Message boards :
News :
stock ATI 58x0 apps updated
Message board moderation
Author | Message |
---|---|
Send message Joined: 30 Aug 07 Posts: 2046 Credit: 26,480 RAC: 0 |
I've updated the stock 58x0 ATI GPU applications. Let me know if they download and work correctly. |
Send message Joined: 19 Feb 10 Posts: 17 Credit: 7,573,117 RAC: 0 |
I've read that 48xx GPU is the one correctly validating against the stock application. But I have problems with my 4890 ati card some wus returning as validation validation inconclusive. So I'll try to run the 0.23 today. |
Send message Joined: 27 Nov 09 Posts: 108 Credit: 430,760,953 RAC: 0 |
After project detach/reattach, it cleaned out all the old code and downloaded the new 0.23 (GPU) and old 0.19 (CPU) apps just fine. It's enqueued 24 GPU WU's here. Is it normal to have to detach the project first? Just restarting BOINC manager/service didn't seem to get the job done. |
Send message Joined: 30 Aug 07 Posts: 2046 Credit: 26,480 RAC: 0 |
I don't know about having to reattach, but I'm glad to hear about them working. From what I've been seeing so far in the validator is that the new ATI application is validating correctly vs stock applications, and not against the older 58x0 application. So I think this will fix the problem once everyone gets updated :) |
Send message Joined: 27 Nov 09 Posts: 108 Credit: 430,760,953 RAC: 0 |
I notice that the old 0.21 app now shows "v0.00" in the task stats even for 4xxx-series cards. |
Send message Joined: 30 Aug 07 Posts: 2046 Credit: 26,480 RAC: 0 |
I notice that the old 0.21 app now shows "v0.00" in the task stats even for 4xxx-series cards. I removed the old app from the database (which typically gets the client to swap over to the new app faster). v0.21 has been updated for 4xxx as well. Also, part of the issue here is I don't think the server really knows the difference between 4xxx and 5xxx as they're running the same application. |
Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0 |
From what I've been seeing so far in the validator is that the new ATI application is validating correctly vs stock applications, and not against the older 58x0 application. So I think this will fix the problem once everyone gets updated :) Yes, I've seen that too. But on some hosts I had a look at it appears to be a mixed bag (or quite a lot of people overclock their GPUs too much). Hmm, maybe we need a little bit more feedback. |
Send message Joined: 30 Aug 07 Posts: 2046 Credit: 26,480 RAC: 0 |
From what I've been seeing so far in the validator is that the new ATI application is validating correctly vs stock applications, and not against the older 58x0 application. So I think this will fix the problem once everyone gets updated :) Would there be an issue if they aren't running the right driver? Or would the application not run at all. |
Send message Joined: 27 Nov 09 Posts: 108 Credit: 430,760,953 RAC: 0 |
I removed the old app from the database (which typically gets the client to swap over to the new app faster).This might be causing a cosmetic issue with the Workunit status pages? All of them I've looked at now say 'This is displayed on the workunit page' followed underneath by 'Database Error'. Were the fitness values removed? |
Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0 |
From what I've been seeing so far in the validator is that the new ATI application is validating correctly vs stock applications, and not against the older 58x0 application. So I think this will fix the problem once everyone gets updated :) It *should* run with any driver version from the last 12 months and *should* error out with older ones (prior to Catalyst 9.3), or actually it shouldn't download the application because the plan_class requires the newer CAL libraries. But I've done no tests for this of course. After two guys from my team reported the new application runs and returned the same value on a HD5850 as on a HD3800/4800 GPU for a test WU I just compiled the three different versions und uploaded it. So maybe there are still some quirks hidden. |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
Getting "ignores" inside the stderr_txt's - about 20% show no ignores, the rest do. Example: No Ignores Contain's ignores Both from the same batch of four. App Info is: <app_info> <app> <name>milkyway</name> </app> <file_info> <name>astronomy_0.23_ATI_x64.exe</name> <executable /> </file_info> <file_info> <name>brook64.dll</name> <executable /> </file_info> <app_version> <app_name>milkyway</app_name> <version_num>23</version_num> <plan_class>ati13ati</plan_class> <flops>1.0e11</flops> <avg_ncpus>0.05</avg_ncpus> <max_ncpus>1</max_ncpus> <coproc> <type>ATI</type> <count>0.5</count> </coproc> <cmdline /> <file_ref> <file_name>astronomy_0.23_ATI_x64.exe</file_name> <main_program /> </file_ref> <file_ref> <file_name>brook64.dll</file_name> </file_ref> </app_version> <app_version> <app_name>milkyway</app_name> <version_num>22</version_num> <flops>1.0e11</flops> <avg_ncpus>0.05</avg_ncpus> <max_ncpus>1</max_ncpus> <coproc> <type>ati13ati</type> <count>0.5</count> </coproc> <cmdline /> <file_ref> <file_name>astronomy_0.23_ATI_x64.exe</file_name> <main_program /> </file_ref> <file_ref> <file_name>brook64.dll</file_name> </file_ref> </app_version> </app_info> No apparent invalids yet - is a problem my end? Regards Zy |
Send message Joined: 30 Aug 07 Posts: 2046 Credit: 26,480 RAC: 0 |
Looks like it was a problem in the validator. When I ported over the boinc validator code to Java there was a limit to the amount of credit that could be granted per CPU second -- of course this ignored time taken on the GPU. This should be fixed now. I think the ones that didn't get credit should go through the validator again. |
Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0 |
Getting "ignores" inside the stderr_txt's - about 20% show no ignores, the rest do. That is a side effect of the plan to omit one parameter file per WU and to use command line parameters for it (set per WU by the project, not in the app_info.xml, so the warning message is not completely correct). In the moment, both approaches are used (command line parameters and input file). The current application simply does not understand those command line parameters yet and warns about it. This will be used in the new MW appplications Travis was talking about. |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
Okie Doke - while I'm here ..... add my thanks for your speedy response to all this re the apps - Kudos :) Regards Zy |
Send message Joined: 24 Feb 09 Posts: 620 Credit: 100,587,625 RAC: 0 |
Found a strange one - not a problem to me - just highlighhting a possible anomoly. Two V23's and a V0.0, mine, a V23, was invalidated. Possible incorrect Validation? Regards Zy |
Send message Joined: 20 Jul 09 Posts: 6 Credit: 3,000,120,629 RAC: 0 |
Good morning Vietnam, I'm trying to get some WUs after reataching to MW but I am unable to initialize the project although servers are working fine, Please let me know what is the issue here, Regards, Greg Granowski |
Send message Joined: 18 Nov 07 Posts: 280 Credit: 2,442,757 RAC: 0 |
Now that the ATI applications have been updated, I'm curious - Cluster Physik mentioned that with a few changes, the CUDA applications could give exactly the same results as the ATI and CPU applications. They are currently close (well within tolerances) but not the same. Do you intend to update the CUDA applications based on his advice, or does the current work updating them with the new science and server communication code prevent this from being a priority? |
Send message Joined: 18 Nov 07 Posts: 280 Credit: 2,442,757 RAC: 0 |
[double post, please delete] |
Send message Joined: 14 Dec 09 Posts: 161 Credit: 589,318,064 RAC: 0 |
My older wus are seen as V0.00 also. The server is out of work. |
Send message Joined: 10 Feb 10 Posts: 6 Credit: 157,756,946 RAC: 0 |
Good day, The new stock application v. 0.23 for ATI GPU has downloaded automatically and is running on my GPU, with several dozens of WU's already finished. Some of those have been validated OK, but some were either considered invalid, or the validation is reported "inconclusive". Thus I guess the problem is still there. Not overclocking. Cheers, ivk |
©2024 Astroinformatics Group