Welcome to MilkyWay@home

Posts by cristipurdel

1) Message boards : Application Code Discussion : Android/Linux GPU app beta (Message 66968)
Posted 15 Jan 2018 by cristipurdel
Post:
Any chance of the getting an Android/Linux GPU app as a beta tester?

If it was done on linux http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=4180, maybe it could be done also on Android
2) Message boards : News : maximum time limit elapsed bug (Message 49973)
Posted 7 Jul 2011 by cristipurdel
Post:


I wouldn't give much on this information, I think this is not a measured value, it's only an information from the App-Server which does your card only identify as "HD 6900 series (Cayman)".

May I ask U another question? Why have you down clocked your GPU-Core instead of the memory. For MW@H only the core speed is signficant for the performance.
The memory speed you can downclock as low as your system runs stable (50 % of stock speed should work fine), for saving energie and lower temperature.

greetings
franz


I'm also running 4 wcg 64 bit applications.
I want to keep my pc as cool as it can be. Right now the gpu hovers around 72C, while the cpu is at 77C.

From my experience with gpu load vs gpu temp, it's much better to lower the clocks than to limit the gpu load (e.g. tthrottle) for the same temperature.
500/1250 is the lowest I can go with amd overdrive.

And I also have to consider the speed of the fan which becomes annoying when it goes into a 'higher' gear :P
3) Message boards : News : maximum time limit elapsed bug (Message 49967)
Posted 7 Jul 2011 by cristipurdel
Post:
Brief note for those who wish to try an optimised application to overcome maximum time limit elapsed bug on ATI GPUs and are unfamiliar with app_info.xml files and the oddities of files compressed on a Mac. This relates to the optimised applications previously mentioned found on arkayn's Crunchers Anonymous forum.

For Windows 64-bit you should use the correct app_info.xml file from the "Win64_0.82_ati\Win64_0.82_ati\Files to install" folder and not the ._app_info.xml file from the "Win64_0.82_ati\__MACOSX\Win64_0.82_ati\Files to install" folder which contains some strange looking Mac zipping stuff:   Mac OS X  2 u  § ATTR z± § œ  œ com.apple.TextEncoding macintosh;0

For Windows 32-bit you should use the correct app_info.xml file from the "Win32_0.82_ati\Win32_0.82_ati\Files to install" folder and not the ._app_info.xml file from the "Win32_0.82_ati\__MACOSX\Win32_0.82_ati\Files to install" folder which contains similar Mac stuff as 64-bit version above.

For both 64-bit and 32-bit versions you do not need to copy the .DS_Store file.

The optimised application itself is the same as the default application automatically downloaded from the MilkyWay server. For those who are not having any problems there is very little to no advantage to use an optimised application, it only allows some parameters to be added/altered for those who have trouble such as a sluggish user interface or those who wish to tweak a bit.


thx ... it worked

Bu, for the record, amd just so I won't die stupid...

For my 6950, at 500/1250 MHz, from http://en.wikipedia.org/wiki/Comparison_of_AMD_graphics_processing_units#Northern_Islands_.28HD_6xxx.29_series my peak SP power should be
2253 which is 1408*800*2

After I down-clocked the GPU from 800 to 500, boinc says I should have 1760GFLOPS
ATI GPU 0: AMD Radeon HD 6900 series (Cayman) (CAL version 1.4.1417, 1024MB, 1760 GFLOPS peak)

When Actually I'm having 1408= 1408*2*500

I know this is for SP, and for DP my peak power is 25% for Cayman.

So, my DP should hover in the vicinity of 352-440GFLOPS

The question is, since I had the TE bug, and the fact that I have 1e11 flops (=100 GFLOPS) in my app_info (which is lower than my DP), in order to bypass the TE bug, the flops value in app_info should be lower than the peak DP of anyone's card?
4) Message boards : News : maximum time limit elapsed bug (Message 49941)
Posted 6 Jul 2011 by cristipurdel
Post:
So on my 6950, with the new boinc 6.12.33, I dwld several GPU WUs.
After it starts crunching normal on the first one(it takes an estimated time of 4min with 500/1250 MHz), at around 40% the WU gets aborted. Than a massive system freeze happens for about 30-60 seconds, in which I presume, the second WU starts, and the first WU finishes as a computational error.
@Arkayn maybe a type of installer like the one from lunatics would be more suitable. With all the flags being inside the installer, along with some explanations, so that everybody could tinker his own app based on community feedback.
5) Message boards : News : maximum time limit elapsed bug (Message 49918)
Posted 5 Jul 2011 by cristipurdel
Post:
Let's try a different approach. Is someone with a 6950 or other ATI cards which doesn't experience this problem and has a lower version of Cat 11.6?

Have a scroll through the top computers list...

A more easier approach...use the card on something that works for me like seti and collatz
6) Message boards : News : maximum time limit elapsed bug (Message 49915)
Posted 5 Jul 2011 by cristipurdel
Post:
Let's try a different approach. Is someone with a 6950 or other ATI cards which doesn't experience this problem and has a lower version of Cat 11.6?
7) Message boards : News : maximum time limit elapsed bug (Message 49894)
Posted 5 Jul 2011 by cristipurdel
Post:
I'm also experiencing the same problem on Win 7 x64, Cat 11.6 on a 6950
How do you reset or increase the DCF?
8) Message boards : Number crunching : ps_separation and ps_test WUs all erroring out (Message 49870)
Posted 4 Jul 2011 by cristipurdel
Post:
On my host http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=299828
all the gpu WUs ps_separation_13_3s ended with computational error.
eg. http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=64434802
Outcome Computation error
Client state Compute error
Exit status -177 (0xffffffffffffff4f)
In other cases I also got the -179 error
Previous I had a FX 580, removed it, using driver sweeper, and insttalled a 6950
My specs are Win 7 64, 6.12.26 x64, Cat 11.6, 6950 downclocked to 500/1250
Seti with ATI OpenCL build seams to be working ok.
Any suggestions?




©2021 Astroinformatics Group