1) Message boards : News : New beta application (Message 44349)
Posted 1 day ago by Profile skgiven
Both Betas went to the same user. That's quite a finite testing methodology. Might take a while to get round everyone though...
2) Message boards : GPUGRID CAFE : What does loss of net neutrality mean for volunteer computing? (Message 44311)
Posted 4 days ago by Profile skgiven
'Net Neutrality' is about sending all types of data at the same rates; without throttling, it's not about cyber crime, hacking, spying...

Obviously we don't want our uploads and downloads throttled in the same way online gamers don't want their connections to be glitchy or intermittent and viewers of streaming media want good resolution, which requires good transfer rates. However, we have the capability to adjust our Boinc Managers to whatever the situation is, so unless we are struggling to meet a 24h or 48h deadline for bonuses we're reasonably ok - we're less vulnerable to throttling/can do something about it, so it won't impact on us as much. Contrast watching an Amazon video and halfway through it cuts out.

PS. That Norse attack map shows that the biggest target is the US but also that the biggest attacker is the US. Next up as an attacker is China, followed by the Netherlands. The second biggest target is the UAE. Russia isn't in the top 10 as an attacker. Not sure how many of these are valid attacks though. Could be 'suspicious activities' rather than 'attacks'.
3) Message boards : Number crunching : ADRIA_2OV5_CONF_CLOSED (Message 44310)
Posted 4 days ago by Profile skgiven
My guess is that the whole batch needs to be built properly for it to be worth much. If too many WU's are duds then I guess the run isn't much use. I would say, "at least you got 'credit' for your effort" but it's like painting a wall and then having to paint it a different colour, and you pay for the paint!
4) Message boards : Number crunching : 8 hours, 42 minutes elapsed, 3 hours CPU time (Message 44306)
Posted 5 days ago by Profile skgiven
Even if you could fully and accurately quantify the progress of a projects scientific research and your contribution towards it, comparing one project to another is qualitative and what the research might lead to in years to come is speculative.

On an i7 you would be freeing up one thread (not core). Doing so should improve GPU performance (reduce the times), but your times are quite good anyway. While threads are entities in themselves they don't entirely operate in isolation; shared cache, controller, queues/buses. The difference in overall CPU performance of a i7-4930K crunching 12 CPU tasks vs 11 CPU tasks is probably negligible compared to the GPU performance improvements that come from not having the CPU constantly over-saturated (in my qualified opinion).

For a GPUGrid task the 'Elapsed' time is basically the time it ran on the GPU.
The CPU time is the time it ran on the CPU (not the GPU).
'Elapsed' time actually includes a few seconds to load the task before it starts and a few seconds at the end (but doesn't include any time is sat in the work queue, was paused, the download, upload or report time periods).
5) Message boards : Number crunching : GTX650 hates long runs and some questions (Message 44305)
Posted 5 days ago by Profile skgiven
The GTX650 is half the 'speed' of a GTX 650 Ti which was an entry level card for here several years ago.
A GTX980Ti should be 8 or 9 times (8.67) faster than a GTX 650 at GPUGrid.

FAQ - Best configurations for GPUGRID
6) Message boards : Number crunching : ADRIA_2OV5_CONF_CLOSED (Message 44304)
Posted 5 days ago by Profile skgiven
Looks like the batch was cancelled early today:

Exit status 202 (0xca) EXIT_ABORTED_BY_PROJECT

https://www.gpugrid.net/workunit.php?wuid=11700325
Too many errors (may have bug) WU cancelled

https://www.gpugrid.net/workunit.php?wuid=11700315
WU cancelled

Some of these tasks actually completed, but many didn't run; they failed to load due to a file read error. Some are still in progress.
7) Message boards : Number crunching : GIANNI_A2 WUs (Message 44301)
Posted 5 days ago by Profile skgiven
There is a new task type from GIANNI (different from the very long tasks):

New - e4s30_e2s22p0f423-GIANNI_A2A-0-1-RND2995_1

Long - e1s80_s1-GIANNI_D3C36bCHL1-0-1-RND9194_3

These new tasks can use a lot of CPU & PCIE bandwidth, but they also utilize the GPU a lot and thus run hot.

Running one of these new task types on a Linux system with 2 of my 3 CPU cores free (an old AMD tri-core) I'm getting high-end throughput:

92 to 95% GPU Utilization (SWAN_SYNC not used)
36% PCIE bandwidth (PCIE2 X16)
988MB GPU RAM used
74C with GPU fan at 85% in an open case (ambient 23C)

While using 2 of the 3 CPU cores for CPU crunching elsewhere the GTX970 Utilization was ~85%, PCIE bandwidth was ~30% and temps were the same but with the GPU fan at 75%.

8% into the run after 79min suggests the task will take ~16h40min to complete - which is the right side of 24h.

They take ~9h or so on a well configured GTX 980 Ti for 345,000 Credits, so these tasks scale well on good systems.
8) Message boards : Graphics cards (GPUs) : New Gianni tasks take loooong time... a warning (8-12-16) (Message 44288)
Posted 7 days ago by Profile skgiven
It's ironic that running the longest tasks simultaneously would be the most beneficial in terms of throughput (for some).
9) Message boards : Graphics cards (GPUs) : New Gianni tasks take loooong time... a warning (8-12-16) (Message 44249)
Posted 10 days ago by Profile skgiven
So it looks like a GTX970 (at stock on a weak system) will take 31 to 32h to crunch these on Linux-x64. That suggests the WDDM overhead for these is at least 12.5% but probably closer to 16%.

A GTX980 is ~17% faster (stock) than a GTX970 so would still take over 24h to complete on Linux (over 26h). If it was overclocked by ~10% then it might be able to just about complete inside 24h if the system was tuned to do so (SWAN_SYNC used, high CPU clock and fast RAM...).

GTX 980 @ 1388MHz, GDDR5 @ 3505 MHz, i3-4160, WinXP, SWAN_SYNC on, no other tasks: 19h 24m 26s
It's almost (~8m) missed the 24h bonus, as it spent 5h 28m in the queue.


GIANNI_D3C36bCHL from Performance

1 Retvari Zoltan 15236101 14.49 NVIDIA GeForce GTX 980 Ti (4095MB) driver: 368.22

14h 30min isn't much over the app description: Long runs (8-12 hours on fastest card) and there is a good chance the GTX1080 (when the CUDA 8 dev kit goes on public release) will manage it within that 12h time frame (on Linux).
10) Message boards : Graphics cards (GPUs) : New Gianni tasks take loooong time... a warning (8-12-16) (Message 44235)
Posted 12 days ago by Profile skgiven
There have been several WUs by Gianni in the recent past. They are really huge and result in a nice credit, but no GPU below a 980Ti can crunch them within 24 hours and get the 20% extra credit.
One of my hosts is a GTX970 - which it took some 36 hours at ~1360MHz.
Hence, a change in this 24hrs rule would be desireable :-)


My Linux system is crunching a Gianni now and it's at 50% after 15h 45min.
My GPU clock (acording to NV X Server is 1278MHz) and I'm using the 361.42 NV driver. X Server also says I'm using 4% PCIE bandwidth on a PCIE2.0 x16 slot. GPU utilization is around 67% but varies from 62% to 70%. My CPU is an AMD A6-3500 APU (2.1/2.4GHz).

So it looks like a GTX970 (at stock on a weak system) will take 31 to 32h to crunch these on Linux-x64. That suggests the WDDM overhead for these is at least
12.5% but probably closer to 16%.

A GTX980 is ~17% faster (stock) than a GTX970 so would still take over 24h to complete on Linux (over 26h). If it was overclocked by ~10% then it might be able to just about complete inside 24h if the system was tuned to do so (SWAN_SYNC used, high CPU clock and fast RAM...).

Note that the bonus is +25% for finishing (and reporting) inside 48h or +50% for finishing and reporting inside 24h.


Next 10