999 Boinc 6:10:25 XP64

All four Bxx - TONI units completed successfully after only two minutes.
One Lxx - TONI unit has just finished normally after only 4h:18m. Excellent performance increase.
Temperatures look normal, however the cpu utilisation of 100% is going to be an issue for me and others running cpu projects alongside GPUgrid.



Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14812 - Posted: 29 Jan 2010 | 23:16:09 UTC - in response to Message 14811.

the cpu utilization should have remained the same. We will look into it.
gdf

Profile Aardvark
Avatar
Send message
Joined: 27 Nov 08
Posts: 28
Credit: 82,362,324
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14816 - Posted: 30 Jan 2010 | 1:33:28 UTC - in response to Message 14812.
Last modified: 30 Jan 2010 | 2:06:33 UTC

I have two of these WU's running on my Windows 7 64 bit Pro', i7 920 with two GTX260 (216 core 55nm) GPU's. After they were about 66% completed I noticed that they were each using 100% of one of the eight multi-threaded cores. I closed down Boinc manager and switched off multithreading. On restarting Boinc manager, each WU was now running 100% of one of the four,non multithreaded cores. The elapsed timer on Boinc manager started again at the point at which it had previously stopped (3.5 hours). The time to completion looks to be about the same as when previously using hyperthreading (approximately 5 hours). I will forward details when completed.

Profile Aardvark
Avatar
Send message
Joined: 27 Nov 08
Posts: 28
Credit: 82,362,324
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14817 - Posted: 30 Jan 2010 | 1:52:50 UTC - in response to Message 14816.
Last modified: 30 Jan 2010 | 2:08:17 UTC

The result of one of the WU's which completed after approximately five hours and seven minutes. The other is similar. Temperatures normal.

Name L18-TONI_TEST2901-0-10-RND0675_0
Workunit 1127722
Created 29 Jan 2010 16:04:17 UTC
Sent 29 Jan 2010 20:14:35 UTC
Received 30 Jan 2010 1:47:32 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 55059
Report deadline 3 Feb 2010 20:14:35 UTC
Run time 18451.701774
CPU time 18065.93
stderr out <core_client_version>6.10.25</core_client_version>
<![CDATA[
<stderr_txt>
# There are 2 devices supporting CUDA
# Device 0: "GeForce GTX 260"
# Clock rate: 1.47 GHz
# Total amount of global memory: 939524096 bytes
# Number of multiprocessors: 27
# Number of cores: 216
# Device 1: "GeForce GTX 260"
# Clock rate: 1.47 GHz
# Total amount of global memory: 939524096 bytes
# Number of multiprocessors: 27
# Number of cores: 216
MDIO ERROR: cannot open file "restart.coor"
# There are 2 devices supporting CUDA
# Device 0: "GeForce GTX 260"
# Clock rate: 1.47 GHz
# Total amount of global memory: 939524096 bytes
# Number of multiprocessors: 27
# Number of cores: 216
# Device 1: "GeForce GTX 260"
# Clock rate: 1.47 GHz
# Total amount of global memory: 939524096 bytes
# Number of multiprocessors: 27
# Number of cores: 216
called boinc_finish

</stderr_txt>
]]>


Validate state Valid
Claimed credit 4503.73958333333
Granted credit 6080.0484375
application version ACEMD beta version v6.05 (cuda)

--------------------------------------------------------------------------------

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14820 - Posted: 30 Jan 2010 | 7:01:02 UTC

I picked up 2 so far. Links are here and here

They only ran for a couple of minutes but finished successfully. They were run on a GTX295.
____________
BOINC blog

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14822 - Posted: 30 Jan 2010 | 9:34:54 UTC - in response to Message 14820.

L6-TONI_TEST2901-0-10-RND2222_2

http://www.gpugrid.net/result.php?resultid=1793679
Completed and validated 37,082.82 35,765.96 4,503.74 6,080.05

The result mirrors Stoneageman's result in that CPU was used almost 100%. Note I am crunching on WCG at the same time (not sure if that is being picked up rather than actual GPUGrid CPU time used)!
That said, great improvement. The GPU is GTS250 and CC1.1 and tasks normally take from 53000s to 60000s. So completion time dropped from 100% to less than 70%. If you take the normal 60000s the WU was 60% faster.

I have one more long Beta task running on a GT 240 CC1.1 - should finish late tonight.

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14825 - Posted: 30 Jan 2010 | 11:31:10 UTC - in response to Message 14822.

GTX285, XP32/ i7-920 HT ON @4.0 GHz.
100% cpu utilization and I did see the estimated 60% runtime improvement.

GTX295 Vista 64 Ultimate/ i7-920 HT OFF @ 4.4 GHz.

On my 295 when I did not suspend other projects to make a full core available for each GPUGrid beta (meaning they had to share) they ran 7.5 hours which is not substantively better than normal, certainly not in the 60% range. I did also get a couple of the real shorties and their CPU utilization was also close to 100%.
It looks like on the i7 there may be an issue with how the processor is handling the instruction set if HT is OFF. Maybe internally the CPU is turning HT on??? Observing task manager if I made sure 2 cores were free then each of the two WUs would use 25% but when I ran 4 CPU WUs from other projects the GPUGrid betas were using 11-13% which is just what a CPU WU looks like when I have HT on. The 7.5 hour runtimes I mentioned above were processed with HT off.


____________
Thanks - Steve

Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14826 - Posted: 30 Jan 2010 | 12:25:18 UTC

Kaboom, in the middle of the night.

Name L38-TONI_TEST2901-0-10-RND1295_0
Workunit 1127742
Created 29 Jan 2010 16:05:13 UTC
Sent 29 Jan 2010 16:13:31 UTC
Received 30 Jan 2010 12:20:21 UTC
Server state Over
Outcome Client error
Client state Compute error
Exit status 3 (0x3)
Computer ID 6133
Report deadline 3 Feb 2010 16:13:31 UTC
Run time 10312.796875
CPU time 5985.359
stderr out <core_client_version>6.10.17</core_client_version>
<![CDATA[
<message>
The system cannot find the path specified. (0x3) - exit code 3 (0x3)
</message>
<stderr_txt>
# There is 1 device supporting CUDA
# Device 0: "GeForce 8800 GT"
# Clock rate: 1.62 GHz
# Total amount of global memory: 536543232 bytes
# Number of multiprocessors: 14
# Number of cores: 112
MDIO ERROR: cannot open file "restart.coor"
SWAN : FATAL : Failure executing kernel sync [frc_sum_kernel] [700]
Assertion failed: 0, file ../swan/swanlib_nv.cpp, line 203

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

</stderr_txt>
]]>


Validate state Invalid
Claimed credit 4503.73958333333
Granted credit 0
application version ACEMD beta version v6.05 (cuda)

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14827 - Posted: 30 Jan 2010 | 13:26:51 UTC
Last modified: 30 Jan 2010 | 13:40:40 UTC

9800 GT, 607MHz/1517MHz/900MHz (512MB) driver: 19062 / QX9650 @3.66GHz / Vista64

B28-TONI_TEST2901-0-5-RND4647_0
Run time 357.2244
CPU time 348.1474
Validate state Valid

B47-TONI_TEST2901-0-5-RND7649_1
Run time 357.2556
CPU time 349.2238
Validate state Valid

B42-TONI_TEST2901-0-5-RND0199_2
Run time 356.179199
CPU time 348.943
Validate state Valid


GTX 295, 701MHz/1509MHz/1086MHz (896MB) driver: 19062 / i7-860 HT @3.8 GHz/ Vista64

L14-TONI_TEST2901-0-10-RND7593_0
Run time 15980.329535
CPU time 15870.54
Validate state Valid

L15-TONI_TEST2901-0-10-RND2988_0
Run time 16040.183401
CPU time 15938.39
Validate state Valid

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14828 - Posted: 30 Jan 2010 | 13:33:25 UTC - in response to Message 14826.

L45-TONI_TEST2901-0-10-RND5880_0 is 43% complete after 5h, so should complete in around 6h (though the estimate time to finish is 12h). On a 2.2GHz opteron Quad with a GT240. On that system typical task turnaround is about 17 or 18h. So it is preforming about 60% faster on that CC1.2 card.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14832 - Posted: 30 Jan 2010 | 15:23:31 UTC - in response to Message 14811.

gtx260 OC 696/1500/999 Boinc 6:10:25 XP64

the cpu utilisation of 100% is going to be an issue for me and others running cpu projects alongside GPUgrid.

8800GT, BOINC v6.10.29, Win7-64, AMD X2

Running with 2 instances of Wieferich@home, beta runs a bit slower than old app, machine is barely responsive, 1 instance of Wieferich stalls. If I free up one complete core (close 1 instance of Wieferich) beta runs faster and machine becomes responsive. With the old app everything ran fine with an instance of Wieferich on each core. The 100% CPU core utilization of the beta is a problem...

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14833 - Posted: 30 Jan 2010 | 15:46:31 UTC

GTX 295, 701MHz/1509MHz/1086MHz (896MB) driver: 19062 / i7-860 HT @3.8 GHz/ Vista64

L13-TONI_TEST2901-1-10-RND4450_0

Aborted by myself - Run time 2726: 100% CPU-load, no GPU-utilization shown by GPU-Z, no progress.

(Dump by BOINC Windows Runtime Debugger)

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14834 - Posted: 30 Jan 2010 | 15:48:13 UTC
Last modified: 30 Jan 2010 | 15:55:24 UTC

9800GTX Stock no o/c Phenom2 o/c 3.2Ghz 8Gb RAM Vista64

Also running Aqua and Freehal concurrently.

L36_TONI_TEST2901-1-10-RND9113_0

Saw the Beta running, so came back to give it a go, I had been blown away by too many errors on the old apps. Grabbed two about 20 mins ago, early days but the speed is impressive.

CPU utilisation is out of wack as noted by others, its using a complete core on the Phenom2 Quad (Task Manager reports cpu utilisation as "23". That needs addressing as its a showstopper because its draining from cpu based Projects (in my case Aqua - has no effect on Freehal).

It managed to muscle out Aqua from one core - quite a feat as the Aqua app is somewhat territorial grabbing all cores usually kicking off others :) Temps are ok.

If the cpu utilisation is resolved its looking like I'll be back again, which is a relief - I have serious GPUGRID withdrawal symptoms :)

Regards
Zy

Profile robertmiles
Send message
Joined: 16 Apr 09
Posts: 503
Credit: 727,920,933
RAC: 155,858
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14835 - Posted: 30 Jan 2010 | 16:27:49 UTC

Looks like this ACEMD beta 6.05 workunit has a severe underestimate of how much CPU time it uses:

1/30/2010 5:28:36 AM GPUGRID Starting L25-TONI_TEST2901-0-10-RND7574_3
1/30/2010 5:28:36 AM GPUGRID Starting task L25-TONI_TEST2901-0-10-RND7574_3 using acemdbeta version 605

It claims to be using 0.18 CPUs + 1.00 NVIDIA GPUs, but it's actually making full use of the CPU core.

Suspending the workunit, then resuming, does not help.

BOINC 6.10.18
64-bit Windows Vista SP2

Processor: 4 GenuineIntel Intel(R) Core(TM)2 Quad CPU Q9650 @ 3.00GHz [Intel64 Family 6 Model 23 Stepping 10]

NVIDIA GPU 0: GeForce 9800 GT (driver version 19038, CUDA version 2030, compute capability 1.1, 1024MB, 336 GFLOPS peak)

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14837 - Posted: 30 Jan 2010 | 19:18:30 UTC - in response to Message 14835.
Last modified: 30 Jan 2010 | 19:31:39 UTC

I have just uploaded acemdbeta6.06 that should use less CPU as before.
It would be interesting to see how much slower it is.
If it is a lot slower what if we also credit for the usage of the CPU? In such case, we would ask for 1 CPU and 1 GPU of course.

Let me know if there are problems, as I cannot test it from home.

gdf

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14838 - Posted: 30 Jan 2010 | 20:23:34 UTC - in response to Message 14837.

L36_TONI_TEST2901-1-10-RND9113_0

Been running for 4hrs 55 mins. The last 50 mins the counter has been at zero.

Abort or keep it going ?

Its on a 9800GTX

Regards
Zy

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14839 - Posted: 30 Jan 2010 | 20:43:24 UTC - in response to Message 14837.

The earlier Beta, L45-TONI_TEST2901-0-10-RND5880_0 completed as expected in 11h 34m.

Downloaded several new Beta tasks including,
L49-TONI_TEST2901-1-10-RND8544_0 (GT 240)
B5-TONI_TEST2901-3-5-RND7970_0 (GTX 260 sp216)
1796887...

Some have already been reported back:

1796949 1130301 30 Jan 2010 20:11:34 UTC 30 Jan 2010 20:28:19 UTC Completed and validated 176.94 34.37 36.03 48.64

Looks like the CPU time may have been sorted!



Thanks,

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14840 - Posted: 30 Jan 2010 | 20:55:43 UTC - in response to Message 14839.

The shorter tasks are all reported.

Now running 3 longer tasks.

Each TEST Tasks using less than 10% CPU of a quad core - so under 40% of one Quad's core.

Please note that no tasks show a rise in % complete, as reported by Zydor.

I will allow them to run, unless advised otherwise.

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14841 - Posted: 30 Jan 2010 | 21:41:18 UTC - in response to Message 14837.
Last modified: 30 Jan 2010 | 22:30:59 UTC

............ If it is a lot slower what if we also credit for the usage of the CPU? In such case, we would ask for 1 CPU and 1 GPU of course.


That could get expensive in pure credit terms, and therefore go headlong into a lot of controversy. For that to work, the cpu element would need to compensate for the loss of one core at the highest rate given by Projects for cpu. Arguably thats Aqua, and currently their high end app on my phenom2 runs for 32 hrs at 32,000 credits - an i7 does it in around a quarter of that time.

The problem comes when lower rated machine run GPUGRID capable cards. A fixed rate would bring a howl of protest as the lower capacity machine gets the higher rate for cpu. I've used extremes to illustrate the issue. As Aqua found out to its cost a few months ago, its a foolish person who ignores this kind of credit issue. Whatever anyone thinks of credits and the reality that they are worthless, the S%$T storm a high fixed rate will cause is going to slam the Project into the deck.

Therefore to get out of that one, the cpu element would need to be calculated using the classic BOINC sliding scale and the whole crazy world of cobblestones, benchmarks and built in elements to the app to make it work.

You take the view that going for a mid level fixed cpu rate would be "fair", you will just get it in the neck both ways "Low end machines get too much" or "high end machines dont get enough". Yup its a silly nightmare, but it will happen, dont even dream you'll avoid it, because you will not. The only way round it all is the classic sliding scale based on the "power" of the machine.

However do the latter, and GPU based Projects will scream blue murder that a GPU Project is giving out far too many credits.......

Theanswer? I have no idea, it will depend how much Flak you are prepared to take ....

Turn left your dead, turn right your dead, go straight ahead your probably dead .... :)

Aint BOINC fun :)

Regards
Zy

Profile [AF>Libristes] Dudumomo
Send message
Joined: 30 Jan 09
Posts: 45
Credit: 425,620,748
RAC: 0
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14842 - Posted: 30 Jan 2010 | 21:46:09 UTC - in response to Message 14837.
Last modified: 30 Jan 2010 | 21:58:22 UTC

I have just uploaded acemdbeta6.06 that should use less CPU as before.
It would be interesting to see how much slower it is.
If it is a lot slower what if we also credit for the usage of the CPU? In such case, we would ask for 1 CPU and 1 GPU of course.

Let me know if there are problems, as I cannot test it from home.

gdf


Thanls gdf and others for what you're doing.
Here is my POV
Usually, a lot of people really like using their GPU on one specific project and their CPU on another. (I do).
I don't really like seeing 100% of CPU usage for feeding a GPU.
Well if it is really slower and then don't really have the choice, then I will accept that. However, crediting CPU usage is a fair way but I dislike this way as 1k/d is almost nothing on this project but so much on others. Usually when a project has a GPU app, I'm stopping my CPUs on these projects and put them in others.

Anyway I cannot compare that as I didn't receive any beta WU on my gentoobox (GTX275)

EDIT :
By the way, with this new app, I guess task are running faster and we will get the same amount of point, then I deduce we will have a higher RAC ? right ?
I'm not sure rather it is the good topic or not, but even if I have to admit I'm crunching with GPU, a bit for the points (Not solely otherwise I would have been on Collatz now, which I did anyway, but I guess I won't), I'm not sure if it is really correct to do so...
It will be very easy for a new project to play with the credit to attract ppl with an easy method : Release first a "crappy" GPU app, get a fair credit/WU and then improve this new app and them increase the global RAC. (if the new app is 10 times faster than the first one...then 10 times more credits...well..this is not good for BOINC I think). I really don't mean you are doing that ! (Not at all !!) so in fact because I know that, I prefer explain here this opinion. May be GPUGrid can once again lead the way xD.
A fair credit system will be not to increase (in my example) 10 times the credit, but far less, just to continue to be consistent.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14843 - Posted: 30 Jan 2010 | 22:11:21 UTC - in response to Message 14841.
Last modified: 30 Jan 2010 | 22:29:47 UTC

My long Beta tasks stayed at Zero % for a while, but one is now at 9%, the other 2 are still at Zero and.

O/T
You are going off topic, but good argument Zydor and well said.
Frankly I agree. However, and for the sake of playing devil’s advocate, I will try to pose a counter argument:
The contribution of a CPU compared to a GPU is insignificant. So it does not really matter how much CPU time is spent or what the power of the CPU is. The CPU performance in the very top multi socket systems is not even as powerful as a GT9800, hence the GPUGrid’s CPU project was suspended until significant improvements can be made on the computational power of CPU’s.
If a CPU is used to facilitate a 60% increase in performance of a GPU it is well worth the sacrifice even without any CPU points.

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14845 - Posted: 30 Jan 2010 | 23:05:54 UTC - in response to Message 14843.
Last modified: 30 Jan 2010 | 23:45:23 UTC

Anyone done a L36_TONI_TEST2901 app? Mine been running for 7hrs 36mins countdown clock not working, showing 0% for ages - have a feeling its a bust but dont want to sell it short if in fact its likely to run through.

Its running on a 9800GTX

EDIT
Must be a bust - aborted at 8hrs 15mins

Regards
Zy

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14846 - Posted: 31 Jan 2010 | 0:07:21 UTC - in response to Message 14845.

Just started L42-TONI_TEST2901-3-10-RND1170_0 acemdbeta version 606 on a 9800GTX, it'll run overnight

CPU utilisation is reported as "2" in Task manager - which is fine, so looks like that problem goes away. Too early to say about likely total completion time.

Countdown percentage clock not working

Regards
Zy

Profile liveonc
Avatar
Send message
Joined: 1 Jan 10
Posts: 292
Credit: 41,567,650
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwat
Message 14847 - Posted: 31 Jan 2010 | 0:08:55 UTC - in response to Message 14845.

All my Beta tests show 0% until they're finished. I'm going through my 4th & 5th ones if I'm not mistaken.
____________

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14849 - Posted: 31 Jan 2010 | 0:48:53 UTC - in response to Message 14843.

If a CPU is used to facilitate a 60% increase in performance of a GPU it is well worth the sacrifice even without any CPU points.


That is quite logical. But some people will value their cpu projects much higher than GPU-Grid and are thus not fine with that sacrifice.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14850 - Posted: 31 Jan 2010 | 0:58:02 UTC

Just had a 6.05 WU that refused to start. Instead the machine downloaded a new WU and left the 6.05 WU sitting "ready to start". I aborted it and it was immediately sent back out as a 6.06 WU. Hopefully 6.06 is better. IMO the old app (6.71) was far preferable to 6.05. The CPU usage really needs to be improved before this gets released. Think what will happen when the beta in it's current form hits a single core machine.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14851 - Posted: 31 Jan 2010 | 1:01:14 UTC - in response to Message 14849.

That is quite logical. But some people will value their cpu projects much higher than GPU-Grid and are thus not fine with that sacrifice.

MrS

I agree ETA. The high CPU usage is a show stopper IMO.

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14852 - Posted: 31 Jan 2010 | 3:05:24 UTC - in response to Message 14851.

Let me get this right ... a GIANNI_BIND use to take 8 hours on my GPU and I get 6118 points ... times 3 per day and I get 18354. Now if I throw in a CPU I can run the same WU in 4.8 hours. So that means I can do 5 WUs per day for 30590 points and you guys are bi^&chin about using 1 CPU core to get the additional 12236???
>10000 is too many points for a CPU per day as it is.
While this conversation is important and we need GPUGrid to be clear on what they are doing and how they are determining how many points a WU is worth, unless they are way off the mark then the ony people who are going to leave because they don't think they are getting enough are the point whores ... yes, I said it! And they will leave anyway when it comes time to hop on a project that has crazy points simply to get attention, not because they have a great piece of science or optimized software.
GDF and team ... from my perspective, I really only care that my system remain in some reasonable state of usefulness while running your WUs. If it take 1 GPU + 1 CPU that's OK by me but I think you should use the CPU as a set resource (like Einstein does) instead of stealing cycles from other processes.
____________
Thanks - Steve

Profile liveonc
Avatar
Send message
Joined: 1 Jan 10
Posts: 292
Credit: 41,567,650
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwat
Message 14854 - Posted: 31 Jan 2010 | 8:12:06 UTC

Wow, I'm on almost 20 hours with L6-TONI_TEST2901-1-10-RND2222_0 & 18 hours with L13-TONI_TEST2901-1-10-RND4450_1

It's as if it went dormate, b4 BoincTasks activated it.

Is someone being funny?
____________

Profile Stoneageman
Avatar
Send message
Joined: 25 May 09
Posts: 224
Credit: 34,057,224,498
RAC: 190
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14855 - Posted: 31 Jan 2010 | 8:50:19 UTC

Mixed results with 6.06. Had two long units complete with the 60% improvement in performance and even less cpu utilisation than a 6.71 unit. Both were awarded 6000 or so points.
However, another 6.06 long unit L37 is performing poorly, as the poster LIVEONC above states. Temperature was as to be expected but it was going to take 21hrs to complete, so I aborted it after 4hrs.
The successful ones were on 260's while the slow one was on a 275, though I doubt that is relevant.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14857 - Posted: 31 Jan 2010 | 11:46:09 UTC - in response to Message 14855.
Last modified: 31 Jan 2010 | 11:49:41 UTC

There were 2 sets of Betas released containing short Betas and long Betas. The short Betas lasted about 5 minutes. CPU time usage was fixed for the second release of the long Betas going by the posts.

My long Betas lasted many hours,


    37,082.82s on a GTS 250 (10 ½h) CC1.1
    29,239.43s on a GTX 260 216sp (8 ½h) CC1.3
    45,470.80 on a GT 240 (12 ½h) CC1.2
    41,658.85 on a GT 240 (11 ½h) CC1.2


Each of my systems run CPU tasks for other Boinc projects, which slows down the GPUGrid tasks somewhat. I have one tasks still running on a 8800 GTS 512 (CC1.1), and still showing 0% complete, but the system was off overnight and I would not expect it to complete under 12h. Anyone with a GT9800 or GT8800 should expect these Betas to last longer than 12h! The estimated times to completion were higher than observed finish times for each of my tasks. For the tasks that did report a % complete, I could accurately work out how long the tasks would actually taks.

Keith
Send message
Joined: 2 Feb 09
Posts: 4
Credit: 13,776,401
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwat
Message 14858 - Posted: 31 Jan 2010 | 11:52:49 UTC

Hi
If you want to run CPU projects then run them.
If you want to run GPU projects then run them.
If runing GPU efficiently involves using 100% CPU then do it.
The projects come first and the points are a separate issue.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14860 - Posted: 31 Jan 2010 | 12:54:07 UTC - in response to Message 14858.

The application with low cpu usage works well, so no problem.

gdf

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14861 - Posted: 31 Jan 2010 | 13:16:18 UTC - in response to Message 14860.

GDF ... you and your team are the best !!!
____________
Thanks - Steve

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14865 - Posted: 31 Jan 2010 | 19:29:31 UTC - in response to Message 14861.
Last modified: 31 Jan 2010 | 19:29:44 UTC

Could you guys check if 6.06 is slower? For some reason the appox elapsed time does not get printed.

gdf

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14866 - Posted: 31 Jan 2010 | 20:05:26 UTC - in response to Message 14865.
Last modified: 31 Jan 2010 | 20:10:06 UTC

OK, I had this all nice but then lost it all as I had timed out :-(

While my GTX285 only provides a small sample, the differences between runs of the same version were very minor so I think the following may be helpful

This GTX285 normally processes a GIANNI_BIND WU in 21600 seconds.

Version 6.05
(over three WUs with less than .05% between each of them)
runtime was 13172 seconds.

Version 6.06
(over two WUs with less than .07% between each of them)
runtime was 14006 seconds.

The difference between 6.05 and 6.06 was -6% which really is quite minor in comparison to the overall estimate you provided of 60% reduction originally.

To round out the numbers, the average CPU time on the 6.06 was 680 seconds which is about half of what the old v6.71 uses for a GIANNI_BIND. The only similar CPU usage on 6.71 is the KASHIF_HIVPR which on average come in at about 550 seconds of CPU usage.
____________
Thanks - Steve

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14867 - Posted: 31 Jan 2010 | 20:08:42 UTC - in response to Message 14865.

Any guesstimates available for how long a L42-TONI_TEST2901-3-10-RND1170_0 acemdbeta version 606 on a 9800GTX would run? Its been going for 19hrs 26mins so far

Regards
Zy

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14868 - Posted: 31 Jan 2010 | 20:13:28 UTC - in response to Message 14852.

... and you guys are bi^&chin about using 1 CPU core to get the additional 12236???


Not me, I'd take the increase in GPU speed any time. But we've had this discussion before in the pioneer time of GPU-Grid when they still needed to figure out how to use less than 1 core. From that I know the CPU power is important to many participants.

If this proves difficult to fix: what if there was a user preference for

- maximum GPU performance, use 1 CPU
- use minor amount of CPU and crunch what you can the GPU

MrS
____________
Scanning for our furry friends since Jan 2002

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14869 - Posted: 31 Jan 2010 | 20:16:34 UTC - in response to Message 14868.

ETA ... read up a few posts ... GDF fixed it !!!
CPU usage is back down very low again. While at this moment it appears at the expense of a few percentage points of the overall improvement (so we might be looking at 55% reduction instead of 60%), I think this is excellent news and look forward to seeing 6.06 in production.
____________
Thanks - Steve

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14870 - Posted: 31 Jan 2010 | 20:21:42 UTC - in response to Message 14867.

Zydor ... I find that the beta is taking about 65% of the time it takes me to process a GIANNI_BIND. Not sure what a 9800 takes but there have not be reports of these betas crashing (in fact they appear to be remarkably stable). I would suggest that if it is version 6.05 that you abort because we have already moved on to 6.06 which uses much less CPU time. If you want the points just let it run.
____________
Thanks - Steve

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14871 - Posted: 31 Jan 2010 | 20:32:39 UTC - in response to Message 14870.

This one ran for 28 hours on a 8800GT with v6.05, went to 100% and then was back to 17% when I aborted it. CPU time was also > 26 hours:

http://www.gpugrid.net/workunit.php?wuid=1129962

So far haven't had any long v6.06 WUs. As you can see this client was very reliable with the v6.71 client:

http://www.gpugrid.net/results.php?hostid=55324

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14872 - Posted: 31 Jan 2010 | 20:33:06 UTC - in response to Message 14865.

Could you guys check if 6.06 is slower? For some reason the appox elapsed time does not get printed.

gdf



One is running on one of my 9800GT.
There is no progress shown, so I have to waite for completition.

But the GPU-load looks very good (low CPU-load).

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14873 - Posted: 31 Jan 2010 | 20:50:48 UTC - in response to Message 14872.
Last modified: 31 Jan 2010 | 21:02:32 UTC

You have to compare speed between 6.05 and 6.06 which are all run on a TONI- workunit (there are two types, one short and one long). What I would like to know is the speed on a fully CPU loaded host, with all cores busy.

AS far as I know the progress bar should work. Does it?

gdf

Profile leprechaun
Send message
Joined: 22 Jun 09
Posts: 8
Credit: 45,224,378
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 14874 - Posted: 31 Jan 2010 | 20:58:03 UTC
Last modified: 31 Jan 2010 | 20:58:37 UTC

I have a L7-TONI_TEST WU here. (6.06)
To see after nearly 7 hours no progress bar. Cancel?

GPU: GTX260

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14875 - Posted: 31 Jan 2010 | 21:18:06 UTC - in response to Message 14874.

I think so.

gdf

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14876 - Posted: 31 Jan 2010 | 21:21:39 UTC - in response to Message 14873.
Last modified: 31 Jan 2010 | 21:50:40 UTC

Running L42-TONI_TEST2901-3-10-RND1170_0 acemdbeta version 606 on a 9800GTX - no progress bar or % counter. Its up to 19hrs 55mins run time at present.

Four cores loaded with Aqua, and 25 x Freehal WUs also running concurrently.

EDIT:
With the speed improvements, this should have completed by now - Aborted at 20hrs 17mins

One odd thing I have noted is that all WUs irrespective of type, length Beta or current Production are coming at the same predicted time of 03:54:45. The predicted time will vary machine to machine of course - vageries of BOINC - but its strange all GPUGRID WUs irrespective of type have the same initial "time to completion". That time to completion will count down to zero - then show as "---" with the Betas, for the duration of the remaining processing time.

Regards
Zy

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14877 - Posted: 31 Jan 2010 | 21:46:14 UTC - in response to Message 14873.

You have to compare speed between 6.05 and 6.06 which are all run on a TONI- workunit (there are two types, one short and one long). What I would like to know is the speed on a fully CPU loaded host, with all cores busy.

AS far as I know the progress bar should work. Does it?

gdf


The results I posted above were run on an i7-920, HT ON, fully loaded with 8 threads of WCG's HFCC subproject.

I am running that unit headless so I don't know if the progressbar was displayed properly.
____________
Thanks - Steve

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14879 - Posted: 31 Jan 2010 | 21:57:23 UTC - in response to Message 14875.
Last modified: 31 Jan 2010 | 21:58:00 UTC

Our tests shows that under full load (all CPUs used), the application is very slow. This is probably true for the standard application as well, a bit less because the kernels are longer there.

If there is no load on the CPU or you have a free core, then the cost is minimal, but then you might just as well use it.
This being confirmed, we will default to 1CPU+1GPU usage.

gdf

Tom Philippart
Send message
Joined: 12 Feb 09
Posts: 57
Credit: 23,376,686
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 14880 - Posted: 31 Jan 2010 | 22:03:40 UTC - in response to Message 14879.


This being confirmed, we will default to 1CPU+1GPU usage.

gdf


sorry, but this will be a very bad move imo...

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14881 - Posted: 31 Jan 2010 | 22:23:49 UTC - in response to Message 14880.

If the hit on the GPU app is so high, then it does not make any sense to have less than 0.1 CPU driving the GPU. You might lose up 100% of performance of 256 GPU cores to save 1 CPU core!
We might go to 0.6 CPU so that two GPUs can share the same CPU. This should be enough and more similar to the reality of the usage. As a matter of fact, many of our calculations do use some CPU anyway for calculations.
Comments?

gdf

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14882 - Posted: 31 Jan 2010 | 22:49:01 UTC - in response to Message 14881.

Previously it worked with comparably little cpu support. Does this method not work any more due to the algorithm changes?

MrS
____________
Scanning for our furry friends since Jan 2002

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14883 - Posted: 31 Jan 2010 | 23:17:11 UTC - in response to Message 14882.

I think that we have a big hit on the previous application as well.
We just did not realize it because on Linux is fine and it's fine on windows if the processor is not fully subscribed.

gdf

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14884 - Posted: 31 Jan 2010 | 23:53:44 UTC - in response to Message 14881.

If the hit on the GPU app is so high, then it does not make any sense to have less than 0.1 CPU driving the GPU. You might lose up 100% of performance of 256 GPU cores to save 1 CPU core!
We might go to 0.6 CPU so that two GPUs can share the same CPU. This should be enough and more similar to the reality of the usage. As a matter of fact, many of our calculations do use some CPU anyway for calculations.
Comments?

gdf


Makes sense in a performance perspective.

0.6 or thereabouts would be fine as by implication dual gpus are involved which by inlarge means a more PC/Technically aware individual who will realise the sense in the trade off. 0.6 should also allow access to that cpu by other apps by single gpu owners, and that will help fend off yelling about credits and interference with other projects.

Its a very obvious statement to say the priority is the needs of the app - but I will repeat it to prevent nugatory spinning off into credit rhetoric here. This is about how to launch this one and the real issues it will face in BOINC Land, not credit Rhetoric.

Utilisation of the cpu/gpu in this way is relatively new in BOINC - SETI does a version of it in one of their apps, but the credit spin there is another world as they are relatively low on credits anyway, and this GPUGRID App is engineered from the ground up - different ball game. The fact remains whatever the rhetoric, a configuration of this nature will open doors to the chattering classes, and if its not approached and "Marketed" correctly, the droves of BOINC Crunchers who love to repeat dramatic rumour, will feed momentum of bad news on its Launch - thats the last thing thats needed when its all there for the right reasons.

Therefore I would strongly suggest a brief "Time Out" at some point to put together the case for it and why the App has ended up the way it does, and why the credits are the way they end up, posting it prominantly as a sticky. GPUGRID members can then refer / link to it in their travels around BOINC Land and nip rumours in the bud.

We can all make a Grand Case in Theory that such a precaution is not required, as all true and rightuous Crunchers naturally march forth to the golden light giving selfless assistance to the goodness of Mankind. However we all know life in BOINC Land is not so simplistic - unfortunately :)

Its not going to take much effort, and will go a long way to deflect critics by being open and transparent. There will be still be some moaning no doubt, but hopefully most will be nipped off by such a move.

Regards
Zy

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14885 - Posted: 31 Jan 2010 | 23:59:21 UTC - in response to Message 14883.

Please, please, please go to 1GPU + 1 CPU!

It is such a pain to try to use 3 cores of a quad in order to leave one free for GPUGrid.
On an i7 what odds is it to only use 7 of the 8 cores? Virtually none. If you were running GPUGrid you were using part of a core anyway, so you actually only lose out on 0.8 threads (10%), and at the same time you are gaining a huge amount for GPUGrid. I think a 60% increase in GPUGrid performance seriously outweighs the loss of 1 thread.
I will put it another way, on an i-7 running WCG or some other Boinc project you lose about 500 points for that one thread (running 24/7). Your GTX 260 (for example) will gain about 9000points! Total Points gain = 8500.

Within 2 months 6core systems with 12 threads will be available. You will even be able to get dual socket versions.
Look to the future, it has dense core populations with multithreading being the norm.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14886 - Posted: 1 Feb 2010 | 0:28:43 UTC - in response to Message 14883.
Last modified: 1 Feb 2010 | 0:38:29 UTC

I think that we have a big hit on the previous application as well.
We just did not realize it because on Linux is fine and it's fine on windows if the processor is not fully subscribed.

gdf

On my systems v6.71 running with a CPU core idle did not speed up the application. V6.71 played very well with every other project, and I currently run more than a dozen on various machines.

I wish people would quit talking about points gained or lost. For many of us it's not about points. It's about being able to run work for other projects too. If ANY project makes that difficult it becomes expendable. I really like GPUGRID and want to keep running it, so if this app makes running other projects more difficult please give us the option of running the old app (via app_info.xml if necessary).

Profile Zydor
Send message
Joined: 8 Feb 09
Posts: 252
Credit: 1,309,451
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 14887 - Posted: 1 Feb 2010 | 0:36:42 UTC - in response to Message 14885.
Last modified: 1 Feb 2010 | 0:46:42 UTC

On a long term prognosis I would agree re 6 cores etc etc - however the transition to multi core niavana has a bear trap.

There are still vast numbers of single core crunchers, they are diminishing quickly of course as machines are replaced, and they cant run the newer cards anyway. However, they are there in large numbers out in BOINC Land, it would be a bad move to cut them off totally, much better in a reputation sense to let them wither on the vine.

Dual Core will be with us in large numbers for years, cutting off one of two cpus would be a bad move.

Aqua debated this recently, and decided that the heavyweight Apps (and there are some real real heavyweight ones there due to the guargantuan amount of data being crunched simulating parts of Quantum Computers) would be restricted to Quad Core only. In fact it was only recently they introduced a light weight App supporting the main app, that they opened the smaller apps to single core.

It was done via Preferences, and in that maybe a way forward here ? The lower gpu useage (say 0.6) could be via one app with an attendent estimate of time to crunch, and second or more app(s) could be there using one cpu and one gpu. Place a test on the faster app to allow access to Dual and Quad core only, and denying download to single core. That way the Dual Core user decides whats right for them, and still leaves the door open for single core/single gpu without arguments spinning round for years.

That will all depend on how the app is put together and how CPU access is granted in the code, obviously I dont know how easy or hard it will be to make such changes without complete rewrites, but I would have thought it would not be too hard to produce two versions around such a premis.

Regards
Zy

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14889 - Posted: 1 Feb 2010 | 1:21:41 UTC

There's lots of speculation and AFAIK few machines that have run many v6.05 & v6.06 WUs. Well here's one:

http://www.gpugrid.net/results.php?hostid=56900&offset=0&show_names=1&state=0

A recap:

v6.05 (GPU then CPU time):

13,190.28 --- 13,184.89
13,136.48 --- 13,131.83


v6.06 (GPU then CPU time):

13,962.64 --- 670.80
14,049.59 --- 689.72

So for around an increase of 15 minutes more GPU time, v6.06 saves over 3.5 hours of CPU time.
This is for a GTX 285. The difference is even more extreme for slower cards. Here's a GT 240:

v6.05 (GPU then CPU time):

41,658.85 --- 40,742.34


v6.06 (GPU then CPU time):

45,470.80 --- 4,785.85

In this case we see an increase of 1:04 of GPU time and a whopping saving of 10:00 hours of CPU time with v6.06.

I certainly prefer v6.06. I can help a lot of projects with that CPU time and will have a more responsive system to work with (especially dual core or slower systems).

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14890 - Posted: 1 Feb 2010 | 1:32:26 UTC - in response to Message 14889.

Our tests shows that under full load (all CPUs used), the application is very slow.

When you say under full load what kind of load are you talking about?
My load, and likely that of other people that are concerned about dedicating a full CPU for each GPU WU , is with other BOINC projects but they all share pretty good and I got very good results with the 6.06 version. If you would like more detailed testing I have a GTX295 and GTX285 that I can configure, set, run anything you would like to help make the best decision for the project.
____________
Thanks - Steve

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14891 - Posted: 1 Feb 2010 | 2:04:53 UTC - in response to Message 14890.

Snow Crash, that was your GTX 285 I used for the comparison above. Hope that's OK.
It looks like v6.06 was running well, even better than the current v6.71 app.

Ross*
Send message
Joined: 6 May 09
Posts: 34
Credit: 443,507,669
RAC: 0
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14892 - Posted: 1 Feb 2010 | 5:51:14 UTC - in response to Message 14885.

[Within 2 months 6core systems with 12 threads will be available. You will even be able to get dual socket versions.
Look to the future, it has dense core populations with multithreading being the norm.]
Hi
to add to this debate. I have 2 i7 920 boxes with a ATI 5970 each plus running WCGrid on 8 threads each as well. Whatever is planned, my view is that the app should be able to run the GPU apps with a CPU app running as well, without any micro management.
I will be running 6 core CPUs in a couple of months hopefully with a new nivida GPU card.
I agree with the use of preferences to fine tune how the box manages its resources.
My future is to 2 gulftowns per MB and 2 double GPUs per box.
Cheers
Ross, New Zealand
____________

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14893 - Posted: 1 Feb 2010 | 7:55:58 UTC - in response to Message 14891.

Snow Crash, that was your GTX 285 I used for the comparison above. Hope that's OK.
I noticed that :-) and that's fine by me. If I didn't want anyone to see them I would make them hidden.
It looks like v6.06 was running well, even better than the current v6.71 app.[/quote]Definately much better ... could I have more please?

____________
Thanks - Steve

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14895 - Posted: 1 Feb 2010 | 8:58:29 UTC - in response to Message 14889.

Beyond,
if this was the situation, it would be fine.
In our tests, loading the CPU with some work, it double the length of the GPU simulation.
So what where you running while you have these sims in progress?

gdf

There's lots of speculation and AFAIK few machines that have run many v6.05 & v6.06 WUs. Well here's one:

http://www.gpugrid.net/results.php?hostid=56900&offset=0&show_names=1&state=0

A recap:

v6.05 (GPU then CPU time):

13,190.28 --- 13,184.89
13,136.48 --- 13,131.83


v6.06 (GPU then CPU time):

13,962.64 --- 670.80
14,049.59 --- 689.72

So for around an increase of 15 minutes more GPU time, v6.06 saves over 3.5 hours of CPU time.
This is for a GTX 285. The difference is even more extreme for slower cards. Here's a GT 240:

v6.05 (GPU then CPU time):

41,658.85 --- 40,742.34


v6.06 (GPU then CPU time):

45,470.80 --- 4,785.85

In this case we see an increase of 1:04 of GPU time and a whopping saving of 10:00 hours of CPU time with v6.06.

I certainly prefer v6.06. I can help a lot of projects with that CPU time and will have a more responsive system to work with (especially dual core or slower systems).


Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14896 - Posted: 1 Feb 2010 | 9:02:26 UTC - in response to Message 14893.

Snow Crash, that was your GTX 285 I used for the comparison above. Hope that's OK.
I noticed that :-) and that's fine by me. If I didn't want anyone to see them I would make them hidden.
It looks like v6.06 was running well, even better than the current v6.71 app.
Definately much better ... could I have more please?
[/quote]

Hi,
run the 6.06 application together with say Seti/Aqua running on all your CPU cores and let's see what are the performance.

GDF

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14897 - Posted: 1 Feb 2010 | 9:56:44 UTC - in response to Message 14895.

Hi GDF,

For the examples below: The GTX 285 belongs to Snow Crash and the GT 240 belongs to SKGiven. In my brief search those were the only machines I came up with that had run long WUs for both v6.05 and v6.06. Maybe they can supply more details about what else was running at the time.

As for my 5 GPUGRID boxes: only 1 of 5 received any long beta WUs, both of which were of the v6.05 variety (even though all are setup to receive test apps). Of those 2 v6.05 WUs: 1 ran for an enormous amount of time, went to 100% and then back to 17% before I aborted it. The other refused to start, something I have never seen before.

Also as I stated somewhere above, with the non-BOINC Wieferich project running on both cores of the dual core machine that got v6.05 WUs: with the normal 2 Wieferich CPU WUs running, 1 stalled and the machine became 95% unresponsive. When I killed 1 instance of Wieferich and let v6.05 have the whole core, the machine became usable but of course I lost 1/2 of the Wieferich production. No other GPU app causes this problem, not v6.71, not Collatz, not MilkyWay, not PrimeGrid, not RC5-72, not SETI. You see the problem some of us have with v6.05...

Regards/Beyond


Beyond,
if this was the situation, it would be fine.
In our tests, loading the CPU with some work, it double the length of the GPU simulation.
So what where you running while you have these sims in progress?

gdf

There's lots of speculation and AFAIK few machines that have run many v6.05 & v6.06 WUs. Well here's one:

http://www.gpugrid.net/results.php?hostid=56900&offset=0&show_names=1&state=0

A recap:

v6.05 (GPU then CPU time):

13,190.28 --- 13,184.89
13,136.48 --- 13,131.83


v6.06 (GPU then CPU time):

13,962.64 --- 670.80
14,049.59 --- 689.72

So for around an increase of 15 minutes more GPU time, v6.06 saves over 3.5 hours of CPU time.
This is for a GTX 285. The difference is even more extreme for slower cards. Here's a GT 240:

v6.05 (GPU then CPU time):

41,658.85 --- 40,742.34


v6.06 (GPU then CPU time):

45,470.80 --- 4,785.85

In this case we see an increase of 1:04 of GPU time and a whopping saving of 10:00 hours of CPU time with v6.06.

I certainly prefer v6.06. I can help a lot of projects with that CPU time and will have a more responsive system to work with (especially dual core or slower systems).



Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14898 - Posted: 1 Feb 2010 | 10:48:46 UTC - in response to Message 14897.

I own the GTX285 that I posted results for.

It is an i7-920 with hyperthreading turned on, running Windows XP Pro 32 bit with BOINCManager version 6.2.28 (WCG Edition)

I was executing 8 World Community Grid WUs, specifically their Help Fight Childhood Cancer subproject, the entire time that the 6.05 and 6.06 GPUGrid betas were executing.
http://www.gpugrid.net/show_host_detail.php?hostid=56900
http://www.gpugrid.net/results.php?hostid=56900


I also have another i7-920 which has hyperthreading turned on, running Windows Vista Ultimate 64 bit with BOINCManager 64 bit 6.10.29. This machine has a GTX295.
http://www.gpugrid.net/show_host_detail.php?hostid=31780
http://www.gpugrid.net/results.php?hostid=31780

I only have valid results for two of the small betas on v6.06 which do show a higher usage of CPU than the results on my GTX285.
GTX285 CPU usage is 6% of CPU usage
GTX295 CPU usage is 16% of CPU usage
I was running a combination of Einstein and ClmiatePrediction on the machine with the GTX295 while the short betas were executing.

I have attached this second PC to SETI and am running 8 of their WUs along with two of GPUGrid's 6.06 betas.
Yes, I aborted a couple of GPUGrid 6.71 WUs so we could get this test started.

I am seeing low CPU usage for the GPUGrid 6.06 betas with between 2-3 percent each as viewed in Task Manager.

The "Status" column reports the WU is running and using 0.28 CPU + 1.00 NVIDIA GPU

The progressbar/ percent complete indicator does not appear to be updating at all.

The Time to Complete is updating.

I have been running for 30 minutes and if the time to complete is accurate then I am not seeing any improvement on runtime at all. I will let this continue and we can follow up with final results when it finishes.

Let's start listing things to test ...
At first blush it looks like it has something to do with the OS?
Perhaps we could test different versions / bits of BOINCManager?

____________
Thanks - Steve

Tom Philippart
Send message
Joined: 12 Feb 09
Posts: 57
Credit: 23,376,686
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 14899 - Posted: 1 Feb 2010 | 14:22:40 UTC - in response to Message 14881.

If the hit on the GPU app is so high, then it does not make any sense to have less than 0.1 CPU driving the GPU. You might lose up 100% of performance of 256 GPU cores to save 1 CPU core!
We might go to 0.6 CPU so that two GPUs can share the same CPU. This should be enough and more similar to the reality of the usage. As a matter of fact, many of our calculations do use some CPU anyway for calculations.
Comments?

gdf


Folding@Home also used 1 cpu core for their ati client. They worked around it using flush intervals, buffering and dualbuffering as environment variables. I don't know if this could be applied to cuda...

I get your point with the performance, but as a whole I think this shouldn't stay this way.

I think at the moment virtually no one runs gpugrid exclusively. Say I have a quadcore with 1 gpu, my objective is to maximize my multiproject participation. To illustrate let's say I run WCG on the cpu and gpugrid on the gpu. I'm currently able to contribute 4 cores to WCG and "1core" (the gpu) to gpugrid.
This allows me to have "5cores" in my quadcore computer. I was able to run 4 WUs of WCG at a time, and when I found out about gpugrid, I was able to contribute to this project on top of my former regular contribution on WCG.

Now, it would mean to me that I'm still contributing to the 2 projects, but it would degrade my WCG production. I'll end up with "4cores" (4WUs running on a quad+gpu).

I'm not talking about credits, in that perspective the new solution would outweight the old, but I'm talking in terms of output. This compared to the old situation and other gpu projects.

Gpu crunching is still in an early stage, but it has till now been seen as an additional source of production: it works on top of what was possible before, as a "5th core".

As somebody before me said, the other projects all use almost no cpu power.
If you have a computer dedicated to gpugrid, the new solution is of course the best one as you're not viewing it from a multiproject point of view.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14901 - Posted: 1 Feb 2010 | 15:38:31 UTC - in response to Message 14899.

When running a 6.71 application Boinc says 0.64CPUs + 1GPU.
Using Boinc and Task Manager you can see that Boinc is running 4 WCG CPU tasks and 1 GPU task (100% CPU usage). If you set Boinc to use 3 of the 4 CPUs, this actually applies to GPUGrid tasks as well. So Boinc will try to run three WCG CPU tasks, and use one of these CPUs for the GPU tasks, leaving one CPU free (useless).
Unless you can set things up so that Boinc thinks it takes 1 full CPU + 1 GPU then the potential GPU efficiencies will not be seen!

Although there are work arounds, to get as much GPU task efficiency as possible, none are particularly handy.

One method is to install 2 versions of Boinc (a 64bit and a 32bit) using the 32bit to run GPUGrid tasks and setting the CPU % usage to 25%, and setting the 64bit client to 75%. Of course you need a 64bit OS to do this.
A second method is similar, but involves using the WCG Bespoke Client rather than Boinc for WCG tasks and setting it to use 3 CPUs. This is obviously no use to anyone running other Boinc tasks.
A third method requires setting up Virtual PCs, eg running XP from within Visa.

Jeremy
Send message
Joined: 15 Feb 09
Posts: 55
Credit: 3,542,733
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 14902 - Posted: 1 Feb 2010 | 16:03:02 UTC - in response to Message 14901.

If you set Boinc to use 3 of the 4 CPUs, this actually applies to GPUGrid tasks as well. So Boinc will try to run three WCG CPU tasks, and use one of these CPUs for the GPU tasks, leaving one CPU free (useless).
Unless you can set things up so that Boinc thinks it takes 1 full CPU + 1 GPU then the potential GPU efficiencies will not be seen!


This is not the case based on my most recent experience. Last night I set BOINC to use 75% of available cores. Turned it loose and I got 100% load on 3 out of 4 cores and an average of about 30% on the fourth with no background tasks active. If I suspend activity on all projects except GPUgrid I get the same result. One core with a 30% load, the rest with nothing.

That's with two 6.71 GPUgrid apps running. BOINC 6.10.18, Win7 Ult64, C2Q @ 3.83 GHz, and two GTX 260-216.

I'm waiting for a few more WUs to finish before I check to see if there's a performance advantage to running this way.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14903 - Posted: 1 Feb 2010 | 17:07:02 UTC

I've been aborting v6.71 tasks on my GTX 260 trying to get a v6.06 WU to test, no luck.
It's hard to test the beta if we can't get WUs. Did manage to get one on the GT 240 however...

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14905 - Posted: 1 Feb 2010 | 17:39:07 UTC

GTX 295, 701MHz, 1509MHz, 1086MHz (896MB) driver: 19062 , i7-860@3.8 GHz/, Vista64

ACEMD beta version v6.05 + 8 CPU-cores(HT)- Spinhenge@home:

Run time: 15771.8694
CPU time: 15753.59


ACEMD beta version v6.06 + 8 CPU-cores(HT)- Spinhenge@home:

Run time: 17494.2924
CPU time: 2444.832

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14906 - Posted: 1 Feb 2010 | 17:47:59 UTC
Last modified: 1 Feb 2010 | 17:52:22 UTC

I just ran this beta test. It errored out instantly, as it did with my wingmen.

GTX 280, Q6600, all 4 cores running tasks from PrimeGrid.

Edit: Didn't realize, but two WUs had run. The ID above is corrected.

Also ran was this test, which completed normally with 203 seconds of run time and 33 CPU seconds. Not sure what else was running at that time.
____________
Want to find one of the largest known primes? Try PrimeGrid. Or help cure disease at WCG.

Profile K1atOdessa
Send message
Joined: 25 Feb 08
Posts: 249
Credit: 370,320,941
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14908 - Posted: 1 Feb 2010 | 19:23:16 UTC

My first test 6.06 WU choked after running for about 1.5 hours. Saw the issue where the progress bar never moved, but saw the message below after it errored out.

"SWAN : FATAL : Failure executing kernel sync [frc_sum_kernel] [999]
Assertion failed: 0, file ../swan/swanlib_nv.cpp, line 203

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information."

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14909 - Posted: 1 Feb 2010 | 19:36:41 UTC

Jeremy wrote:
If you set Boinc to use 3 of the 4 CPUs, this actually applies to GPUGrid tasks as well. So Boinc will try to run three WCG CPU tasks, and use one of these CPUs for the GPU tasks, leaving one CPU free (useless).

This is not the case based on my most recent experience.

I second that. Running MW on a HD4870 and a C2Q. If I set BOINC 6.10.29 to use 75% CPU is launches 3 CPU tasks and one MW@ATI. This way performance is much better than at 4 CPU + 1 MW, even though MW itself uses little CPU. The catch here is that it needs cpu support often and in precise intervals. So effectively you have to dedicate one core here as well.. or live with a slower GPU.


Tom Philippart wrote:
I'm not talking about credits, in that perspective the new solution would outweight the old, but I'm talking in terms of output.

You're measuring with two different gauges here. A reduction of your WCG output by a factor of 1.33 does count, but a GPU-Grid output increase by a factor of 1.66 does not count? You must not count it purely in terms of "cores", as that can be quite misleading. Or is one core of a Celeron 266 MHz worth as much as one of an i7? Or a GTX260 or a GTX380?


Zydor wrote:
... and second or more app(s) could be there using one cpu and one gpu. Place a test on the faster app to allow access to Dual and Quad core only, and denying download to single core.

Don't deny it, just don't make it the default. Otherwise dedicated cruncher boxes will hate you ;)

MrS
____________
Scanning for our furry friends since Jan 2002

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14910 - Posted: 1 Feb 2010 | 19:40:25 UTC - in response to Message 14881.

You might lose up 100% of performance of 256 GPU cores to save 1 CPU core!

Comments?

gdf


The reason the CPU performance is important to many of is, well, look down at my sig. We run other applications, most of which don't run on GPUs. Many, many problems simply don't lend themselves to parallelization very well.

While helping with medical research is fine and noble, so are other tasks such as preventing the spread of Malaria or keeping an asteroid from falling on my head. One of the criteria which I use to decide whether or not to run a project is whether it will interfere with other projects I run. A GPU-based project is already consuming a highly valuable resource. I don't want it to also gobble up the a CPU core that would otherwise be used by projects that can't run on the GPU. Likewise, I won't run something like Einstein@Home's APB1/2 GPU task because it wastes the GPU, which could be put to better use by projects such as GPUGRID.


____________
Want to find one of the largest known primes? Try PrimeGrid. Or help cure disease at WCG.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14911 - Posted: 1 Feb 2010 | 21:35:10 UTC - in response to Message 14909.

Running MW on a HD4870 and a C2Q. If I set BOINC 6.10.29 to use 75% CPU is launches 3 CPU tasks and one MW@ATI. This way performance is much better than at 4 CPU + 1 MW, even though MW itself uses little CPU. The catch here is that it needs cpu support often and in precise intervals. So effectively you have to dedicate one core here as well.. or live with a slower GPU.

This hasn't been my experience at all. I'm running 6 HD4770 ATI cards on 4 machines. MW tasks reliably take 210 seconds CPU time and 212 seconds elapsed time no matter what project is running on the 4 CPU cores (Athlon II 620, Win7-64, ATI v9.12). So thought I'd give it a test to check out your theory. The machine I tested is running 4 instances of SIMAP on the CPU (10 WUs run each way). Times:

MW + 4 instances SIMAP: 3:30 CPU time --- 3:32 elapsed time
MW + 0 instances SIMAP: 3:29 CPU time --- 3:31 elapsed time

There you have it. The dedicated quad with nothing running but the OS saves 1 second or about .47%, not much at all. As a trade off each CPU core pumps out a SIMAP WU every 28 minutes.

If your results differ much from these you might try adding -b to your MW app_info.xml commandline parameters:-)

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14912 - Posted: 1 Feb 2010 | 22:18:18 UTC - in response to Message 14911.
Last modified: 1 Feb 2010 | 22:43:11 UTC

Good news. We have found the problem of the hanging.
Ready for 6.07.

gdf

Edit. Uploaded.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14913 - Posted: 1 Feb 2010 | 23:11:34 UTC - in response to Message 14912.

Should we abort the v6.06 WUs that haven't started yet?

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14915 - Posted: 2 Feb 2010 | 8:24:30 UTC - in response to Message 14913.

There was a problem 7.07. So, we removed for now.

gdf

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14916 - Posted: 2 Feb 2010 | 9:35:35 UTC - in response to Message 14911.

Beyond, thanks for that post. My test were done a few months back, can't remember if it was 0.20 or 0.19. Now I'm running a C2Q, Win7 64, Cat 9.11 and MW 0.21 without app_info.

MW + 3 instances SIMAP: 170s CPU time --- 171.7 +/- 0.3 elapsed time
MW + 4 instances SIMAP: 173s CPU time --- 175 - 176 elapsed time

That's much better than last time I checked and just about 2% slower (-> 2k credits/day). Would you mind sending me your app_info via PM?

MrS
____________
Scanning for our furry friends since Jan 2002

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14919 - Posted: 2 Feb 2010 | 11:34:05 UTC - in response to Message 14916.

I have been trying to track the various beta WUs
but the results seem to be all over the place, meaning that sometimes with the CPOU fully utilized by other BOINC projects (WCG, Einstein, Climate Prediction, Seti) the betas seems to run about the same as 6.71. Other times they process very quickly with a big decrease in runtime (-55%) , and finally I have had a couple instances where they ran worse than usual (+55%). I have seen all three scenarios on both my GTX285 and on my GTX295. I'll keep trying to find the magic combination.
I have one running right now that is displaying the "elapsed time", "progressbar", and the "estimated time to completion" accurately (boicmanager 6.2.28, WCG edition).
ps If you added the ability to sort the results page by Sent Time and also by Returned Time it would make tracking this stuff much easier because your taskIDs are no longer in sequence.
____________
Thanks - Steve

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14921 - Posted: 2 Feb 2010 | 11:47:56 UTC - in response to Message 14919.
Last modified: 2 Feb 2010 | 11:58:42 UTC

The timing of 6.05 and 6.05 could be wrong as there was a problem in the restarting. Only the fast times are correct.

Just uploaded acemdbeta6.08.

Enjoy.

gdf
EDIT. Abort any job with acemdbeta6.06 still running.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14925 - Posted: 2 Feb 2010 | 15:21:10 UTC - in response to Message 14916.

Beyond, thanks for that post. My test were done a few months back, can't remember if it was 0.20 or 0.19. Now I'm running a C2Q, Win7 64, Cat 9.11 and MW 0.21 without app_info.

MW + 3 instances SIMAP: 170s CPU time --- 171.7 +/- 0.3 elapsed time
MW + 4 instances SIMAP: 173s CPU time --- 175 - 176 elapsed time

That's much better than last time I checked and just about 2% slower (-> 2k credits/day). Would you mind sending me your app_info via PM?

MrS


You should have a PM...

Also sorry, made a typo in the previous post. The commandline should be:

<cmdline>b-1</cmdline>

not -b

That should get you a bit more GPU usage :-)



Profile robertmiles
Send message
Joined: 16 Apr 09
Posts: 503
Credit: 727,920,933
RAC: 155,858
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14927 - Posted: 2 Feb 2010 | 16:28:59 UTC

A problem with an ACEMD beta version 6.06 (cuda) workunit:

At 13.791% progress, this error message was shown on the
screen; progress then stopped, but the elapsed time kept
rising.

acemd_6.06_windows.intelx86__cuda has
stopped working

A problem caused the program to stop working correctly.
Windows will close the program and notify you if a solution is
available.

(A button labelled) Close program

The workunit was L2-TONI_TEST0202-0-RND6466_2.

http://www.gpugrid.net/result.php?resultid=1805197

After clicking on the button, the error message window
disappeared, with no apparant immediate change in the way
the workunit was running; but within a minute, a computation
error was reported and the GPU was left idle for a few minutes
it took to get another GPUGRID workunit.

<core_client_version>6.10.18</core_client_version>
<![CDATA[
<message>
The extended attributes are inconsistent. (0xff) - exit code 255 (0xff)
</message>
<stderr_txt>
# There is 1 device supporting CUDA
# Device 0: "GeForce 9800 GT"
# Clock rate: 1.50 GHz
# Total amount of global memory: 1073741824 bytes
# Number of multiprocessors: 14
# Number of cores: 112
MDIO ERROR: cannot open file "restart.coor"
SWAN : FATAL : Failure executing kernel sync [frc_sum_kernel] [999]
Assertion failed: 0, file ../swan/swanlib_nv.cpp, line 203

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

</stderr_txt>

BOINC 6.10.18
64-bit Windows Vista SP2

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14931 - Posted: 2 Feb 2010 | 17:13:33 UTC
Last modified: 2 Feb 2010 | 17:40:02 UTC

6.08 is looking very nice!!!
Windows XP Pro 32 bit, Boinc Manager 6.2.28 (WCG edition)
Core i7-920, GTX285 w/ 191.07 driver

Constant CPU load with 8 CPU WUs for WCG (all different types, HFCC, HCMD2, HPF2, NRFW, FAAH) and also a couple of Einstein GW6 WUs.

GPUGrid 6.08
Elapsed Time, Progressbar, To Completion all working properly.
The "Result" column says .29 CPU but Task Manager barely registers any CPU usage for this WU at all. After 3 hours 50 minutes elapsed time the WU "Properties" reports CPU usage at 10 minutes 36 seconds.
Currently this WU is 92.416% complete and looks to be done in another 18 minutes. This GPU typically proceses a GIANNI_BIND in 6 hours.

Let me know if there is anything else I can do to help.

Best regards,
Steve

<edit>
My first 6.08 completion
runtime = 15,021.08
cputime = 692.70

It looks like we still get a nice improvement (30%) even when all CPU cores are loaded.

If I am around and see another 6.08 show up on thios machine I will shut down a couple of CPU tasks and see if there is any further improvement unless GDF can say that test would be a waste of time.

</edit>

Tom Philippart
Send message
Joined: 12 Feb 09
Posts: 57
Credit: 23,376,686
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 14933 - Posted: 2 Feb 2010 | 18:25:18 UTC - in response to Message 14909.


Tom Philippart wrote:
I'm not talking about credits, in that perspective the new solution would outweight the old, but I'm talking in terms of output.

You're measuring with two different gauges here. A reduction of your WCG output by a factor of 1.33 does count, but a GPU-Grid output increase by a factor of 1.66 does not count? You must not count it purely in terms of "cores", as that can be quite misleading. Or is one core of a Celeron 266 MHz worth as much as one of an i7? Or a GTX260 or a GTX380?
MrS


As I said, as nearly no one is running gpugrid as their only project, I want to run it on top of the regular production: stay with the normal output of my primary project and only take minor losses to contribute with the gpu. This is a higher output than running nothing on the gpu. My point is that I want to run it "on top" of my primary project. I don't want to touch the output of my primary focus.

That's also the reason why I didn't run the folding ati client until they build the workaround to reduce cpu usage

Anyways I think aiming to reduce cpu usage, by for instance using buffering techniques (as far as it's possible) is the best solution. Einstein@home is also working on reducing the cpu usage of their app.

Profile K1atOdessa
Send message
Joined: 25 Feb 08
Posts: 249
Credit: 370,320,941
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14934 - Posted: 2 Feb 2010 | 18:25:46 UTC
Last modified: 2 Feb 2010 | 18:46:11 UTC

Had two 6.08 Beta's fail within 3 seconds: Beta WU 1 and Beta WU 2

Both had the message "SWAN: FATAL : swanMalloc failed"

-----------------EDIT--------------------

May have been just something weird with BOINC switching tasks or something. I rebooted and two new 6.08 Beta's are crunching along fine now.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14935 - Posted: 2 Feb 2010 | 18:35:51 UTC - in response to Message 14931.

Please go ahead with the test.
You have too look at the elapsed time or in the result output ms/step to compare WUs of the same name. The unit we are sending around is not a GIANNI_BIND, but they are all the same for the beta app.

gdf

6.08 is looking very nice!!!
Windows XP Pro 32 bit, Boinc Manager 6.2.28 (WCG edition)
Core i7-920, GTX285 w/ 191.07 driver

Constant CPU load with 8 CPU WUs for WCG (all different types, HFCC, HCMD2, HPF2, NRFW, FAAH) and also a couple of Einstein GW6 WUs.

GPUGrid 6.08
Elapsed Time, Progressbar, To Completion all working properly.
The "Result" column says .29 CPU but Task Manager barely registers any CPU usage for this WU at all. After 3 hours 50 minutes elapsed time the WU "Properties" reports CPU usage at 10 minutes 36 seconds.
Currently this WU is 92.416% complete and looks to be done in another 18 minutes. This GPU typically proceses a GIANNI_BIND in 6 hours.

Let me know if there is anything else I can do to help.

Best regards,
Steve

<edit>
My first 6.08 completion
runtime = 15,021.08
cputime = 692.70

It looks like we still get a nice improvement (30%) even when all CPU cores are loaded.

If I am around and see another 6.08 show up on thios machine I will shut down a couple of CPU tasks and see if there is any further improvement unless GDF can say that test would be a waste of time.

</edit>

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14936 - Posted: 2 Feb 2010 | 19:13:06 UTC

GTX260 (754 MHz, Shader: 1568 MHz, Speicher: 1211 MHz) (WinXP_32, Kentsfield, 4xSpinhenge@home)


ACEMD beta version v6.08:

Run time 16629.203125
CPU time 608.4688
Time per step: 26.600 ms


ACEMD beta version v6.06:

Run time 17283.195
CPU time 2136.792
Time per step: 27.649 ms

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14937 - Posted: 2 Feb 2010 | 19:34:17 UTC

I got a 6.08 result which, unfortunately, was consuming a significant portion of a C2Q core (7% on task manager, so about 28% of the CPU time of one core.) Typical usage is 0 to 1 percent in Task Manager.

I aborted the WU, and also turned off the beta WUs for now.


Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14938 - Posted: 2 Feb 2010 | 20:11:03 UTC - in response to Message 14937.

CPU usage is not important anymore. Just look at the time/step with and without load. So far, it seems very good.

gdf

Profile Stoneageman
Avatar
Send message
Joined: 25 May 09
Posts: 224
Credit: 34,057,224,498
RAC: 190
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14939 - Posted: 2 Feb 2010 | 20:18:48 UTC

Running three 6.08 units and eight WCG units together on one machine using 275's and looks fine. CPU load for 6.08 units is 17-18% and WCG units average 89%. GPUgrid tasks showing 60% performance increase.
Job well done GDF and team.

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14943 - Posted: 2 Feb 2010 | 22:31:27 UTC - in response to Message 14937.

I got a 6.08 result which, unfortunately, was consuming a significant portion of a C2Q core (7% on task manager, so about 28% of the CPU time of one core.) Typical usage is 0 to 1 percent in Task Manager.


I need to correct my own statement.

Either I'm misremebering how each project behaves, or something is different on this box today. Both are fairly likely. :)

The production (6.71) application is showing 5 to 6 percent CPU utilization in Task Manager, so the difference between the CPU usage between the two is insignificant.

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14944 - Posted: 2 Feb 2010 | 23:23:55 UTC
Last modified: 2 Feb 2010 | 23:56:59 UTC

Fully loaded with 8 CPU WUs, one 6.71 GPU and one 6.08 GPU.

While watching task manager on my system with an old style GTX295 (2 physical cards) I had one 6.71 WU and one 6.08 WU. I bump the Priority up to Normal on the 6.71 and it starts using about 5%-10% CPU. After watching that for a few minutes I thinking about what might be going on so I bump the beta 6.08 priority up to Normal which I noticed is at "Below Normal" (most BOINC WUs are set to Low) and the 6.71 returns to using 2%-3% (sometimes up to 5%) like it was before I bumped it to Normal. So I then push 6.71 down to Below Normal and the 6.08 starts to take up more CPU, 5%-8% and occasionally 10%-12%.

Perhaps because it is a GTX295 it can only use 1 core for both cards and not only is it competing against other processes on that core but the two WUs are competing against each other. I'll see if the behavior is any different when I have two 6.08 WUs running.
The above behavior only is apparent when running fully loaded with 8 CPU WUs running. If I reduce to 7 CPU WUs then the CPU usage stays fairly constatnt between 2-3 for each GPU WU no matter what I set the Priority to. I will see what the CPU usage is long term, but wonder if it will run slower overall because it has to wait for CPU cycles to be available which I am assuming is what the spikes to 8%-12% I am observing are.

<edit>and if you change the Priority of the WUs enough you WILL crash them. I just lost two betas this way, I will just let them run instead of changing them.</edit>
____________
Thanks - Steve

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14945 - Posted: 3 Feb 2010 | 1:13:18 UTC - in response to Message 14944.

Running 2 beta 6.08 WUs at the same time on a GTX295 takes longer than they would if you processed them with 6.71.
You get the same result if you are fully loaded with 8 other CPU WUs or if you reduce to 7.
____________
Thanks - Steve

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14949 - Posted: 3 Feb 2010 | 10:55:49 UTC - in response to Message 14945.

Could somebody with one of those failing GTX260 see if the beta app works for them?

gdf

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14950 - Posted: 3 Feb 2010 | 12:03:38 UTC - in response to Message 14949.

Just started running one IBUCH task using the 6.08 Beta application on a GTX260 sp216:
No problems after 53minutes, just over 15% complete. Task says it is using 33%CPU.
From resource monitor I can see that the GPUGrid task is using about 7% of the total CPU (Phenom II 940). It fluctuates from about 3% to 11%, but mostly sits around 7%, which is 28% of one core and close enough to stated 33% (more accurate than my ability to read a percentage line bouncing up and down).

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14953 - Posted: 3 Feb 2010 | 14:53:05 UTC - in response to Message 14950.

Was this GTX260 causing problems for the FFT bug before?

gdf

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14954 - Posted: 3 Feb 2010 | 14:56:18 UTC - in response to Message 14949.
Last modified: 3 Feb 2010 | 15:09:00 UTC

Could somebody with one of those failing GTX260 see if the beta app works for them?

gdf


Picked up 4 IBUCH_1000smd on my machine with dual GTX260's (65nm). Current estimate is 3 hours 40 minutes to completion. Will let you know in a few hours how they go. Link to host here.
____________
BOINC blog

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14957 - Posted: 3 Feb 2010 | 15:43:23 UTC

GTX260 (754 MHz, Shader: 1568 MHz, Speicher: 1211 MHz) (WinXP_32, Kentsfield)

100% load on all cores (4xSpinhenge@home)

TONI_TEST -ACEMD beta version v6.08

Process Priority: below normal

Time per step: 26.600 ms
Run time 16629.203125
CPU time 608.4688


Process Priority: above normal

Time per step: 25.016 ms
Run time 15637.953125
CPU time 628.4375

Process Priority: high

Time per step: 24.406 ms
Run time 15257.40625
CPU time 659.7969

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14962 - Posted: 3 Feb 2010 | 17:17:10 UTC - in response to Message 14957.

Was this GTX260 causing problems for the FFT bug before?
gdf


No. The FFT bug is not a problem on my 55nm GTX 260, as it uses the G200b revision. The problem was seen on the earlier 65nm G200 versions of the GTX260 – both the 216sp and 192sp card versions. I no longer have one of these versions.

Still, the 55nm card can act as a comparison reference.

MarkJ has an earlier 65nm (G200) GTX260 card, which would have been subject to intermittent FFT errors, and is running a Beta.

Siegfried Niklas, Which version is your card?
- GTX 260 sp216 55nm, GTX 260 sp216 65nm, or GTX 260 sp192 65nm?

For reference,
my 144-IBUCH_1000smd_pYEEI_100202-0-10-RND5155_1 task completed in 5h 34min. http://www.gpugrid.net/result.php?resultid=1810596
Completed and validated:
Run Time 20,202.83
CPU Time 5,240.67
(Credit claimed 3,977.21, Credit granted 5,369.23)

- My GTX 260 (G200b) 216shaders:
GPU 625MHz, Memory 1100MHz (X2), Shaders 1348MHz (Factory Clocked)

- The NVidia Reference clock rates for the GTX 260 cards:
GPU 576MHz, Memory 1998MHz, Shaders 1242MHz.
Most cards are somewhat factory, if not user, overclocked compared to these.

As the GTX 280 also uses the 65nm G200 core technology, it too was presumably subject to the FFT bug.

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 14965 - Posted: 3 Feb 2010 | 17:50:52 UTC - in response to Message 14962.


...
Siegfried Niklas, Which version is your card?
- GTX 260 sp216 55 nm, GTX 260 sp216 65nm, or GTX 260 sp192 65nm?




It is a GTX 260 sp216 55nm Rev. B1

I run a GTX 260 sp216 65nm Rev. A2 on a other host (same OC). No significant difference at "process priority: below normal"

(I had only one "TONI_TEST -ACEMD beta version v6.08" on the "65nm Rev. A2" card.
I never had problems with the "FFT bug")

dsred
Send message
Joined: 15 Jan 10
Posts: 1
Credit: 27,072,760
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwat
Message 14970 - Posted: 3 Feb 2010 | 18:55:14 UTC - in response to Message 14965.

sorry, to ask this stupid question:
I just would like to ask for support how to add my system to run the test applications? I enabled 'Run test applications?' in my GPUGRID preferences but this seemed to have no effect. Is there anything else to do?

system is an ubuntu 9.10 64bit with 2 x EVGA GTX275 FTW

Best regards,
dsred

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14971 - Posted: 3 Feb 2010 | 19:14:43 UTC - in response to Message 14970.

dsred, you have done all you have to do.
You could watch for Beta releases and abort any queued GPUGrid tasks to force a task request. You would however need to be watching the forums closely otherwise you would get a normal task, in which case you would have wasted your time, and the GPUGrid server’s time. If you are going to do things like that it is more beneficial if you chip in by posting up your beta result details.

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14972 - Posted: 3 Feb 2010 | 20:28:18 UTC - in response to Message 14954.

Picked up 4 IBUCH_1000smd on my machine with dual GTX260's (65nm). Current estimate is 3 hours 40 minutes to completion. Will let you know in a few hours how they go. Link to host here.


1st wu (RND8704_2) is on 89% and thinks another 40 mins to complete

2nd wu (RND4096_1) stuck at 7.12%. Suspend/resume in BOINC seems to have got it going again and its now up to 9%.

3rd and 4th wu waiting to run.

Host is also running 8 Einstein GW searches on the cpu, so that might have slowed things down a bit.
____________
BOINC blog

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 14974 - Posted: 3 Feb 2010 | 20:57:26 UTC - in response to Message 14970.

sorry, to ask this stupid question:
I just would like to ask for support how to add my system to run the test applications? I enabled 'Run test applications?' in my GPUGRID preferences but this seemed to have no effect. Is there anything else to do?

system is an ubuntu 9.10 64bit with 2 x EVGA GTX275 FTW

Best regards,
dsred


It is just for windows now.
gdf

Keith
Send message
Joined: 2 Feb 09
Posts: 4
Credit: 13,776,401
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwat
Message 14975 - Posted: 3 Feb 2010 | 21:16:37 UTC

Does anyone know if running a gtx295 as one card
instead of 2 x gtx200's has any efect on process time?

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14976 - Posted: 3 Feb 2010 | 21:25:02 UTC - in response to Message 14972.

1st wu (RND8704_2) is on 89% and thinks another 40 mins to complete

2nd wu (RND4096_1) stuck at 7.12%. Suspend/resume in BOINC seems to have got it going again and its now up to 9%.

3rd and 4th wu waiting to run.

Host is also running 8 Einstein GW searches on the cpu, so that might have slowed things down a bit.


1st wu completed successfully. Link to wu here

2nd wu failed. Had 2 popup windows on the console saying app has stopped responding (seems to be new with Win 7). Link to wu here

3rd and 4th wu now running.

Cards are GTX260 (65nm) 216sp revision A2
____________
BOINC blog

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 14979 - Posted: 4 Feb 2010 | 8:31:10 UTC

Hi would test it, but:

<core_client_version>6.6.28</core_client_version>
<![CDATA[
<message>
Unzul�ssige Funktion. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
# There is 1 device supporting CUDA
# Device 0: "GeForce 9600M GT"
# Clock rate: 1.25 GHz
# Total amount of global memory: 536870912 bytes
# Number of multiprocessors: 4
# Number of cores: 32
MDIO ERROR: cannot open file "restart.coor"

</stderr_txt>
]]>

http://www.gpugrid.net/result.php?resultid=1814376)


or watch also:
http://www.gpugrid.net/result.php?resultid=1814070
http://www.gpugrid.net/result.php?resultid=1808555


Best RS

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 14982 - Posted: 4 Feb 2010 | 11:18:27 UTC - in response to Message 14979.

I'm afraid the GeForce 9600M GT (laptop) is not suitable for the project... thanks anyway.

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 14984 - Posted: 4 Feb 2010 | 11:40:18 UTC - in response to Message 14982.

I'm afraid the GeForce 9600M GT (laptop) is not suitable for the project... thanks anyway.


Hi.. thx for information.

But i'd crunched several WU's til yet (switching to Beta) with the normal Cuda-App.?!


best RS

Tom Philippart
Send message
Joined: 12 Feb 09
Posts: 57
Credit: 23,376,686
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 14985 - Posted: 4 Feb 2010 | 14:00:56 UTC

you can still run another gpu project like folding@home

according to the chart the gpu is supported, just slow... strange

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 14986 - Posted: 4 Feb 2010 | 14:06:03 UTC - in response to Message 14985.

you can still run another gpu project like folding@home

according to the chart the gpu is supported, just slow... strange


hm... i understand..

...but i HAD crunched here at GPUgrid with THIS GPU!
Would my card at this moment not further supported at all? Or only with the new beta-App?


best RS
(Sorry for poor english)

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14990 - Posted: 4 Feb 2010 | 16:56:56 UTC - in response to Message 14986.

Your card IS supported -- but is considered too slow to reliably return WUs quickly enough to make the WU deadline.

That speed determination is a generalization. Since you have your computers hidden, I can't determine how long it was actually taking your computer to crunch the WUs.

If you have been crunching WUs already (and returning them within the deadlines) then you'll continue to be able to do so. Since the new application is faster, it will be somewhat easier for slower cards to meet the deadline -- assuming the GPUGRID folks don't either increase the amount of work done in each WU and/or lower the deadlines as a result of the faster algorithm.

If you're missing the deadline, you're A) slowing down the project (bad for the science) and B) possibly not getting credit for the WU (bad, if that's important to you.)

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 14991 - Posted: 4 Feb 2010 | 17:22:53 UTC - in response to Message 14990.

Your card IS supported -- but is considered too slow to reliably return WUs quickly enough to make the WU deadline.

That speed determination is a generalization. Since you have your computers hidden, I can't determine how long it was actually taking your computer to crunch the WUs.

If you have been crunching WUs already (and returning them within the deadlines) then you'll continue to be able to do so.


I had succesfull... means within the deadline.


Since the new application is faster, it will be somewhat easier for slower cards to meet the deadline -- assuming the GPUGRID folks don't either increase the amount of work done in each WU and/or lower the deadlines as a result of the faster algorithm.

If you're missing the deadline, you're A) slowing down the project (bad for the science) and B) possibly not getting credit for the WU (bad, if that's important to you.)


Because the card is not really fast i would try out the new Beta-App. - but all Wu's crashing with error within 2 seconds.

May be my card isn't supported for the new BETA-App?


Best RS

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14992 - Posted: 4 Feb 2010 | 17:29:47 UTC - in response to Message 14991.

It *would* help if you unhid your computers. Otherwise we're just guessing. :)

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 14994 - Posted: 4 Feb 2010 | 18:47:01 UTC - in response to Message 14992.

It *would* help if you unhid your computers. Otherwise we're just guessing. :)


done: http://www.gpugrid.net/hosts_user.php?userid=424

best RS

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14995 - Posted: 4 Feb 2010 | 19:02:50 UTC - in response to Message 14976.

3rd and 4th wu now running.

Cards are GTX260 (65nm) 216sp revision A2


Machine was off during day due to weather.

I suspended all the Einstein wu. This makes the wu progress a lot faster, so sharing a cpu core is clearly a major performance issue.

3rd wu completed successfully. Link to wu here

4th wu still running. However it too froze. A suspend/resume in BOINC got it going again. Also machine rebooted for no apparent reason. According to the Win7 event log it refers to an application error. Details are:

acemdbeta_6.08_windows_intelx86__cuda
0.0.0.0
4b680f5f
acemdbeta_6.08_windows_intelx86__cuda
0.0.0.0
4b680f5f
40000015
0003274d
a74
01caa5930a0dce07
C:\ProgramData\BOINC\projects\www.gpugrid.net\acemdbeta_6.08_windows_intelx86__cuda
C:\ProgramData\BOINC\projects\www.gpugrid.net\acemdbeta_6.08_windows_intelx86__cuda
c5ce415c-119f-11df-bbfe-00248c1ddc91
____________
BOINC blog

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 14996 - Posted: 4 Feb 2010 | 19:58:38 UTC - in response to Message 14994.

It *would* help if you unhid your computers. Otherwise we're just guessing. :)


done: http://www.gpugrid.net/hosts_user.php?userid=424

best RS


One thing I noticed about your machine:

Measured integer speed 1536.79 million ops/sec

You're running a Core2 CPU at a higher clock rate than I am, and my machine gets around 5000 for that benchmark. Same version of Windows, too.

So, why do you see about 25% of what I do on that benchmark? 25% happens to be the speed that most laptops cut the CPU to when it's going into power saving mode. Since the benchmarks will only run when BOINC is running, this means BOINC is running during the time that the laptop is taking steps to conserve power and/or reduce heat (pretty much the same thing).

It's possible, therefore, that the reason the WUs are failing is because of the power saving mode affecting the GPU. To test this, you need to go into the Windows power settings and make sure the laptop is always running at full power and then try running one of the beta WUs again.

This might be completely wrong, of course. But unless someone from the project says "nope, that card is not supported", at least this is something you could try.

Second observation:

The most recent stable version of the BOINC client is 6.10.18. Not sure if this has anything to do with your problem. I don't think it does, but it's a possibility. I'd install 6.10.18 anyway.

Third observation:

There's more recent versions of the video drivers. People have been saying that the 195 and 196 versions are slower, but 191 should be good. I'm running 191.07. Again, I don't think this is the cause of the problem.

P.S. That's a pretty fast CPU for a laptop, and a pretty good mobile GPU too. Nice machine you have there!
____________
Want to find one of the largest known primes? Try PrimeGrid. Or help cure disease at WCG.

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 14997 - Posted: 4 Feb 2010 | 21:01:38 UTC

Santa: I suspect you need a newer driver.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 15002 - Posted: 5 Feb 2010 | 7:27:15 UTC

thx @ all.

Will try as next the newer NVDIA-Driver.


Best RS

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 15004 - Posted: 5 Feb 2010 | 8:29:26 UTC

UPDATE:

after installing 195.62 it seems, that now my GPU could crunch the new beta-APP.

...til yet 9 minutes running without errors...


will watch out this ;)


Best RS

Profile Stoneageman
Avatar
Send message
Joined: 25 May 09
Posts: 224
Credit: 34,057,224,498
RAC: 190
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15007 - Posted: 5 Feb 2010 | 9:45:19 UTC

How soon before 6.08 becomes the standard application?

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15008 - Posted: 5 Feb 2010 | 9:50:10 UTC - in response to Message 15007.

For stability it seems fine. We have to check the results. Probably next week we will create a new application acemd2 which is not beta to go along side the old one.

gdf

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15014 - Posted: 5 Feb 2010 | 11:10:53 UTC - in response to Message 14995.

4th wu still running. However it too froze. A suspend/resume in BOINC got it going again. Also machine rebooted for no apparent reason. According to the Win7 event log it refers to an application error. Details are:

acemdbeta_6.08_windows_intelx86__cuda
0.0.0.0
4b680f5f
acemdbeta_6.08_windows_intelx86__cuda
0.0.0.0
4b680f5f
40000015
0003274d
a74
01caa5930a0dce07
C:\ProgramData\BOINC\projects\www.gpugrid.net\acemdbeta_6.08_windows_intelx86__cuda
C:\ProgramData\BOINC\projects\www.gpugrid.net\acemdbeta_6.08_windows_intelx86__cuda
c5ce415c-119f-11df-bbfe-00248c1ddc91


Well the 4th one finished. As above caused a reboot plus a couple of "not responding" popups. It managed to validate, despite the error messages. Wu can be found here
____________
BOINC blog

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15020 - Posted: 5 Feb 2010 | 13:07:36 UTC - in response to Message 15014.

This is the CUDA FFT bug on 260 cards.
g

Rabinovitch
Avatar
Send message
Joined: 25 Aug 08
Posts: 143
Credit: 64,937,578
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15021 - Posted: 5 Feb 2010 | 15:09:46 UTC - in response to Message 14545.

On windows it does not work well. We will come out with Linux at first.
gdf


Well, how about Linux' beta app now?

____________
From Siberia with love!

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15028 - Posted: 5 Feb 2010 | 19:27:40 UTC

this beta app for windows only? I'll try 3day to get beta WU on linux
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15029 - Posted: 5 Feb 2010 | 19:58:44 UTC - in response to Message 15028.

I was out of town.
Linux app coming soon, just the time to compile it.

gdf

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15030 - Posted: 5 Feb 2010 | 20:29:02 UTC - in response to Message 15029.

I was out of town.
Linux app coming soon, just the time to compile it.

gdf

let me know when it will be ready :-) GTX275 waiting for it :-)
____________

Tom Philippart
Send message
Joined: 12 Feb 09
Posts: 57
Credit: 23,376,686
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 15031 - Posted: 5 Feb 2010 | 21:53:21 UTC

I've had a WU failing today on the new app:

http://www.gpugrid.net/result.php?resultid=1808770

It happened while I was using the computer, the driver had to recover...

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15033 - Posted: 5 Feb 2010 | 23:54:18 UTC - in response to Message 15020.

This is the CUDA FFT bug on 260 cards.
g


Well I guess it didn't fix that then :-)
____________
BOINC blog

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15035 - Posted: 6 Feb 2010 | 4:25:28 UTC

So far have run 19 v6.08 WUs on 5 different GPUs (GTX 260 / GT 240 / 9600GSO). 18 were successful. The 1 failure was on a GT 240.
A CPU app errored just before the failure and I think it caused the GPU app to fail also.

Tom Philippart
Send message
Joined: 12 Feb 09
Posts: 57
Credit: 23,376,686
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 15036 - Posted: 6 Feb 2010 | 10:18:16 UTC - in response to Message 15031.

I've had a WU failing today on the new app:

http://www.gpugrid.net/result.php?resultid=1808770

It happened while I was using the computer, the driver had to recover...


as you can see ut us a 9600gt on vista x64

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 15042 - Posted: 6 Feb 2010 | 17:24:33 UTC
Last modified: 6 Feb 2010 | 17:35:45 UTC

Hi,

til yet 32,5h (94%) done .

... hope the last minutes are also successfull on the 9600M GT.

btw: a lot faster then the old app.. so you can edit the list of supported GPU's because the deadline is NOW no problem!


Best RS

Profile RealSanta
Send message
Joined: 17 Aug 07
Posts: 9
Credit: 347,550
RAC: 0
Level

Scientific publications
watwatwat
Message 15043 - Posted: 6 Feb 2010 | 19:48:12 UTC
Last modified: 6 Feb 2010 | 19:49:40 UTC

Update.

successfully done: http://www.gpugrid.net/result.php?resultid=1818728


Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 37854
Report deadline 10 Feb 2010 8:15:45 UTC
Run time 124240.524799
CPU time 4816.328
stderr out

<core_client_version>6.6.28</core_client_version>
<![CDATA[
<stderr_txt>
# There is 1 device supporting CUDA
# Device 0: "GeForce 9600M GT"
# Clock rate: 1.25 GHz
# Total amount of global memory: 536870912 bytes
# Number of multiprocessors: 4
# Number of cores: 32
MDIO ERROR: cannot open file "restart.coor"
# There is 1 device supporting CUDA
# Device 0: "GeForce 9600M GT"
# Clock rate: 1.25 GHz
# Total amount of global memory: 536870912 bytes
# Number of multiprocessors: 4
# Number of cores: 32
MDIO ERROR: cannot open file "restart.coor"
# Time per step: 198.761 ms
# Approximate elapsed time for entire WU: 124225.814 s
called boinc_finish

</stderr_txt>
]]>



One question: what happens/means here: " MDIO ERROR: cannot open file "restart.coor" " ?



best RS

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 15044 - Posted: 6 Feb 2010 | 19:53:25 UTC - in response to Message 15043.

That's normal. I believe it's stating that it can't find a checkpoint file, therefore it's starting from the beginning. This will always happen when starting a WU for the first time.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15047 - Posted: 7 Feb 2010 | 0:09:55 UTC

The beta is running great here on 4 different types of cards: GTX 260, GT 240, 8800 GT and 9600GSO.
My BIG problem is having to abort a boatload of v6.71 WUs in order to get the v6.08 ones.
Any chance of making this easier on us?

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15050 - Posted: 7 Feb 2010 | 9:03:11 UTC - in response to Message 15047.

We will move this beta to a standard acemd2 application as soon as possible.

gdf

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,606,436,851
RAC: 8,676,721
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15053 - Posted: 7 Feb 2010 | 9:51:37 UTC - in response to Message 15050.

We will move this beta to a standard acemd2 application as soon as possible.

gdf

Fine by me: reporting successful completion of a IBUCH_1000smd_pYEEI with the Beta app and a 9800GTX+.

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15056 - Posted: 7 Feb 2010 | 11:55:59 UTC
Last modified: 7 Feb 2010 | 13:15:25 UTC

Completed 2 successfully on a GTX275 with 196.21 drivers.

Link to wu 1 here and wu 2 here


And 2 more on a GTX295 with 196.21 drivers

Link to wu 1 here and wu 2 here

Also are we going to try the cuda 2.3 DLL's (ie plan class cuda23)? I don't think they will offer any speed advantage but you never know.
____________
BOINC blog

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15059 - Posted: 7 Feb 2010 | 17:10:27 UTC

MarJ - can you tell us a little about the macine/ os/ drivers/ anything else going on on your PC with the 295? Looking at the processing time / time per step for the two WUs you posted was painful ... Time per step: 95.802 ms

I have found that when two betas are running at the same time on the same 295 the processing is very slow and it looks like your machine is having the same issue. Any details you have might help provide some insight.

____________
Thanks - Steve

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15067 - Posted: 8 Feb 2010 | 9:43:59 UTC - in response to Message 15059.

Linux beta application uploaded.
gdf

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15069 - Posted: 8 Feb 2010 | 13:43:39 UTC - in response to Message 15067.

Linux beta application uploaded.
gdf

what I need to do? I changed "Run test applications?" to "yes" but still WUs for 6.70 app.
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15070 - Posted: 8 Feb 2010 | 13:58:48 UTC - in response to Message 15069.

Wait... for the right one.

Linux beta application uploaded.
gdf

what I need to do? I changed "Run test applications?" to "yes" but still WUs for 6.70 app.

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15071 - Posted: 8 Feb 2010 | 15:25:25 UTC - in response to Message 15070.

Wait... for the right one.

ok, let's wait
____________

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15072 - Posted: 8 Feb 2010 | 15:48:05 UTC - in response to Message 15047.
Last modified: 8 Feb 2010 | 15:58:32 UTC

The beta is running great here on 4 different types of cards: GTX 260, GT 240, 8800 GT and 9600GSO.
My BIG problem is having to abort a boatload of v6.71 WUs in order to get the v6.08 ones.
Any chance of making this easier on us?

Now up to 33 successful v6.08 WUs on 5 cards (GTX 260, GT 240, 8800 GT, 9600GSO and another GT 240). Still only the single error mentioned above and I think that was caused by something not related to GPUGRID (another project had put out some bad WUs and the bad CPU WU crashed pretty much everything including the GPUGRID WU). I think v6.08 is more stable than v6.71, and that wasn't bad.

Rabinovitch
Avatar
Send message
Joined: 25 Aug 08
Posts: 143
Credit: 64,937,578
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15074 - Posted: 8 Feb 2010 | 18:38:52 UTC - in response to Message 15067.
Last modified: 8 Feb 2010 | 18:39:59 UTC

Linux beta application uploaded.
gdf


Got one. 19 min = 0.704% Is it OK? GTS250, Ubuntu 9.10 x64, 195.30.
____________
From Siberia with love!

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15075 - Posted: 8 Feb 2010 | 20:20:29 UTC

I also got one. Later 2day there will be results.
____________

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15077 - Posted: 8 Feb 2010 | 22:12:19 UTC

1h01m - 2.704% only. Is it ok?
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15078 - Posted: 8 Feb 2010 | 22:36:57 UTC - in response to Message 15077.

For a GTX275 it seems quite slow.

gdf

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15080 - Posted: 8 Feb 2010 | 23:51:45 UTC - in response to Message 15078.
Last modified: 9 Feb 2010 | 0:10:18 UTC

For a GTX275 it seems quite slow.

gdf

2h41m - 9.024%. Something wrong...

my GTX275 is OC'd @702/1584/1260

upd.: sorry, I cancelled that WU... 3 hours and 10%...
____________

Rabinovitch
Avatar
Send message
Joined: 25 Aug 08
Posts: 143
Credit: 64,937,578
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15083 - Posted: 9 Feb 2010 | 2:07:03 UTC

8 hr, 12% (GTS250). Will it be finished untill the deadline?..
____________
From Siberia with love!

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15087 - Posted: 9 Feb 2010 | 9:33:53 UTC - in response to Message 15080.

For a GTX275 it seems quite slow.

gdf

2h41m - 9.024%. Something wrong...

my GTX275 is OC'd @702/1584/1260

upd.: sorry, I cancelled that WU... 3 hours and 10%...



Reboot after every driver-crash!

If the problems do not disappear:

Deinstall old drivers via windows, then boot into safe-mode, run Guru3d driver sweeper, reboot, and install the new drivers.

Profile [AF>Libristes>Jip] Elgran...
Avatar
Send message
Joined: 16 Jul 08
Posts: 45
Credit: 78,618,001
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15090 - Posted: 9 Feb 2010 | 10:28:35 UTC - in response to Message 15067.

Linux beta application uploaded.
gdf

I have got two workunits (1 for GTX275 and 1 for GTX260 (192sp)).
Let's crunch it.

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15092 - Posted: 9 Feb 2010 | 10:48:46 UTC - in response to Message 15087.

Reboot after every driver-crash!

If the problems do not disappear:

Deinstall old drivers via windows, then boot into safe-mode, run Guru3d driver sweeper, reboot, and install the new drivers.


Driver works fine.

And... I using linux. Should I install windows in order to reinstall driver? lol
____________

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15094 - Posted: 9 Feb 2010 | 11:52:37 UTC - in response to Message 15092.

Reboot after every driver-crash!

If the problems do not disappear:

Deinstall old drivers via windows, then boot into safe-mode, run Guru3d driver sweeper, reboot, and install the new drivers.


Driver works fine.

And... I using linux. Should I install windows in order to reinstall driver? lol


Then forget "driver sweeper" - and I replace "driver-crash" by "WU-crash".
It is a common advice to reboot after every failed WU.

(I am running dual-boot sytems (Win/Linux) on my whole "farm".)

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15100 - Posted: 9 Feb 2010 | 15:39:29 UTC - in response to Message 15094.
Last modified: 9 Feb 2010 | 15:40:38 UTC

It can be good advice to reboot after a failed WU; where a failure is not common! If a Beta fails, or a task fails after a few seconds, this is probably not the case, as some Betas are expected to fail, and some tasks can crash (bad batch) when running under some conditions.
So, if you do not normally get failures, and certainly if you get two or three, it is a good idea to reboot. This eliminates some possibilities of system or application errors causing a problem, and removes the possibility of runaway failures from such problems.

The longer a system is on, the more likely an odd or random problem with the operating system will occur. Many network, driver and application problems are solved by a simple reboot.

Profile [AF>Libristes>Jip] Elgran...
Avatar
Send message
Joined: 16 Jul 08
Posts: 45
Credit: 78,618,001
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15102 - Posted: 9 Feb 2010 | 17:24:20 UTC

8% of cpu utilization for gpu calculation is great with this ACEMD beta version v6.09 (cuda) Linux version.

Profile [AF>Libristes>Jip] Elgran...
Avatar
Send message
Joined: 16 Jul 08
Posts: 45
Credit: 78,618,001
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15105 - Posted: 9 Feb 2010 | 18:50:47 UTC - in response to Message 15102.

8% of cpu utilization for gpu calculation is great with this ACEMD beta version v6.09 (cuda) Linux version.

Here, my first wu calculated without problems.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1112
Credit: 6,162,416,256
RAC: 0
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15111 - Posted: 9 Feb 2010 | 23:22:58 UTC - in response to Message 15072.

The beta is running great here on 4 different types of cards: GTX 260, GT 240, 8800 GT and 9600GSO.
My BIG problem is having to abort a boatload of v6.71 WUs in order to get the v6.08 ones.
Any chance of making this easier on us?

Now up to 33 successful v6.08 WUs on 5 cards (GTX 260, GT 240, 8800 GT, 9600GSO and another GT 240). Still only the single error mentioned above and I think that was caused by something not related to GPUGRID (another project had put out some bad WUs and the bad CPU WU crashed pretty much everything including the GPUGRID WU). I think v6.08 is more stable than v6.71, and that wasn't bad.

Up to 50 successful v6.08 with only the one failure mentioned above :-)
That's on a varied mix of GPUs. Older cards seem to benefit as much as the G200 series.

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15116 - Posted: 10 Feb 2010 | 9:24:29 UTC - in response to Message 15100.

It can be good advice to reboot after a failed WU; where a failure is not common! If a Beta fails, or a task fails after a few seconds, this is probably not the case, as some Betas are expected to fail, and some tasks can crash (bad batch) when running under some conditions.
So, if you do not normally get failures, and certainly if you get two or three, it is a good idea to reboot. This eliminates some possibilities of system or application errors causing a problem, and removes the possibility of runaway failures from such problems.

The longer a system is on, the more likely an odd or random problem with the operating system will occur. Many network, driver and application problems are solved by a simple reboot.


You are completely right. I only add the information that a slowdown of computation is a known "runaway failure" from cuda failures.

Look here for example.

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15133 - Posted: 10 Feb 2010 | 23:14:34 UTC - in response to Message 15116.
Last modified: 10 Feb 2010 | 23:15:19 UTC

I only add the information that a slowdown of computation is a known "runaway failure" from cuda failures.


you meant that the reason for massive slowdown of 6.09 app happened to me to be fixed by restarting computer?

I don't so coz after I aborted 6.09 WUs I've got 6.70 ones. And u know what? they worked w/o any slowdown: 6 hours and the job done (as usual). And furthermore - to complete 10% of 6.09 WU took me 3 hours, so I need 30 hours to complete the whole WU, i.e. 6.09 WU is 5 times slower. Taking in consideration that new app is 1.6 time faster, my card was 8 time slower then it should be.


I think that smth wrong with linux app, i.e. 6.09
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15134 - Posted: 10 Feb 2010 | 23:41:38 UTC - in response to Message 15133.
Last modified: 10 Feb 2010 | 23:42:10 UTC

the main change for Linux is that now the new app uses the same driver mechanism to reduce CPU usage of Windows. Maybe the problem is in your driver or Linux kernel. The application under Linux does normally work as expected, fast.

gdf

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15136 - Posted: 11 Feb 2010 | 3:49:42 UTC
Last modified: 11 Feb 2010 | 3:50:01 UTC

GDF,
system: ubuntu 9.10 x64,
kernel - at that time - 2.6.31-19,
driver - 195.30
cuda 2.3

what's necessary to change?

BTW, I tried to "catch" 6.09 - no luck :(
____________

Rabinovitch
Avatar
Send message
Joined: 25 Aug 08
Posts: 143
Credit: 64,937,578
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15142 - Posted: 11 Feb 2010 | 9:48:04 UTC

Completed. Is such a run time OK?!
____________
From Siberia with love!

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15143 - Posted: 11 Feb 2010 | 11:01:30 UTC - in response to Message 15142.

No, it's too slow.
Are you sure that your card is running at full speed during these runs?

gdf

Profile liveonc
Avatar
Send message
Joined: 1 Jan 10
Posts: 292
Credit: 41,567,650
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwat
Message 15144 - Posted: 11 Feb 2010 | 11:23:38 UTC

141-IBUCH_1000smd_pYEEI_100202-2-10-RND7134_1 was the first ACEMD beta version v6.09 (cuda)that I ran on a Linux PC (Ubuntu/Mint64bit). Damn it took forever! http://www.gpugrid.net/result.php?resultid=1835513
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15147 - Posted: 11 Feb 2010 | 12:58:07 UTC - in response to Message 15144.

Maybe there is a specific problem with Ubuntu.
We have run fine on CentOS and Fedora.

We can reintroduce the same mechanism as before for Linux.

gdf

Rabinovitch
Avatar
Send message
Joined: 25 Aug 08
Posts: 143
Credit: 64,937,578
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15149 - Posted: 11 Feb 2010 | 18:27:40 UTC

"Maximum performance", as usually... But it's really SLOWWW.....

Ohhh, that diversity of Linuxes!.. Ohh, that Ubuntu!..

Crunch3r said: "Any Linux Distribution is better than Ubuntu and all it's derivates since it causes to many trouble due their weired kernel experiments.

Personally i use Gentoo and am running openSuSE on production servers.
Fedora is a good choice as well."

Fedora's ISO is waiting the Ubuntu's first full crash to take it's place...

Or may be I'll switch back to Debian.
____________
From Siberia with love!

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15159 - Posted: 11 Feb 2010 | 23:15:00 UTC
Last modified: 12 Feb 2010 | 0:11:41 UTC

looks I'm not alone facing the slowdown with 6.09 ...

I don't think hat the problem is ubuntu kernel, otherwise why 6.70 works just fine???

update.
I've just got new app - "acemd2_6.02_x86_64-pc-linux-gnu__cuda"

There is progress in comparison with 6.09 - 18.5-19 hours against 30 hours (6.09), but still worse then 6.70.

BUT: CPU usage - ZERO :-) That's awesome :-)
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15166 - Posted: 12 Feb 2010 | 9:34:25 UTC - in response to Message 15159.

6.70 and 6.09 are exactly different in the use of CPU. So it could be the kernel.

gdf

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15180 - Posted: 12 Feb 2010 | 15:24:39 UTC

I put "No' to "run test apps" option, but I'm still getting 6.02 apps. I aborted 2 of them and now
"Fri 12 Feb 2010 10:22:03 AM EST GPUGRID Message from server: Project has no jobs available"

and again: 0h59m and 2.3% only...
____________

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15203 - Posted: 13 Feb 2010 | 6:00:34 UTC

I'm still receiving 6.02... How I can avoid this?
____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15205 - Posted: 13 Feb 2010 | 9:39:48 UTC - in response to Message 15203.

Well you should stop receiving it if you set "no test application".
It's odd that you keel receiving those jobs.
Anyway, that queue is empty.

gdf

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15278 - Posted: 17 Feb 2010 | 23:38:38 UTC - in response to Message 15205.
Last modified: 17 Feb 2010 | 23:41:21 UTC

today i've got 6.04 app even "No" to "run test apps" option is still there.

In terms of the speed it looks like 6.70 app (0h15min and 6.384%) but consumes the whole core... Temperature is 4-5 *C higher then while 6.70 app runs (68 against 63-64), but that's not big deal.
____________

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15282 - Posted: 18 Feb 2010 | 0:39:56 UTC - in response to Message 15278.

The 6.03 (Windows) apps seem to be similar to before. Core usage is normal (about 28% of one of my 4 Phenom II 940 cores supporting an ATX260sp216, less for my i7 supporting 2 GT240s).

Profile [AF>Libristes>Jip] Elgran...
Avatar
Send message
Joined: 16 Jul 08
Posts: 45
Credit: 78,618,001
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15293 - Posted: 18 Feb 2010 | 12:40:04 UTC - in response to Message 15278.
Last modified: 18 Feb 2010 | 12:40:52 UTC

today i've got 6.04 app even "No" to "run test apps" option is still there.

In terms of the speed it looks like 6.70 app (0h15min and 6.384%) but consumes the whole core... Temperature is 4-5 *C higher then while 6.70 app runs (68 against 63-64), but that's not big deal.

Same problem with this version 6.04, 91% of one core of my cpu used for the calculation, it's unbelievable.
It's worse than the previous 6.70 version 50% of one core of my cpu.
This project becomes more and more elitist : GTX260 bug, GTX295 (several compute errors, and another bug).
I am very tired of it.

CTAPbIi
Send message
Joined: 29 Aug 09
Posts: 175
Credit: 259,509,919
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15313 - Posted: 18 Feb 2010 | 18:28:02 UTC - in response to Message 15293.

Same problem with this version 6.04, 91% of one core of my cpu used for the calculation, it's unbelievable.
It's worse than the previous 6.70 version 50% of one core of my cpu.
This project becomes more and more elitist : GTX260 bug, GTX295 (several compute errors, and another bug).
I am very tired of it.

well, in my case 6.04 going the job and it's OK. willing to get new 6.10/6.11 test app

____________

Profile robertmiles
Send message
Joined: 16 Apr 09
Posts: 503
Credit: 727,920,933
RAC: 155,858
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15399 - Posted: 24 Feb 2010 | 1:09:14 UTC - in response to Message 15020.
Last modified: 24 Feb 2010 | 1:29:42 UTC

This is the CUDA FFT bug on 260 cards.
g


Can we have more information on how to tell which 260 cards are affected, and under which versions of CUDA?

For example, I now have CUDA 3.0 available and haven't seen anything on whether this version is affected.

196.21 driver, compute capability 1.1, BOINC 6.10.18, not a 260 card

I think I remember seeing something saying that CUDA 2.1 is not affected, and that 280 cards with the same graphics chip as the 260 cards (except no shaders disabled) work just fine.

Is is possible to build an alternate version of the application using at least part of the CUDA 2.1 SDK instead, or is too much missing from that SDK? If it's possible, that version could be used on 260 cards and any other cards no longer adequately supported by recent CUDA versions.

Is it possible to build an alternate version using the CUDA 3.0 SDK, at least for testing whether that fixes any known problems?

Can you ask Nvidia to check whether recent versions of CUDA leave out the part needed for proper handling of the way shaders were disabled on the early chips used on some 260 cards?

Also, is there a way to ask the graphics chip for details about what type of graphics chip was used, including how many shaders were disabled and perhaps even which ones? Checking that information might allow better information on which 260 cards are affected.

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15400 - Posted: 24 Feb 2010 | 7:47:54 UTC - in response to Message 15399.

This is the CUDA FFT bug on 260 cards.
g


Can we have more information on how to tell which 260 cards are affected, and under which versions of CUDA?


Download a copy of GPU-Z and run it. It will tell you various details about your gpu. If it says you have a 65nm GTX260 with a G200 chip then it has the FFT bug courtesy of nVidia. Driver versions above 181.21 have the bug.

The fact you have a cuda 3.0 capable driver doesn't mean the app is using the 3.0 DLL's as its designed to be able to run older apps. The current apps use either cuda 2.2 or cuda 2.3 (see the plan class shown in brackets in the app name, under the tasks tab in BOINC manager).

The cuda 3.0 DLL's are currently only available to registered developers and cannot be given to the public due to the pending release of the Fermi cards. Once they are released it will be publicly available. Given that nvidia doesn't seem willing or able to correct the bug suggests that even cuda 3.0 apps are still going to have the same problem, but you never know.
____________
BOINC blog

Profile robertmiles
Send message
Joined: 16 Apr 09
Posts: 503
Credit: 727,920,933
RAC: 155,858
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15416 - Posted: 25 Feb 2010 | 1:43:50 UTC - in response to Message 15399.
Last modified: 25 Feb 2010 | 2:01:51 UTC

I still don't have an answer to this part:

Is is possible to build an alternate version of the application using at least part of the CUDA 2.1 SDK instead, or is too much missing from that SDK? If it's possible, that version could be used on 260 cards and any other cards no longer adequately supported by recent CUDA versions.


Are you still trying to find out yourself?

It would require those with the affected 260 and other cards to install rather old drivers, but wouldn't that be better than not being able to use their cards at all?

Rabinovitch
Avatar
Send message
Joined: 25 Aug 08
Posts: 143
Credit: 64,937,578
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 15527 - Posted: 1 Mar 2010 | 9:57:34 UTC

Hi all!

Is there a way to make ACEMD 6.04 (Linux app) run by default with the same (lowest) priority as other BOINC applications? It's not true way to set the priority by hands each time, I suppose...
____________
From Siberia with love!

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15528 - Posted: 1 Mar 2010 | 10:46:54 UTC - in response to Message 15527.

all the application should run with the same system priority. This is set by BOINC.
gdf

Profile Michael Goetz
Avatar
Send message
Joined: 2 Mar 09
Posts: 124
Credit: 46,573,744
RAC: 837,894
Level
Val
Scientific publications
watwatwatwatwatwatwatwat
Message 15531 - Posted: 1 Mar 2010 | 14:10:59 UTC - in response to Message 15527.

Hi all!

Is there a way to make ACEMD 6.04 (Linux app) run by default with the same (lowest) priority as other BOINC applications? It's not true way to set the priority by hands each time, I suppose...


Short answer: not really.

BOINC sets the priorities for all the WUs to the same (low) value, except for GPU WUs. The CPU portion of a GPU WU is set to a slightly higher priority than the pure CPU WUs so that it runs ahead of the CPU tasks. The intent is to keep the GPU from being starved of work by the other BOINC WUs.

If you're running any other CPU-only projects on your machine, you don't want to lower the priority anyway, at least not unless you dedicate a CPU core to feeding the GPU.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15534 - Posted: 1 Mar 2010 | 15:45:52 UTC - in response to Message 15531.

Thanks, I had forgotten about it.

gdf

Post to thread

Message boards : Graphics cards (GPUs) : New nvidia beta application

//