Advanced search

Message boards : News : Geforce 10 / Pascal app coming soon

Author Message
Profile MJH
Project administrator
Project developer
Project scientist
Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 44031 - Posted: 22 Jul 2016 | 11:00:54 UTC

hi Crunchers

Just to let you know that there'll be a new app coming to support the ascal GPUs just as soon as Nvidia make a public release of CUDA 8.

Matt

Jim1348
Send message
Joined: 28 Jul 12
Posts: 460
Credit: 1,130,761,180
RAC: 18,722
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwat
Message 44032 - Posted: 22 Jul 2016 | 12:23:41 UTC - in response to Message 44031.

Will it help out the Maxwell cards?

The drivers ever since 361.43, released on December 21, 2015 are said to have included CUDA 8, for whatever that is worth. But as posted elsewhere, the earlier CUDA 7.5 driver are faster on the old app, so I wonder whether it would be useful to upgrade to a recent driver for the Maxwell cards.

Lluis
Send message
Joined: 22 Feb 14
Posts: 23
Credit: 424,878,457
RAC: 597,341
Level
Gln
Scientific publications
watwatwatwatwatwatwat
Message 44034 - Posted: 22 Jul 2016 | 13:25:55 UTC - in response to Message 44031.

hi Crunchers

Just to let you know that there'll be a new app coming to support the ascal GPUs just as soon as Nvidia make a public release of CUDA 8.

Matt


Thank you.

Eagerly awaiting !!!

Profile [CSF] Georg Stoifl
Avatar
Send message
Joined: 15 Nov 12
Posts: 10
Credit: 461,401,419
RAC: 442,307
Level
Gln
Scientific publications
watwatwatwatwatwatwat
Message 44040 - Posted: 24 Jul 2016 | 20:47:47 UTC

+1 My Geforce 1070 can't await it... :-)

Michael P. Gainor
Send message
Joined: 23 Jan 16
Posts: 15
Credit: 69,388,188
RAC: 0
Level
Thr
Scientific publications
wat
Message 44042 - Posted: 25 Jul 2016 | 16:36:44 UTC

Anythoughts if this will incress GPU load percentage? I find that the GPU load is quite low compared to Folding@home.

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 588
Credit: 2,037,147,925
RAC: 1,495,102
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44043 - Posted: 25 Jul 2016 | 16:54:50 UTC - in response to Message 44042.

Anythoughts if this will incress GPU load percentage? I find that the GPU load is quite low compared to Folding@home.


If you go back to the 359.6 driver your gpu load will rise by between 4% to 5%

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44135 - Posted: 11 Aug 2016 | 4:44:28 UTC - in response to Message 44031.

hi Crunchers

Just to let you know that there will be a new app coming to support the Pascal GPUs just as soon as NVIDIA make a public release of CUDA 8.

Matt

27 May 2016
https://www.phoronix.com/scan.php?page=news_item&px=CUDA-8-Release-Candidate

https://developer.nvidia.com/cuda-toolkit
____________
1 Corinthians 9:16 "For though I preach the gospel, I have nothing to glory of: for necessity is laid upon me; yea, woe is unto me, if I preach not the gospel!"
Ephesians 6:18-20, please ;-)
http://tbc-pa.org

Profile skgiven
Volunteer moderator
Project tester
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,834,518,624
RAC: 292,156
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44138 - Posted: 11 Aug 2016 | 20:35:52 UTC - in response to Message 44135.

Yeah, the Release Candidate has been around for some time, but you can't legally use it in publicly available apps...
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44141 - Posted: 11 Aug 2016 | 23:18:28 UTC - in response to Message 44138.

Yeah, the Release Candidate has been around for some time, but you can't legally use it in publicly available apps...

Judging by the time went by since then it's probably had a couple bugs to fix.
I recall that there were issues even with the CUDA subsystem of released drivers in the past.

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44155 - Posted: 12 Aug 2016 | 20:44:32 UTC - in response to Message 44141.

It could still be used for building the code and working out some bugs in house then release the full product with very minimal testing once a bug free final CUDA 8 is released. I just don't want to see us waiting til a public official release comes out before touching it at all, which I assume is not happening anyway.

Profile skgiven
Volunteer moderator
Project tester
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,834,518,624
RAC: 292,156
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44169 - Posted: 14 Aug 2016 | 19:22:47 UTC - in response to Message 44155.

That has been happening. If there were no bugs then the Cuda 8 dev kit would have already been released for public use.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

eXaPower
Send message
Joined: 25 Sep 13
Posts: 265
Credit: 1,043,017,367
RAC: 1,762,909
Level
Met
Scientific publications
watwatwatwatwatwat
Message 44190 - Posted: 16 Aug 2016 | 11:26:40 UTC

Bulletin for crunchers with Pascal GPU's and XP:

Microsoft Windows XP/Windows XP 64-bit driver support for GeForce GPUs has been deprecated with driver branch R370_00 and onward. Driver 368.81 is the last driver to support Windows XP/Windows XP 64-bit.


http://us.download.nvidia.com/Windows/372.54/372.54-win10-win8-win7-desktop-release-notes.pdf

https://forums.geforce.com/default/topic/957568/geforce-drivers/official-372-54-game-ready-whql-display-driver-feedback-thread-released-8-15-16-/1/


xixou
Send message
Joined: 8 Jun 14
Posts: 18
Credit: 18,757,357
RAC: 0
Level
Pro
Scientific publications
wat
Message 44198 - Posted: 16 Aug 2016 | 20:48:24 UTC - in response to Message 44190.
Last modified: 16 Aug 2016 | 20:49:42 UTC

Can we crunch with today nvidia drivers on win10 on pascal gpu ? are short run available to test ? Or only long run ?

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2671
Credit: 750,826,474
RAC: 478,673
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44200 - Posted: 16 Aug 2016 | 20:51:38 UTC - in response to Message 44198.

Not yet, but I'd like to ask pretty much the same thing: today driver 372.54 was released, including official CUDA 8 support. Is that enough yet?

MrS
____________
Scanning for our furry friends since Jan 2002

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44201 - Posted: 16 Aug 2016 | 21:32:01 UTC - in response to Message 44190.

Bulletin for crunchers with Pascal GPU's and XP:

Microsoft Windows XP/Windows XP 64-bit driver support for GeForce GPUs has been deprecated with driver branch R370_00 and onward. Driver 368.81 is the last driver to support Windows XP/Windows XP 64-bit.

I thought that this will happen. It's time for me to install Linux on one of my hosts.

xixou
Send message
Joined: 8 Jun 14
Posts: 18
Credit: 18,757,357
RAC: 0
Level
Pro
Scientific publications
wat
Message 44207 - Posted: 17 Aug 2016 | 5:01:10 UTC

Still failing with 372.54 Nvidia drivers and gtx 1070/1060

17/08/2016 07:02:51 | | Starting BOINC client version 7.6.22 for windows_x86_64
17/08/2016 07:02:51 | | log flags: file_xfer, sched_ops, task, sched_op_debug
17/08/2016 07:02:51 | | Libraries: libcurl/7.45.0 OpenSSL/1.0.2d zlib/1.2.8
17/08/2016 07:02:51 | | Data directory: C:\boinc_data
17/08/2016 07:02:51 | | Running under account lee_l
17/08/2016 07:02:52 | | CUDA: NVIDIA GPU 0: GeForce GTX 1070 (driver version 372.54, CUDA version 8.0, compute capability 6.1, 4096MB, 3562MB available, 6998 GFLOPS peak)
17/08/2016 07:02:52 | | CUDA: NVIDIA GPU 1: GeForce GTX 1060 6GB (driver version 372.54, CUDA version 8.0, compute capability 6.1, 4096MB, 3566MB available, 4374 GFLOPS peak)
17/08/2016 07:02:52 | | CUDA: NVIDIA GPU 2: GeForce GTX 1070 (driver version 372.54, CUDA version 8.0, compute capability 6.1, 4096MB, 3562MB available, 6998 GFLOPS peak)
17/08/2016 07:02:52 | | OpenCL: NVIDIA GPU 0: GeForce GTX 1070 (driver version 372.54, device version OpenCL 1.2 CUDA, 8192MB, 3562MB available, 6998 GFLOPS peak)
17/08/2016 07:02:52 | | OpenCL: NVIDIA GPU 1: GeForce GTX 1060 6GB (driver version 372.54, device version OpenCL 1.2 CUDA, 6144MB, 3566MB available, 4374 GFLOPS peak)
17/08/2016 07:02:52 | | OpenCL: NVIDIA GPU 2: GeForce GTX 1070 (driver version 372.54, device version OpenCL 1.2 CUDA, 8192MB, 3562MB available, 6998 GFLOPS peak)
17/08/2016 07:02:52 | | Host name: xixou-win10
17/08/2016 07:02:52 | | Processor: 6 GenuineIntel Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz [Family 6 Model 63 Stepping 2]
17/08/2016 07:02:52 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 dca pbe fsgsbase bmi1 smep bmi2
17/08/2016 07:02:52 | | OS: Microsoft Windows 10: x64 Edition, (10.00.10586.00)
17/08/2016 07:02:52 | | Memory: 31.91 GB physical, 32.01 GB virtual
17/08/2016 07:02:52 | | Disk: 465.21 GB total, 41.05 GB free
17/08/2016 07:02:52 | | Local time is UTC +2 hours
17/08/2016 07:02:52 | | VirtualBox version: 5.0.12

17/08/2016 07:02:55 | GPUGRID | Starting task e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0
17/08/2016 07:02:55 | pogs | Starting task J010727.2+321911_area31916982_0
17/08/2016 07:02:57 | GPUGRID | [sched_op] Deferring communication for 00:01:27
17/08/2016 07:02:57 | GPUGRID | [sched_op] Reason: Unrecoverable error for task e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0
17/08/2016 07:02:57 | GPUGRID | Computation for task e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0 finished
17/08/2016 07:02:57 | GPUGRID | Output file e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0_1 for task e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0 absent
17/08/2016 07:02:57 | GPUGRID | Output file e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0_2 for task e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0 absent
17/08/2016 07:02:57 | GPUGRID | Output file e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0_3 for task e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0 absent

17/08/2016 07:03:00 | GPUGRID | Started upload of e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0_0
17/08/2016 07:03:00 | GPUGRID | Started upload of e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0_7
17/08/2016 07:03:01 | GPUGRID | Finished upload of e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0_0
17/08/2016 07:03:01 | GPUGRID | Finished upload of e39s6_e28s20p0f482-GERARD_FXCXCL12RX_1153966_1-0-1-RND9279_0_7

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44208 - Posted: 17 Aug 2016 | 6:34:47 UTC - in response to Message 44207.

Still failing with 372.54 Nvidia drivers and gtx 1070/1060

Yes, it is not the drivers on your computer that counts at this point. It is the programing of the tasks to be done on that new driver (CUDA version) needed for the new cards. NVIDIA has to complete, test, and release a bug free version (a gold version) of the CUDA SDK for GPUGRID to make the code in the project work with your card and driver. The ball on this is in the NVIDIA court currently to release that after making a version that they feel confident in releasing.

Really, most companies that make an actual product makes the drivers and the SDK available to manufacturers of the products that work with their product and software companies that make software to work with that product before releasing the product and its user end drivers. In the case of this whole line of products NVIDIA seems to have done this backwards and released the product, the user drivers and not the toolkit to make products and software to use the GTX10 line of cards and the CUDA 8 platform.

So until NVIDIA releases that, the tasks coming out of GPUGRID will continue to fail on these new cards. That is unless GPUGRID uses the buggy/non-gold SDK available to compile new tasks risking more failed tasks for the sake of the processing power going untapped. I would think that unless they actually release the amount of work that surpasses the need for that extra unused power in the installed cards that want to crunch GPUGRID, they should not take that risk and should just let the NVIDIA timeline dictate the use of the cards. If they can release enough work to surpass the user's current non-GTX10 cards and actually need the GTX10 cards in production, then I would say there is even a discussion to be had on the matter of taking that risk.

gianni
Send message
Joined: 11 Jul 08
Posts: 16
Credit: 105,098
RAC: 0
Level

Scientific publications
watwatwat
Message 44209 - Posted: 17 Aug 2016 | 10:31:55 UTC - in response to Message 44208.

We have received information that one important bug for us has been fixed.

You should expect new applications compatible with pascal soon after (a day) cuda8 is released. Initially for Linux and then for Windows (maybe a week later depending on how easy was the Linux release).

gdf

xixou
Send message
Joined: 8 Jun 14
Posts: 18
Credit: 18,757,357
RAC: 0
Level
Pro
Scientific publications
wat
Message 44213 - Posted: 17 Aug 2016 | 13:07:57 UTC - in response to Message 44210.

We have received information that one important bug for us has been fixed.

You should expect new applications compatible with pascal soon after (a day) cuda8 is released. Initially for Linux and then for Windows (maybe a week later depending on how easy was the Linux release).

gdf

great ! hoppefully long runs are not flagged as error as it was sometime the case on titan x maxwell. short runs were always ok.

xixou
Send message
Joined: 8 Jun 14
Posts: 18
Credit: 18,757,357
RAC: 0
Level
Pro
Scientific publications
wat
Message 44278 - Posted: 24 Aug 2016 | 4:44:44 UTC - in response to Message 44209.

We have received information that one important bug for us has been fixed.

You should expect new applications compatible with pascal soon after (a day) cuda8 is released. Initially for Linux and then for Windows (maybe a week later depending on how easy was the Linux release).

gdf


Hi Gianni,
How went the linux port ?

KSUMatt
Avatar
Send message
Joined: 11 Jan 13
Posts: 214
Credit: 831,004,493
RAC: 0
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwat
Message 44322 - Posted: 29 Aug 2016 | 18:49:55 UTC

Just checking in. I have a 1080 and 1070 waiting for a Windows app. They've been busy on Einstein in the meantime.

Profile MJH
Project administrator
Project developer
Project scientist
Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 44342 - Posted: 30 Aug 2016 | 10:01:40 UTC - in response to Message 44322.

Just checking in. I have a 1080 and 1070 waiting for a Windows app. They've been busy on Einstein in the meantime.


Still waiting on the release of CUDA 8.5

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44438 - Posted: 6 Sep 2016 | 11:34:43 UTC - in response to Message 44342.

Just checking in. I have a 1080 and 1070 waiting for a Windows app. They've been busy on Einstein in the meantime.


Still waiting on the release of CUDA 8.5

This is worse than I've feared.
A quick google search of "CUDA 8.5" gave the following two results as best match:
http://www.cudabrand.com/cuda-8-5-titanium-bonded-bent-needle-nose-pliers.html
http://www.cudabrand.com/cuda-8-5-titanium-bonded-dehooker.html
which left me unimpressed.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1073
Credit: 4,496,456,504
RAC: 414,796
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44439 - Posted: 6 Sep 2016 | 13:46:14 UTC - in response to Message 44438.

Still waiting on the release of CUDA 8.5

This is worse than I've feared.
A quick google search of "CUDA 8.5" gave the following two results as best match:
http://www.cudabrand.com/cuda-8-5-titanium-bonded-bent-needle-nose-pliers.html
http://www.cudabrand.com/cuda-8-5-titanium-bonded-dehooker.html
which left me unimpressed.

Zoltan, there has to be a "catch and release" joke in there somewhere...

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44441 - Posted: 6 Sep 2016 | 14:34:04 UTC - in response to Message 44439.

HAH!

eXaPower
Send message
Joined: 25 Sep 13
Posts: 265
Credit: 1,043,017,367
RAC: 1,762,909
Level
Met
Scientific publications
watwatwatwatwatwat
Message 44452 - Posted: 7 Sep 2016 | 10:13:37 UTC - in response to Message 44438.

Just checking in. I have a 1080 and 1070 waiting for a Windows app. They've been busy on Einstein in the meantime.


Still waiting on the release of CUDA 8.5

This is worse than I've feared.
A quick google search of "CUDA 8.5" gave the following two results as best match:
http://www.cudabrand.com/cuda-8-5-titanium-bonded-bent-needle-nose-pliers.html
http://www.cudabrand.com/cuda-8-5-titanium-bonded-dehooker.html
which left me unimpressed.

Speculation: CUDA 8.5 publicly available near (GP107) GTX 1050 release (October) or further down the road when (GP102) GTX 1080ti appears in early 2017 or late 2016.

http://ambermd.org/gpus/benchmarks.htm

CUDA 8.0 AMBER benchmarks show what we're missing in potential throughput with Pascal.
For example: GTX 1070 Single job throughput equal to Maxwell Titan X or GTX 980ti.

klepel
Send message
Joined: 23 Dec 09
Posts: 136
Credit: 1,829,755,020
RAC: 1,443,951
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44458 - Posted: 8 Sep 2016 | 3:07:29 UTC

My strategy seems to work: I restricted my purchase of a latest generation Nvidia GPU to the availability of a working app for this generation on my priority project GPUGRID. The price of these cards goes in the right direction… down:-)

Lesson learned from the GTX 670 introduction…

DigitalParasite
Send message
Joined: 20 Jul 16
Posts: 3
Credit: 158,913,025
RAC: 348,284
Level
Ile
Scientific publications
wat
Message 44472 - Posted: 10 Sep 2016 | 10:39:26 UTC

I must be missing an important detail, but is there a reason why the GPUGRID code for NVIDIA GPUs will not run on the new Pascal cards? Is the hardware not backwards compatible that it can run older code like other projects such as Einstein or POEM which have been working without the release of CUDA 8.0/8.5 support on the 1080/1070?

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44473 - Posted: 10 Sep 2016 | 11:18:49 UTC - in response to Message 44472.

...is there a reason why the GPUGRID code for NVIDIA GPUs will not run on the new Pascal cards?

Judging only by the error messages the app throws, it uses different libraries for different Compute Capabilities (to maximize efficiency I guess), and the one which is necessary for CC6.1 is simply not present in the CUDA6.5 code (obviously because there was no CC6.1 back then when the CUDA6.5 came out).

Is the hardware not backwards compatible that it can run older code

They are backwards compatible, and they could run older code.

... it can run older code like other projects such as Einstein or POEM which have been working without the release of CUDA 8.0/8.5 support on the 1080/1070?

Probably the CUDA 6.5 code could be made to work on the Pascals, but when you do that, your app could be missing a significant part of the hardware improvements, thus it could be slower and/or less energy efficient than a new code.

xixou
Send message
Joined: 8 Jun 14
Posts: 18
Credit: 18,757,357
RAC: 0
Level
Pro
Scientific publications
wat
Message 44474 - Posted: 10 Sep 2016 | 16:18:54 UTC - in response to Message 44473.

I would say better less than 100% efficient but to have at least something.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44475 - Posted: 10 Sep 2016 | 18:34:30 UTC - in response to Message 44474.

I would say better less than 100% efficient but to have at least something.

A fellow cruncher called the CUDA6.5 client "ancient" in another thread. Other projects go the way according to the "to have at least something" mentality, resulting in CUDA3.2, CUDA5.0, and CUDA5.5 apps. What would that fellow call those clients? This project has chosen the other way: to have state of the art app (because it's using a proprietary code provided by Acellera, so it has to be competitive). To have clients serving both mentalities requires doubled development teams, but it's gratifying to have at least one such team. Thus there always could be a group of people who can pick on the way a project goes, but I'm quite satisfied with the way this project goes, however the pace could be faster, but it's limited by 3rd party (NVidia).

jlhal
Send message
Joined: 1 Mar 10
Posts: 147
Credit: 1,077,535,540
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44476 - Posted: 10 Sep 2016 | 19:00:14 UTC - in response to Message 44475.

Hi everybody !
Excuse my "newbie" question but would it be possible that all apps be written with OpenCL in mind instead of proprietary CUDA ?
____________
Lubuntu 16.04.1 LTS x64

Tomas Brada
Send message
Joined: 3 Nov 15
Posts: 37
Credit: 1,934,925
RAC: 293
Level
Ala
Scientific publications
wat
Message 44489 - Posted: 12 Sep 2016 | 18:35:36 UTC - in response to Message 44476.

Excuse my "newbie" question but would it be possible that all apps be written with OpenCL in mind instead of proprietary CUDA ?

That would be ideal to have all applications in OpenCL.
But the acemd application is written with CUDA and there is no intention to rewrite it to OpenCL, which is not trivial task and part of it code is not open so not even possible. On the other hand some benchmarks show that OpenCL is bit slower on nvidia cards (citation-needed).
____________

jlhal
Send message
Joined: 1 Mar 10
Posts: 147
Credit: 1,077,535,540
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44501 - Posted: 12 Sep 2016 | 21:44:22 UTC - in response to Message 44489.

Thanks, I understand your answer but "the bit slower" could be forgotten if AMD GPUs could also crunch these WUs because more GPUs crunching at the same time means that the batch is sooner finished !

I'm planning to buy a GTX 10 series card.
For Lubuntu I have to install a driver from ppa:graphics-drivers/ppa because the actual Xenial distrib is limited to 361.42 version and this does not support GTX 10 series
After that, I'll have to install the latest release of CUDA either from this ppa (actually only 7.5 is availale in Xenial, or from Nvidia .

____________
Lubuntu 16.04.1 LTS x64

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44513 - Posted: 13 Sep 2016 | 7:20:13 UTC - in response to Message 44501.

Two things.

1) They have been testing with a Python app to run on the GPU so it can port to AMD GPUs. This MAY be, if successful, also a port to native Intel chipset GPUs as well.

2) distriuted.net has an OpenCL and a CUDA client and the OpenCL runs a lot more keys per second than the CUDA client on the same NVIDIA GPU. Now this may be a difference in the way dnetc and a true GPU client may work or it may be a good benchmark, I am not sure, but that is what I have found when experimenting.
____________
1 Corinthians 9:16 "For though I preach the gospel, I have nothing to glory of: for necessity is laid upon me; yea, woe is unto me, if I preach not the gospel!"
Ephesians 6:18-20, please ;-)
http://tbc-pa.org

Werkstatt
Send message
Joined: 23 May 09
Posts: 116
Credit: 122,267,748
RAC: 35,976
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44517 - Posted: 13 Sep 2016 | 22:50:49 UTC

Maybe this could be of interest:
https://software.intel.com/en-us/blogs/2016/09/08/intel-distribution-for-python

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44518 - Posted: 14 Sep 2016 | 6:51:24 UTC - in response to Message 44517.

Yup. That would prolly be it. :-)

jlhal
Send message
Joined: 1 Mar 10
Posts: 147
Credit: 1,077,535,540
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44520 - Posted: 14 Sep 2016 | 12:48:44 UTC - in response to Message 44517.

Maybe this could be of interest:
https://software.intel.com/en-us/blogs/2016/09/08/intel-distribution-for-python

Hope that Anaconda GPUGRID distrib is not only for INTEL !!...
____________
Lubuntu 16.04.1 LTS x64

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2671
Credit: 750,826,474
RAC: 478,673
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44527 - Posted: 14 Sep 2016 | 19:34:16 UTC

@Jihal: the last time they tried to port their code to OpenCL was probably 2-3 years ago. At that point the app was really slow. If I remember correctly the AMD GPUs were about a factor of 10 behind the "gaming equivalent" nVidias running CUDA code. Some libraries were clearly not optimized well. But worst was that they couldn't make it stable, i.e. the app would crash randomly. At that point they stopped these efforts, after having spent a few months on it I guess. Things will have improved by now, but I can understand if they're not keep on trying OpenCL again.

@caffeineyellow5: the general conception is that OpenCL is mostly inferior to CUDA. But this doens't mean it always has to be like that. Depending on how hard you try, you can have the best programming tools and screw up and write bad code. I'm not saying that's what DNETC is doing, but their code is rather simple (since their task is very easy and regular), so one could easily see one hand-optimization outperform the other one, despite the latter having far better librariers (which would matter in more complex problems).

And if you consider CPUs, you don't argue "an Intel Skylake is good for Fortan, but if you want to run C you should use an AMD" for good reason. The libraries and compilers have some influence, but it's mostly the code itself which governs which hardware it fits well. E.g. "branchy game-AI loves CPUs with low branch mispredit penalty and good branch prediction" or "in-memory big data analysis loves memory bandwidth".

@Topic: I suspect the Maxwell code would work very well for Pascal, considering the minor changes to the SM logic blocks in this generation. However, the project team currently seems to be rather limited by scientific manpower than computational power.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44592 - Posted: 28 Sep 2016 | 7:28:04 UTC
Last modified: 28 Sep 2016 | 7:28:33 UTC

So with the release of the CUDA 8 toolkit, we can assume support for the new cards is forthcoming very shortly?
https://developer.nvidia.com/cuda-toolkit

eXaPower
Send message
Joined: 25 Sep 13
Posts: 265
Credit: 1,043,017,367
RAC: 1,762,909
Level
Met
Scientific publications
watwatwatwatwatwat
Message 44599 - Posted: 29 Sep 2016 | 19:40:17 UTC - in response to Message 44592.

So with the release of the CUDA 8 toolkit, we can assume support for the new cards is forthcoming very shortly?
https://developer.nvidia.com/cuda-toolkit

http://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html

Deprecated Features:

•Fermi Architecture Support. Fermi architecture support is being deprecated in the CUDA 8.0 Toolkit, which will be the last toolkit release to support it. Future versions of the CUDA Toolkit will not support the architecture and are not guaranteed to work on that platform. Note that support for Fermi is being deprecated in the CUDA Toolkit but not in the driver. Applications compiled with CUDA 8.0 or older will continue to work on Fermi with newer NVIDIA drivers.

•Windows Server 2008 R2 Support. Support for Windows Server 2008 R2 is now deprecated and will be removed in a future version of the CUDA Toolkit.

•32-bit Linux CUDA Applications. CUDA Toolkit support for 32-bit Linux CUDA applications has been dropped. Existing 32-bit applications will continue to work with the 64-bit driver, but support is deprecated.

A note about 8.0.44 toolkit: driver 369.30 (non r370 branch) is included in package which most likely allow's for Windows XP support (368.81 for XP is a CUDA 8.0 driver) if the project chooses to develop with now publicly available 8.0 toolkit. Waiting until 8.5CUDA could cease XP support completely unless a workaround is available.

JoergF
Avatar
Send message
Joined: 20 Apr 15
Posts: 206
Credit: 296,317,561
RAC: 1,079,146
Level
Asn
Scientific publications
watwat
Message 44627 - Posted: 5 Oct 2016 | 9:50:50 UTC
Last modified: 5 Oct 2016 | 10:23:38 UTC

Seems that CUDA 8 has been released recently. Our Pascal GPUs are waiting eagerly to get off the starting blocks ;-)

jlhal
Send message
Joined: 1 Mar 10
Posts: 147
Credit: 1,077,535,540
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44628 - Posted: 5 Oct 2016 | 10:30:25 UTC - in response to Message 44627.

Seems that CUDA 8 has been released these days. Our Pascal GPUs are waiting eagerly to get off the starting blocks ;-)

The Linux X64 NVidia driver 370.28 is CUDA 8.0 capable and I'm using it on my GTX TITAN Black GPU.
Non Pascal GPUs could be able to crunch too with CUDA 8.0 driver ?

____________
Lubuntu 16.04.1 LTS x64

jlhal
Send message
Joined: 1 Mar 10
Posts: 147
Credit: 1,077,535,540
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44629 - Posted: 5 Oct 2016 | 11:26:33 UTC - in response to Message 44628.
Last modified: 5 Oct 2016 | 11:40:17 UTC

In addition yo my preceeding post about 370.28 driver :
For Ubuntu (debian) users add this PPA with this command line :

sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt-get update

Yoy may then use the Synaptic packets manager to install the latest available proprietary driver from NVidia .
You don't need any more a manual installation !

BOINC journal :
... CUDA: NVIDIA GPU 0: GeForce GTX TITAN Black (driver version 370.28, CUDA version 8.0, compute capability 3.5, 4096MB, 4010MB available, 6396 GFLOPS peak)
... OpenCL: NVIDIA GPU 0: GeForce GTX TITAN Black (driver version 370.28, device version OpenCL 1.2 CUDA, 6074MB, 4010MB available, 6396 GFLOPS peak)

Hope this helps...
____________
Lubuntu 16.04.1 LTS x64

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44630 - Posted: 5 Oct 2016 | 12:27:11 UTC - in response to Message 44627.

Seems that CUDA 8 has been released recently. Our Pascal GPUs are waiting eagerly to get off the starting blocks ;-)

According to this post, we're waiting for the release of CUDA8.5, so we still have to be patient.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44631 - Posted: 5 Oct 2016 | 12:30:15 UTC - in response to Message 44628.

Non Pascal GPUs could be able to crunch too with CUDA 8.0 driver ?

They are working with CUDA8.0 driver, however on Windows systems it seems to be slower than the latest CUDA7.5 driver (359.06).

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44633 - Posted: 6 Oct 2016 | 5:30:30 UTC - in response to Message 44031.

hi Crunchers

Just to let you know that there'll be a new app coming to support the ascal GPUs just as soon as Nvidia make a public release of CUDA 8.

Matt

Still waiting on the release of CUDA 8.5


So we need clarification then. Which were we waiting on?
____________
1 Corinthians 9:16 "For though I preach the gospel, I have nothing to glory of: for necessity is laid upon me; yea, woe is unto me, if I preach not the gospel!"
Ephesians 6:18-20, please ;-)
http://tbc-pa.org

JoergF
Avatar
Send message
Joined: 20 Apr 15
Posts: 206
Credit: 296,317,561
RAC: 1,079,146
Level
Asn
Scientific publications
watwat
Message 44636 - Posted: 7 Oct 2016 | 15:32:45 UTC

Cripes! My new 1070 and 1080 are bored beyond belief. So I have to redeploy them at Folding@Home for a while. Too bad that I pensioned my 980 off recently :-(
____________
I would love to see HCF1 protein folding and interaction simulations to help my little boy... someday.

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2671
Credit: 750,826,474
RAC: 478,673
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44644 - Posted: 8 Oct 2016 | 16:03:18 UTC
Last modified: 8 Oct 2016 | 16:03:37 UTC

They definitely said CUDA 8.5 - it's right in this thread. My GTX 1060 is off to Folding@Home, too, since POEM quit a few days ago.

MrS
____________
Scanning for our furry friends since Jan 2002

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 588
Credit: 2,037,147,925
RAC: 1,495,102
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44646 - Posted: 8 Oct 2016 | 16:31:54 UTC - in response to Message 44644.

They definitely said CUDA 8.5 - it's right in this thread. My GTX 1060 is off to Folding@Home, too, since POEM quit a few days ago.

MrS


Makes you wonder as to the value of some projects when they QUIT! Doesn't it?

10esseeTony
Send message
Joined: 4 Feb 15
Posts: 4
Credit: 576,470,875
RAC: 59,018
Level
Lys
Scientific publications
wat
Message 44647 - Posted: 8 Oct 2016 | 16:49:58 UTC - in response to Message 44646.
Last modified: 8 Oct 2016 | 16:53:06 UTC

.....since POEM quit a few days ago....


Makes you wonder as to the value of some projects when they QUIT! Doesn't it?


Ah, but in reality you should be quite curious as to WHY they quit. Pretty exciting, although sad to lose a favorite project without much notice.

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 588
Credit: 2,037,147,925
RAC: 1,495,102
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44648 - Posted: 8 Oct 2016 | 17:07:40 UTC - in response to Message 44647.

.....since POEM quit a few days ago....


Makes you wonder as to the value of some projects when they QUIT! Doesn't it?


Ah, but in reality you should be quite curious as to WHY they quit. Pretty exciting, although sad to lose a favorite project without much notice.


No, I don't because as they say on their site they now have enough computational resources to do what the need without out you. So thank you for eight years and P***s O**

Or to translate:

We now have no need for "something for nothing"

If GPUGrid goes the same way I'm out for good after 16 years+

BOINC has ultimately proven good only for unfunded projects to exist to some extent until proven to be worthy of funding so BOINC will go on to be a massive waste of our resources in order to preserve theirs.

I am still around for GPUGrid so hope to be proved wrong!

10esseeTony
Send message
Joined: 4 Feb 15
Posts: 4
Credit: 576,470,875
RAC: 59,018
Level
Lys
Scientific publications
wat
Message 44649 - Posted: 8 Oct 2016 | 17:21:33 UTC - in response to Message 44648.

Makes you wonder as to the value of some projects when they QUIT! Doesn't it?


....BOINC has ultimately proven good only for unfunded projects to exist to some extent until proven to be worthy of funding....



So, POEM was of questionable value or not? LOL. Sorry to be a thorn in your side, well, not really, I like poking people a bit. :D

I do fully understand your frustration, though. We donate freely, in hopes of real breakthroughs, and longer/better lives for all mankind, and then those real breakthroughs seem to get patented and cost $100 to $50,000 per dose. Ain't right.

Jim1348
Send message
Joined: 28 Jul 12
Posts: 460
Credit: 1,130,761,180
RAC: 18,722
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwat
Message 44650 - Posted: 8 Oct 2016 | 17:36:18 UTC - in response to Message 44648.

BOINC has ultimately proven good only for unfunded projects to exist to some extent until proven to be worthy of funding so BOINC will go on to be a massive waste of our resources in order to preserve theirs.

They have funding for researchers (student or otherwise) and servers. If they had funding for big computers, they wouldn't need BOINC, or us.

But most of them will publish papers and never get funded beyond that. Whether that is worthwhile depends on the project. I consider some of them a waste of electricity. Others are studies that are very basic, long before the commercial stage, but could be very important eventually.

jlhal
Send message
Joined: 1 Mar 10
Posts: 147
Credit: 1,077,535,540
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44651 - Posted: 8 Oct 2016 | 17:50:35 UTC - in response to Message 44648.

Hi all !
This interesting discussion about BOINC/projects future would surely better fit in its own thread ;-)
We are talking here about CUDA 8 and (soon ? ) upcoming apps
Regards
____________
Lubuntu 16.04.1 LTS x64

Profile skgiven
Volunteer moderator
Project tester
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,834,518,624
RAC: 292,156
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44652 - Posted: 8 Oct 2016 | 18:05:33 UTC - in response to Message 44649.
Last modified: 8 Oct 2016 | 18:11:42 UTC

[OT] When science is performed by corporations the objective is profit, not cure.
Boinc is open source. POEM's results will be available to the public and they are making the GPU code available to the scientific community. If they have enough results to analyse and publish their work then their objective is complete & the project was successful.
Most scientists are capable of performing experiments, analysing the results and publishing their finding, but are not in a position to make cures from their work. They do what they can. We do what we can...[/OT]

Alas, until the CUDA 8.5 Dev Kit is released to the public there can be no movement on using Pascal's here. We are in the worst case scenario I envisaged prior to the release of mainstream Pascals. The ball's in NV's park and they are playing tippy-tappy (possession) football.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

kain
Send message
Joined: 3 Sep 14
Posts: 110
Credit: 132,491,966
RAC: 18,921
Level
Cys
Scientific publications
watwatwatwat
Message 44653 - Posted: 8 Oct 2016 | 18:34:29 UTC - in response to Message 44648.

Sorry for OT but I really dont understand some aspects...


No, I don't because as they say on their site they now have enough computational resources to do what the need without out you. So thank you for eight years and P***s O**


Yes, and the problem is? They have money now, they will publish their work done with BOINC, now you can support scientists who does not have enough funding. Isn't that a sincere and good approach?!


BOINC has ultimately proven good only for unfunded projects to exist to some extent until proven to be worthy of funding so BOINC will go on to be a massive waste of our resources in order to preserve theirs.


I'm so confused... For some projects it is true, and again - what is your problem with that? That is a proof that work done with BOINC was interesting, useful and worth of support. Support of not only a bunch of nerds like you and me, with computers who usually don't even know exactly what they are actually crunching, but also support of people and institutions with (really) big money and influence. Thats great, not bad.


I am still around for GPUGrid so hope to be proved wrong!


If GPUGRID would for some reason get IDK, NASA support with enough power to thank us I will be very happy. Of course, it's sad to "lose" a project but dont get irrational.

Profile skgiven
Volunteer moderator
Project tester
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,834,518,624
RAC: 292,156
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44654 - Posted: 8 Oct 2016 | 18:50:58 UTC - in response to Message 44653.

The POEM forum is open if you want to continue the discussion there.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2671
Credit: 750,826,474
RAC: 478,673
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44731 - Posted: 16 Oct 2016 | 20:11:23 UTC

To other Admins & Mods: I unlocked the thread because I see no reason to lock it. If this is wrong from your point of view, feel free to lock it again but please provide some message here.

MrS
____________
Scanning for our furry friends since Jan 2002

Profile [CSF] Georg Stoifl
Avatar
Send message
Joined: 15 Nov 12
Posts: 10
Credit: 461,401,419
RAC: 442,307
Level
Gln
Scientific publications
watwatwatwatwatwatwat
Message 44748 - Posted: 17 Oct 2016 | 12:04:20 UTC

Hey,

Any news or timeline for Pascal Nvidia cards support?

Cheers,
Georg

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 788
Credit: 1,422,060,845
RAC: 1,410,932
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44749 - Posted: 17 Oct 2016 | 12:57:32 UTC

I noticed today that I've had two separate replication _3 long tasks issued.

e6s17_e5s7p0f556-GERARD_CXCL12CHALCLD_mol39_1-0-1-RND3782
e26s8_e16s16p0f492-GERARD_CXCL12CHALCYDIM_chalcone266_1-0-1-RND4729

Five of the six previous attempts failed because they were sent to incompatible Pascal-class GPUs. But the system couldn't do anything about that without sending out all the files, waiting for the host to attempt processing them, and receiving an error message back. That's a heck of a waste of time and resources, for both the project and the volunteer.

These cards have been on public sale for a while now, and are getting quite common. If the software really hasn't caught up with the hardware yet, perhaps you could consider excluding Pascal GPUs from work allocation for the time being? The plan_class_spec.xml file on the server allows for

Fields for NVIDIA GPU apps

<max_nvidia_compcap>MMmm</max_nvidia_compcap>
maximum compute capability

which would do the trick.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44753 - Posted: 17 Oct 2016 | 18:15:57 UTC - in response to Message 44749.

... perhaps you could consider excluding Pascal GPUs from work allocation for the time being?

+1000

DigitalParasite
Send message
Joined: 20 Jul 16
Posts: 3
Credit: 158,913,025
RAC: 348,284
Level
Ile
Scientific publications
wat
Message 44778 - Posted: 19 Oct 2016 | 10:13:58 UTC

I am getting some work units but they error out immediately:

<core_client_version>7.6.22</core_client_version>
<![CDATA[
<message>
(unknown error) - exit code -59 (0xffffffc5)
</message>
<stderr_txt>
# GPU [GeForce GTX 1070] Platform [Windows] Rev [3212] VERSION [65]
# SWAN Device 0 :
# Name : GeForce GTX 1070
# ECC : Disabled
# Global mem : 4095MB
# Capability : 6.1
# PCI ID : 0000:01:00.0
# Device clock : 1683MHz
# Memory clock : 4004MHz
# Memory width : 256bit
# Driver version : r370_00 : 37306
#SWAN: FATAL: cannot find image for module [.nonbonded.cu.] for device version 610

</stderr_txt>
]]>

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44784 - Posted: 19 Oct 2016 | 17:42:58 UTC - in response to Message 44778.

I am getting some work units but they error out immediately:

It's because the GPUGrid app does not support Pascal GPUs yet.

mixolyd
Send message
Joined: 11 Oct 16
Posts: 3
Credit: 9,154,850
RAC: 0
Level
Ser
Scientific publications
wat
Message 44785 - Posted: 19 Oct 2016 | 17:44:47 UTC

I'm curious why F@H supports Pascal GPUs just fine but GPUGRID doesn't?

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2671
Credit: 750,826,474
RAC: 478,673
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44786 - Posted: 19 Oct 2016 | 20:32:49 UTC - in response to Message 44785.

Pascal also works just fine at every other DC project that I know, with CUDA 8.0 driver. However, GPU-Grid targets CUDA 8.5 for the next app and is waiting for its release. We don't know what CUDA 8.5 offers them beyond what CUDA 8.0 already provides.

MrS
____________
Scanning for our furry friends since Jan 2002

mmonnin
Send message
Joined: 2 Jul 16
Posts: 38
Credit: 48,937,070
RAC: 1,364,637
Level
Val
Scientific publications
wat
Message 44791 - Posted: 20 Oct 2016 | 14:31:23 UTC

FAH uses OpenCL, not CUDA, so the same WUs can run on any card.

But there's no answer here why 8.5 is needed vs another version from what I've seen.

Jim1348
Send message
Joined: 28 Jul 12
Posts: 460
Credit: 1,130,761,180
RAC: 18,722
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwat
Message 44794 - Posted: 20 Oct 2016 | 16:02:22 UTC - in response to Message 44792.

But there's no answer here why 8.5 is needed vs another version from what I've seen.

I think MJH said (or implied) earlier that the bug they needed fixed was not fixed in CUDA 8.0, as they had originally hoped, but won't be implemented until CUDA 8.5. There seems to be no speculation on when that will be released, so Nvidia must keep a tight leash on their developers.

xixou
Send message
Joined: 8 Jun 14
Posts: 18
Credit: 18,757,357
RAC: 0
Level
Pro
Scientific publications
wat
Message 44807 - Posted: 21 Oct 2016 | 21:10:38 UTC
Last modified: 21 Oct 2016 | 21:12:13 UTC

Poem@home being stopped, I think GPUGRID will get much more GPU power if available for Pascal.


http://boincstats.com/en/stats/-5/user/detail/839477/projectList

Profile [CSF] Georg Stoifl
Avatar
Send message
Joined: 15 Nov 12
Posts: 10
Credit: 461,401,419
RAC: 442,307
Level
Gln
Scientific publications
watwatwatwatwatwatwat
Message 44815 - Posted: 22 Oct 2016 | 18:50:25 UTC - in response to Message 44807.

I've recognized too, they stopped some weeks ago. I had been running Poem on my 1070 for the last month, waiting for GPU grid to support Pascal. Now, it's doing Seti.
Anyone can give an outlook for Pascal support at GPU grid ?

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44820 - Posted: 23 Oct 2016 | 9:54:04 UTC
Last modified: 23 Oct 2016 | 9:58:20 UTC

There are speculations (on other projects [not BOINC, but development on projects like Debian/Ubuntu releases and such]) that the new branch of GCC was supposed to be supported in CUDA 8 and then it was put on speculation as Q1 2017. Could that be the answer we are looking for and can't get? Are we waiting on the GCC6 support with CUDA 8+? If so, the Q1 2017 speculation and a release of CUDA is speculated to follow that in short order just like the original GCC6 had the CUDA 8 release within about 20 days after realizing it was not going to have that support. Does BOINC, or some other component of what we are doing, like the MD/ACEMD rely on GCC support and to upgrade to the new architecture of NVIDIA GPUs in a "perfect world moving forward release" capacity, we need a CUDA 8 (or better) with GCC6 (or better) support?

I only thought of this after re-reading the threads here on this and came across this one https://www.gpugrid.net/forum_thread.php?id=4378&nowrap=true#44483 that I read, but missed the one part I just caught about GCC 5.4 breaking CUDA 8 RC. I think I missed it because I didn't understand it. I still can't say I do, but after searching around, I think I do a little.

There are 2 aspects of this questioning.
1) It is speculation on timing of CUDA 8.5 and GCC6 being in the first quarter of 2017. This may give us a time to expect this since there seems to be a blackout of information from NVDIA and those inside the CUDA toolkit developers teams on projects such as ours on when that release will come, so we must rely on the speculation of some of those developers.
2) It is a question, not a speculation on my part. Meaning that I am questioning the need of GCC6 compiling to be some part of what we need in a CUDA 8+ release. And if the answer to all of this that I wrote is simply, "No, that is not a factor", then the whole thing I just wrote is useless to read in conjunction with the rest of this thread. But if the answer is (or could be), "Yes. We do/might need that support in a CUDA release", then the speculation of the timeline of Q1 2017 for what we need is more clear and we can at least say that it is speculated to be around that time when people continue to come asking that "when" and "why" questions. That would also mean that the onus is not on GPUGRID (which we already kenw), but it is also not on NVIDIA (which we have been telling people), but on GNU to release the GCC6 version in order for NVIDIA to release a CUDA library with that GCC compiler compatibility to be compatible with the new applications compiled for the new hardware architecture.

P.S. I hope all this was clear to the reader as it was tough on me as the writer to keep/make it clear. I am on pain meds for kidney stones right now and keeping/making clear thoughts is difficult.) If it was not clear, please ask for clarification and I can re-read what I wrote and try to make it more clear to what I mean. Thanks. - Mike

John C MacAlister
Send message
Joined: 17 Feb 13
Posts: 178
Credit: 132,357,411
RAC: 2,487
Level
Cys
Scientific publications
watwatwatwatwatwatwatwat
Message 44821 - Posted: 23 Oct 2016 | 10:47:21 UTC - in response to Message 44820.

Hello, Mike:

First, I hope you soon recover from your painful affliction.

I do not pretend to understand much of the material in this thread and have this question:

Will I be able to continue contributing to GPUGrid research with these two rigs:

1. AMD FX-8350 with two GTX 660 Ti cards,

2. AMD 1090T with two GTX-650 Ti cards

Thanks,

John

There are speculations (on other projects [not BOINC, but development on projects like Debian/Ubuntu releases and such]) that the new branch of GCC was supposed to be supported in CUDA 8 and then it was put on speculation as Q1 2017. Could that be the answer we are looking for and can't get? Are we waiting on the GCC6 support with CUDA 8+? If so, the Q1 2017 speculation and a release of CUDA is speculated to follow that in short order just like the original GCC6 had the CUDA 8 release within about 20 days after realizing it was not going to have that support. Does BOINC, or some other component of what we are doing, like the MD/ACEMD rely on GCC support and to upgrade to the new architecture of NVIDIA GPUs in a "perfect world moving forward release" capacity, we need a CUDA 8 (or better) with GCC6 (or better) support?

I only thought of this after re-reading the threads here on this and came across this one https://www.gpugrid.net/forum_thread.php?id=4378&nowrap=true#44483 that I read, but missed the one part I just caught about GCC 5.4 breaking CUDA 8 RC. I think I missed it because I didn't understand it. I still can't say I do, but after searching around, I think I do a little.

There are 2 aspects of this questioning.
1) It is speculation on timing of CUDA 8.5 and GCC6 being in the first quarter of 2017. This may give us a time to expect this since there seems to be a blackout of information from NVDIA and those inside the CUDA toolkit developers teams on projects such as ours on when that release will come, so we must rely on the speculation of some of those developers.
2) It is a question, not a speculation on my part. Meaning that I am questioning the need of GCC6 compiling to be some part of what we need in a CUDA 8+ release. And if the answer to all of this that I wrote is simply, "No, that is not a factor", then the whole thing I just wrote is useless to read in conjunction with the rest of this thread. But if the answer is (or could be), "Yes. We do/might need that support in a CUDA release", then the speculation of the timeline of Q1 2017 for what we need is more clear and we can at least say that it is speculated to be around that time when people continue to come asking that "when" and "why" questions. That would also mean that the onus is not on GPUGRID (which we already kenw), but it is also not on NVIDIA (which we have been telling people), but on GNU to release the GCC6 version in order for NVIDIA to release a CUDA library with that GCC compiler compatibility to be compatible with the new applications compiled for the new hardware architecture.

P.S. I hope all this was clear to the reader as it was tough on me as the writer to keep/make it clear. I am on pain meds for kidney stones right now and keeping/making clear thoughts is difficult.) If it was not clear, please ask for clarification and I can re-read what I wrote and try to make it more clear to what I mean. Thanks. - Mike

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44827 - Posted: 24 Oct 2016 | 7:52:57 UTC - in response to Message 44821.

John, this is surely not the place for the question, but since you asked, I'll answer.

If you look on this page https://www.gpugrid.net/performance.php at the bottom, you will see both of your cards listed in the currently "top ranked" cards while even some of mine doing the work quite well are not. The one has a listed 'longest run' of about 65 hours (about 2.75 days) and the timeout for work units is 5 days (120 hours). Short runs take much less time and are currently available, though that has not been the usual over the past year. Availability may change and the length of time needed to complete a task might change, but I think currently and maybe for quite a while, those cards will do the tasks in more than enough time. The nuances will happen depending on your mother board, PCI Express slot speeds, CPU availability, OS, and other tweaks and things that can be done and are listed on other threads.

The new cards are offering more opportunity, not a replacement of older opportunity, although with time, some architectures do become "extinct" after many years. As you can see (not clearly, but its there) on this page https://www.gpugrid.net/graphs/ that your level of Kepler cards are still doing most of the work for GPUGRID currently and Fermi cards are falling off with their being "taken off the market" in new code and general usefulness to complete work on time.

Crunch away sir!

gianni
Send message
Joined: 11 Jul 08
Posts: 16
Credit: 105,098
RAC: 0
Level

Scientific publications
watwatwat
Message 44851 - Posted: 26 Oct 2016 | 7:56:08 UTC - in response to Message 44827.

we will have support for Pascal very soon, in a matter of days.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44853 - Posted: 26 Oct 2016 | 9:55:03 UTC - in response to Message 44851.

we will have support for Pascal very soon, in a matter of days.

This is good news.
Could you share more details about the new app?
What CUDA / driver version will be needed for it?
The old apps (CUDA60 & CUDA65) will still be available after the new app is released?
Will it work on older cards?
Will it work on older OS? (especially Windows XP - but I guess it depends on the driver)
For what OSes will it be available? (I guess it will come out for Linux first)

KSUMatt
Avatar
Send message
Joined: 11 Jan 13
Posts: 214
Credit: 831,004,493
RAC: 0
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwat
Message 44854 - Posted: 26 Oct 2016 | 12:14:14 UTC - in response to Message 44851.

we will have support for Pascal very soon, in a matter of days.


Very good news! Time to start reducing my cache of Einstein tasks.

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwat
Message 44858 - Posted: 26 Oct 2016 | 14:14:13 UTC - in response to Message 44851.

we will have support for Pascal very soon, in a matter of days.

Sweet! Now is finally the time to start looking into the cards available to upgrade.... but need to get my pcs stable!!! (my seeming constant struggle I keep losing)

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 44859 - Posted: 26 Oct 2016 | 16:45:20 UTC - in response to Message 44858.

we will have support for Pascal very soon, in a matter of days.

Sweet! Now is finally the time to start looking into the cards available to upgrade.... but need to get my pcs stable!!! (my seeming constant struggle I keep losing)



What's wrong with the PC?
____________
Cruncher/Learner in progress.

eXaPower
Send message
Joined: 25 Sep 13
Posts: 265
Credit: 1,043,017,367
RAC: 1,762,909
Level
Met
Scientific publications
watwatwatwatwatwat
Message 44860 - Posted: 26 Oct 2016 | 17:11:30 UTC - in response to Message 44851.

we will have support for Pascal very soon, in a matter of days.

Any recent change in the server scheduling for Pascal (also with Maxwell) blacklisting?
Unable to download tasks during last couple days on my (2) GTX 1070 / (1) GTX 1060 / (2) GTX 970 / (1) 384core GT 630 system.
Windows 8.1 refuses to see more than 4 GPU's since r370 branch (including r375). As of late a couple of 7/8 GPUs Window 10 (enterprise) were on the performance page.

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 271
Credit: 1,305,108,981
RAC: 5,303,877
Level
Met
Scientific publications
watwat
Message 44862 - Posted: 27 Oct 2016 | 9:39:16 UTC

What are we waiting for at this point? The developer of your software to update to the new Cuda?

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 271
Credit: 1,305,108,981
RAC: 5,303,877
Level
Met
Scientific publications
watwat
Message 44863 - Posted: 27 Oct 2016 | 9:39:23 UTC

What are we waiting for at this point? The developer of your software to update to the new Cuda?

Profile MJH
Project administrator
Project developer
Project scientist
Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 44865 - Posted: 27 Oct 2016 | 9:57:35 UTC - in response to Message 44863.

What are we waiting for at this point? The developer of your software to update to the new Cuda?


Our Windows app has until now been built as a 32bit Windows XP application. CUDA 8 has dropped support for this platform, so we are having to move to a 64bit Windows 7 development platform. This is causing some problems, but it should be done in a while.

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 44867 - Posted: 27 Oct 2016 | 12:15:02 UTC - in response to Message 44853.
Last modified: 27 Oct 2016 | 12:20:22 UTC

Hi Matt,

May you please answer Zoltan's question a few posts back? I am curious as well as to what the details of the new app will be.

Most importantly to me and maybe some other's that Zoltan asked:

Will we have to upgrade from windows XP, or can we stick with it and keep NVidia driver 359.09?

Thanks

I'm asking about windows xp because I'm wondering if we can still use the older app.

Logan

Profile MJH
Project administrator
Project developer
Project scientist
Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 44868 - Posted: 27 Oct 2016 | 12:53:18 UTC - in response to Message 44867.


Hi Matt,

May you please answer Zoltan's question a few posts back? I am curious as well as to what the details of the new app will be.

Most importantly to me and maybe some other's that Zoltan asked:

Will we have to upgrade from windows XP, or can we stick with it and keep NVidia driver 359.09?

Thanks

I'm asking about windows xp because I'm wondering if we can still use the older app.

Logan


To spoil my next post -- If you have a Pascal GPU you will need driver 360+, Windows 7 and 64bit.

Older GPUs will keep the current 65 app.

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 44917 - Posted: 27 Oct 2016 | 20:32:10 UTC - in response to Message 44868.

Okay thank you

I Appreciate it.
____________
Cruncher/Learner in progress.

Bedrich Hajek
Send message
Joined: 28 Mar 09
Posts: 340
Credit: 3,819,818,009
RAC: 929,462
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44942 - Posted: 27 Oct 2016 | 22:03:29 UTC - in response to Message 44791.

FAH uses OpenCL, not CUDA, so the same WUs can run on any card.

But there's no answer here why 8.5 is needed vs another version from what I've seen.




What are the pluses and minuses in using OpenCL over CUDA?

Or was that already posted somewhere in this form?


Michal Kinďura
Send message
Joined: 22 Jun 11
Posts: 2
Credit: 14,998,367
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwat
Message 44989 - Posted: 28 Oct 2016 | 20:31:59 UTC

When you will support Pascal GPU on Linux hosts?

Profile skgiven
Volunteer moderator
Project tester
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,834,518,624
RAC: 292,156
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 44992 - Posted: 28 Oct 2016 | 21:10:28 UTC - in response to Message 44989.

Testing should start next week, or the following week (depending on Matt's availability and results testing on Windows).

https://www.gpugrid.net/forum_thread.php?id=4405&nowrap=true#44930
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

[AF>P4G] anthony
Send message
Joined: 14 Mar 10
Posts: 13
Credit: 275,902,573
RAC: 370,161
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 45018 - Posted: 29 Oct 2016 | 19:29:16 UTC

Hello,

Good news, I finaly change my laptop with a 850M for a desktop (HP Omen 870) with a 1080 founders edition. Yesterday, I have tested GPUGRID but with short I had errors after 2 seconds of computation.

I have intalled CUDA 8 avec 375.63 drivers.

Until the app arrive, I chrunch for PimeGrid.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45030 - Posted: 29 Oct 2016 | 23:31:27 UTC - in response to Message 45018.
Last modified: 29 Oct 2016 | 23:49:08 UTC

Until the app arrive, I chrunch for PimeGrid.

You can have another try now, as the version 9.14 app is working fine.
Your host was running the previous versions, which weren't working.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45031 - Posted: 29 Oct 2016 | 23:37:02 UTC - in response to Message 44342.
Last modified: 29 Oct 2016 | 23:49:53 UTC

Just checking in. I have a 1080 and 1070 waiting for a Windows app. They've been busy on Einstein in the meantime.

Still waiting on the release of CUDA 8.5

Well, in the light of yesterday's events, could you please explain this post?

[AF>P4G] anthony
Send message
Joined: 14 Mar 10
Posts: 13
Credit: 275,902,573
RAC: 370,161
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 45034 - Posted: 30 Oct 2016 | 6:58:54 UTC
Last modified: 30 Oct 2016 | 7:03:56 UTC

Good morning,

Thanks Retvari Zoltan, the 9.14 short runs APP works fine, 73 % of charge on my 1080 at 1809 Mhz.

I allow test applications and short runs. Usually, I prefer long runs, these apps give more credit.

Another question, I understand that Fermi GPU are out of support with GPUGRID, is it true ?

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45047 - Posted: 30 Oct 2016 | 17:40:40 UTC - in response to Message 45034.
Last modified: 30 Oct 2016 | 18:35:43 UTC

I understand that Fermi GPU are out of support with GPUGRID, is it true ?

Basically yes.
There are some batches which will work on Fermi cards, but question which batches are these can be answered only by experience (=there will be a lot of failed workunits on Fermi GPUs). Fermi cards have so poor energy efficiency compared to recent generations that they are rather not recommended.

EDIT: I've found a host with a working GTX 480, here's its tasklist
EDIT2: Host 176489 with a working GTX 480, its tasklist
EDIT3: Host 205349 with a working GTX 480, its tasklist

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 1844
Credit: 10,627,187,944
RAC: 9,830,358
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45048 - Posted: 30 Oct 2016 | 18:52:29 UTC - in response to Message 45047.
Last modified: 30 Oct 2016 | 19:36:26 UTC

Host 158961 with a working GTX 580, its tasklist
Host 160688 with a working GTX 580, its tasklist
Host 163864 with a working GTX 580, its tasklist
Host 174593 with a working GTX 580, its tasklist
Host 185272 with a working GTX 580, its tasklist
Host 191599 with a working GTX 580, its tasklist
Host 194099 with a working GTX 580, its tasklist
Host 218071 with a working GTX 580, its tasklist
Host 236670 with a working GTX 580, its tasklist
Host 247225 with a working GTX 580, its tasklist
Host 271784 with a not so well working GTX 580, its tasklist
Host 272487 with a working GTX 580, its tasklist
Host 281538 with a working GTX 580, its tasklist
Host 286412 with a working GTX 580, its tasklist
Host 290202 with a working GTX 580, its tasklist
Host 302800 with a working GTX 580, its tasklist
Host 371474 with a working GTX 580, its tasklist

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 30,647,175
RAC: 32,853
Level
Val
Scientific publications
wat
Message 45049 - Posted: 30 Oct 2016 | 19:42:49 UTC - in response to Message 45048.

Just throwing it out there but you can look on my 2nd PC as well that has a GTX 460.

https://www.gpugrid.net/results.php?hostid=388795

Although it is far slower compared to my GTX 960, I've had success so far. I plan on this only being a temporary GPU, as I will upgrade within the next month or two.




____________
Cruncher/Learner in progress.

[AF>P4G] anthony
Send message
Joined: 14 Mar 10
Posts: 13
Credit: 275,902,573
RAC: 370,161
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 45058 - Posted: 31 Oct 2016 | 7:39:34 UTC

Hello,

Thanks for this precisions, I have a GTX 580 :
https://www.gpugrid.net/show_host_detail.php?hostid=102868
But the motherboard is out, and I put this GPU into a computer with Vista :
https://www.gpugrid.net/show_host_detail.php?hostid=358931

But the driver is not supported (I have also another old card into and the 322 driver is the latest driver can support the 2 cards) , and I think it is not necessary to intall linux to use this old card with gpugrid.

Maybe this winter if I need more heat system power.
Yes I heat my house with computers.......

pvh
Send message
Joined: 17 Mar 10
Posts: 23
Credit: 427,605,206
RAC: 764,810
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45844 - Posted: 22 Dec 2016 | 20:05:41 UTC

I have a GTX 1060 card and noticed that the server sends both cuda65 and cuda80 tasks to my machine. The cuda65 tasks obviously won't run due to the well known incompatibility issue. This is a waste of perfectly good WUs (which are hard to come by anyway). Is there something I can do to prevent getting cuda65 tasks?

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 271
Credit: 1,305,108,981
RAC: 5,303,877
Level
Met
Scientific publications
watwat
Message 45857 - Posted: 23 Dec 2016 | 11:52:46 UTC - in response to Message 45844.

I have a GTX 1060 card and noticed that the server sends both cuda65 and cuda80 tasks to my machine. The cuda65 tasks obviously won't run due to the well known incompatibility issue. This is a waste of perfectly good WUs (which are hard to come by anyway). Is there something I can do to prevent getting cuda65 tasks?


Hmm, I do not know why it would be sending you those but for now I would check the WUs and abort the cuda65 WU as this allows someone else with the appropriate hardware to grab it.

Profile skgiven
Volunteer moderator
Project tester
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,834,518,624
RAC: 292,156
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45878 - Posted: 23 Dec 2016 | 20:35:55 UTC - in response to Message 45844.

I have a GTX 1060 card and noticed that the server sends both cuda65 and cuda80 tasks to my machine. The cuda65 tasks obviously won't run due to the well known incompatibility issue. This is a waste of perfectly good WUs (which are hard to come by anyway). Is there something I can do to prevent getting cuda65 tasks?

What version of Linux and what driver are you using?
Have you upgraded your GPU or driver?
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

pvh
Send message
Joined: 17 Mar 10
Posts: 23
Credit: 427,605,206
RAC: 764,810
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45879 - Posted: 23 Dec 2016 | 22:43:25 UTC
Last modified: 23 Dec 2016 | 23:02:45 UTC

I am running openSUSE 42.1 with the 375.20 driver. The card is new.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 788
Credit: 1,422,060,845
RAC: 1,410,932
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45883 - Posted: 23 Dec 2016 | 23:45:21 UTC - in response to Message 45879.

I am running openSUSE 42.1 with the 375.20 driver. The card is new.

It's also worth noting that you're using BOINC v7.2.42, which is getting quite elderly (released ~ 28-Feb-2014). It's possible that version may not report all the card and driver information required for the server to make a proper allocation, especially with such a new card. But I haven't checked that yet.

pvh
Send message
Joined: 17 Mar 10
Posts: 23
Credit: 427,605,206
RAC: 764,810
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45887 - Posted: 24 Dec 2016 | 4:48:31 UTC

Boinc v7.2.42 is still the recommended version. It detects the card correctly, so there is no reason it should not be able to report this to the server:

15-Dec-2016 21:47:50 [---] CUDA: NVIDIA GPU 0: GeForce GTX 1060 3GB (driver version unknown, CUDA version 8.0, compute capability 6.1, 3036MB, 2931MB available, 6167 GFLOPS peak)
15-Dec-2016 21:47:50 [---] OpenCL: NVIDIA GPU 0: GeForce GTX 1060 3GB (driver version 375.20, device version OpenCL 1.2 CUDA, 3036MB, 2931MB available, 6167 GFLOPS peak)


The string "driver version unknown" is surprising, especially since it does detect it correctly on the openCL line, but the card version and compute capability information is there...

Profile skgiven
Volunteer moderator
Project tester
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,834,518,624
RAC: 292,156
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45888 - Posted: 24 Dec 2016 | 9:45:43 UTC - in response to Message 45887.
Last modified: 25 Dec 2016 | 14:59:46 UTC

7.6.31 is the recommended version for Linux, if that's the issue (the app reads the driver directly at GPUGrid). It's likely in the repository, if you installed Boinc that way.

Alternatively, there might be an issue with that 375 driver - I reverted from 375.26 to 370.28 after experiencing problems with it on Ubuntu. I got the impression 375.26 wasn't fully/correctly built.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

jlhal
Send message
Joined: 1 Mar 10
Posts: 147
Credit: 1,077,535,540
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45890 - Posted: 24 Dec 2016 | 11:06:35 UTC - in response to Message 45888.
Last modified: 24 Dec 2016 | 11:55:54 UTC

Merry Christmas everybody !

Do you know you may allways have the latest Nvidia drivers for Ubuntu by simply executing this command :

sudo add-apt-repository ppa:graphics-drivers/ppa

See this link for details :

http://www.omgubuntu.co.uk/2015/08/ubuntu-nvidia-graphics-drivers-ppa-is-ready-for-action

Currently running with 370.28

Hope this helps...
____________
Lubuntu 16.04.1 LTS x64

pvh
Send message
Joined: 17 Mar 10
Posts: 23
Credit: 427,605,206
RAC: 764,810
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45897 - Posted: 24 Dec 2016 | 17:32:10 UTC - in response to Message 45888.

7.6.33 is the recommended version for Linux, if that's the issue (the app reads the driver directly at GPUGrid).


I think you are confused with Windows there. The recommended version for Linux is 7.2.42. I am sure the app communicates directly with the driver, but by then it is too late. The server decides the app version, and the cuda65 app cannot handle my hardware.

Alternatively, there might be an issue with that 375 driver.


I don't see any issues with the driver so far. It seems to be fine. The server decides to send tasks for the wrong version of the app to me. That is where the problem is.

captainjack
Send message
Joined: 9 May 13
Posts: 112
Credit: 820,718,399
RAC: 1,200,373
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwat
Message 45906 - Posted: 25 Dec 2016 | 3:34:28 UTC

7.6.33 is the recommended version for Linux, if that's the issue (the app reads the driver directly at GPUGrid).



I think you are confused with Windows there. The recommended version for Linux is 7.2.42. I am sure the app communicates directly with the driver, but by then it is too late. The server decides the app version, and the cuda65 app cannot handle my hardware.


If you are looking at the Berkeley page (which doesn't get much attention these days), the recommended version is 7.2.42. Using the software installer in Ubuntu Linux, I get 7.6.33.

Profile Beyond
Avatar
Send message
Joined: 23 Nov 08
Posts: 1073
Credit: 4,496,456,504
RAC: 414,796
Level
Arg
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45908 - Posted: 25 Dec 2016 | 6:12:07 UTC - in response to Message 45890.

Merry Christmas everybody !

I'll 2nd that!

Profile skgiven
Volunteer moderator
Project tester
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,834,518,624
RAC: 292,156
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45912 - Posted: 25 Dec 2016 | 14:58:41 UTC - in response to Message 45908.
Last modified: 25 Dec 2016 | 15:27:40 UTC

Happy Christmas.

OpenSUSE 42.1 was released in Nov 2015. Any Boinc version placed into repositories around that time would have been adapted for 42.1 from the latest Berkely versions prior to Nov 2015. However, if someone upgraded to 42.1 then I guess they might be using an even older version.

Perhaps this will prove useful,

http://www.rechenaugust.de/boinc/

http://www.rechenaugust.de/boinc/boinc_7.6.31_x86_64-suse-linux-gnu.sh

The Boinc Manager version in the Ubuntu 16.04 x64 LTS repository is also 7.6.31. That's the recommended version for Ubuntu 16.04 x64 (and according to rechenaugust also works for 42.1). Other similar Linux builds will likely have similar Boinc versions available from repositories.

Generally, repository version available to Linux users depend on the flavour and version of Linux being used. If its a dated version of Linux (or not well supported) then there might not be a new/recent Boinc version in the repositories. If it's a newer/updated version of Linux then the Boinc repo version is likely to be recent.

PS. Yes I meant the repo 7.6.31 Linux version earlier (rather than the 7.6.33 Win version) and yes the Berkeley page isn't updated or as helpful as it might be for Linux users, but it does advise to install from a repository and when you visit from a Linux system and go to the download page (which is visibly different in appearance for Win users) it does again recommend to install a distribution-specific package instead of 7.2.42 (which is unfortunately still described as the recommended version?!?).

I'd updated from 367.57 to 370.28 and then to 375.26 (which didn't work for me), but I also swapped out a 970 and replaced with a 1060-3GB.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

ETQuestor
Send message
Joined: 11 Jul 09
Posts: 26
Credit: 777,353,686
RAC: 568,974
Level
Glu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45916 - Posted: 25 Dec 2016 | 18:03:35 UTC - in response to Message 45897.
Last modified: 25 Dec 2016 | 18:05:24 UTC

The issue with detecting NVIDIA/CUDA driver versions on linux was fixed in BOINC 7.3.16, circa late April 2014. You must run a newer version than that (and I don't believe this fix was ever checked into the 7.2.x branch).

The BOINC home page recommendation is for an older version (7.2.42 is circa Feb 2014).

What is recommended for use here is something like 7.6.31 or 7.6.33.

Link to commit that fixed this:

https://github.com/BOINC/boinc/commit/450f6808800bf6b3f1c0b5d8cfcbf84b0650987a

pvh
Send message
Joined: 17 Mar 10
Posts: 23
Credit: 427,605,206
RAC: 764,810
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45943 - Posted: 28 Dec 2016 | 22:22:17 UTC - in response to Message 45912.

OpenSUSE 42.1 was released in Nov 2015. Any Boinc version placed into repositories around that time would have been adapted for 42.1 from the latest Berkely versions prior to Nov 2015. However, if someone upgraded to 42.1 then I guess they might be using an even older version.


openSUSE Leap 42.1 and 42.2 both ship v7.2.42, while openSUSE Tumbleweed ships v7.6.32 (but Tumbleweed is useless for running GPGPU work as the proprietary drivers would be forever breaking due to the very aggressive kernel updates). So openSUSE Leap seems to be following the Boinc recommendation.

PS - it makes no difference if you installed openSUSE from scratch or upgraded from an older version, you will always get the latest version of a package in the repository (it would be too difficult to manage the security fixes otherwise). But if you installed boinc from a third-party site, then the openSUSE installer will leave that alone of course.

http://www.rechenaugust.de/boinc/boinc_7.6.31_x86_64-suse-linux-gnu.sh


I installed this, and so far it seems to do the trick and solve my problem. It is worrying though that I have to rely on a third-party site to get a working version of Boinc. This is not exactly what I would think linux support should look like...

Thanks for the link!

Profile MJH
Project administrator
Project developer
Project scientist
Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 45968 - Posted: 29 Dec 2016 | 17:23:52 UTC - in response to Message 45943.


openSUSE Leap 42.1 and 42.2 both ship v7.2.42, while openSUSE Tumbleweed ships v7.6.32 (but Tumbleweed is useless for running GPGPU work as the proprietary drivers would be forever breaking due to the very aggressive kernel updates). So openSUSE Leap seems to be following the Boinc recommendation.


The Nvidia driver will register itself with DKMS (if that's installed) so that it gets rebuilt every time the machine is rebootedf with a new kernel installed.

Matt

pvh
Send message
Joined: 17 Mar 10
Posts: 23
Credit: 427,605,206
RAC: 764,810
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 45974 - Posted: 29 Dec 2016 | 19:46:36 UTC - in response to Message 45968.

The Nvidia driver will register itself with DKMS (if that's installed) so that it gets rebuilt every time the machine is rebooted with a new kernel installed.


Indeed, but that will only work if the kernel is supported. If the kernel is too new, it can break the build. I fear that will happen quite a lot with Tumbleweed due to its very aggressive bleeding edge kernel support (haven't tried this, but I am pretty sure that will happen). The people at Suse themselves warn that they make no effort to keep the proprietary drivers alive in Tumbleweed.

But the automatic rebuilds work quite well in openSUSE Leap.

Post to thread

Message boards : News : Geforce 10 / Pascal app coming soon