21) Message boards : News : New CUDA4.2 applications are out for Kepler GPUs (Message 25879)
Posted 4345 days ago by Profile K1atOdessa
My two 570's (in separate machines) are factory OC'd to 797 Mhz and 780 Mhz. The 780 Mhz card finished a long WU about 30% faster with cuda4.2 than cuda3.1, and I didn't have any issues with errors. Also, given it is summer that card is running warmer than normal.

My 797 Mhz card is working on it's first long WU (did a short cuda4.2 WU yesterday w/ no issues), so we'll see how that goes. Maybe I am just lucky on the OC front.


I am liking the 90%+ utilization in Win7.
22) Message boards : News : New CUDA4.2 applications are out for Kepler GPUs (Message 25878)
Posted 4345 days ago by Profile K1atOdessa
I will let that run and see if I get any more cuda3.1 WU's.


Well, 1st long cuda4.2 finished (and quite faster than cuda3.1, as promised). It immediately downloaded another cuda4.2. All seems well, though 2 WU's hardly indicates a pattern. Will keep an eye on it.


On my 2nd rig, I did the same thing after the cuda3.1 WU finished. "Project reset" in BOINC, and got a cuda4.2. There may be something to that, or again maybe just 3 for 3 luck.

If anyone isn't getting any cuda4.2's (or a mix), could be worth a shot to try it and see.
23) Message boards : News : New CUDA4.2 applications are out for Kepler GPUs (Message 25868)
Posted 4345 days ago by Profile K1atOdessa
On my 2nd rig, I saw the current cuda3.1 WU getting close to finishing. After it completed and uploaded, I clicked the "reset project" button in BOINC. Now, it could be a complete coincidence, but I did get a WU using cuda4.2 afterward.

I will let that run and see if I get any more cuda3.1 WU's.
24) Message boards : News : New CUDA4.2 applications are out for Kepler GPUs (Message 25863)
Posted 4345 days ago by Profile K1atOdessa
Interesting... Only one of my two rigs has completed a cuda3.1 WU since this announcement and thus become eligible for the long cuda4.2 app. However, the newly downloaded WU is labeled as using cuda3.1 still.

For kicks yesterday I disabled long WU's and selected only short WU's to test the system with the cuda4.2 app. It completed one WU without issue. So, I re-enabled only Long WU's yesterday afternoon.

Any reason why it would use a cuda3.1 app when the cuda4.2 app is available, for a GTX570 running 301.42 drivers? Should I try a detach/reattach after this WU's ends?
25) Message boards : GPUGRID CAFE : The milestone thread (Message 25816)
Posted 4349 days ago by Profile K1atOdessa
100M!
26) Message boards : Number crunching : Replaced fans with blowers (Message 25535)
Posted 4364 days ago by Profile K1atOdessa
I gotta get more of these things!


Have a link to the product you'd like to share?
27) Message boards : Server and website : Multiple WU's sent to one host (Message 25411)
Posted 4370 days ago by Profile K1atOdessa
7.0.27 is a deprecated beta. 7.0.28 fixes a few things. Might be worth trying, and reporting.


Thanks. I will try that first...likely tomorrow.
28) Message boards : Server and website : Multiple WU's sent to one host (Message 25358)
Posted 4371 days ago by Profile K1atOdessa
Only 1 GPU is allowed to use GPUGrid, with the other 3 older cards designated for other, less demanding projects (using the appropriate <exclude_gpu> options in the cc_config.xml file).

Based on the output files, GPU0 (GTX570) is the only one doing any processing. So, at least that works. The <exclude_gpu> option has been running for nearly a month on this host with no issues, until I changed the driver recently. That's why it seems like a driver-related issue, somehow confusing the server into sending WU's for all GPU's and not just those specified by BOINC options.

The other host, with the same options (but only has 1 other, older GPU) is running normally. Again, it has driver 295.73, versus the problem host with the 301.24 beta.

At this point, I'm only losing bonus credits as all the WU's are completing successfully. When I'm able (next couple days), I will upgrade to the newer 301.42 WHQL drivers and see what happens. If no change, I'll drop back to match the other host running without issues.
29) Message boards : Server and website : Multiple WU's sent to one host (Message 25339)
Posted 4372 days ago by Profile K1atOdessa
Anyone else having issues getting more than 1 or 2 WU's at a time? One host I have (see host) had about 4 or 5 WU's a couple days ago, and after completing those now has 7 outstanding. Even with a GTX570, those are still going to take 3+ days to crank through.

The only different between the two hosts (besides the actual guts of the machines) is the NVIDIA driver. The version on the host with the issues is 301.24 beta, whereas the the host without issues is running 295.75. All the BOINC (7.0.27) settings are the same, and I have low numbers in the "keep as much work for" fields to minimize this issue.

Not sure what is going on, but it seems like the server is spitting out more tasks for the newer beta driver -- though it didn't appear to happen immediately after installation (only began 2 days ago, several days after the installation of the 301.24 drivers). Before I try rolling forward to the WHQL driver 301.42 or back to the 295.73, I wanted to see if anyone else had the same issue.
30) Message boards : News : Install new drivers for new application (Message 25243)
Posted 4376 days ago by Profile K1atOdessa
Let's say around Thursday next week.

If you have the old driver, the old app will still be running, but later only the cuda4.2 application will be out for the long queue.

gdf


OK. That sounds fine. I should be able to update them next Friday (1 week from today), so it sounds like I will still be able to contribute with both machines without interruption.

Good deal.


Previous 10 | Next 10
//