Posts by Marty

\n studio-striking\n
1) Message boards : News : Switched to granting static credit (Message 452)
Posted 23 May 2011 by Profile Marty
Post:
I also see different credit amounts for different combinations.
My single GPU host (HD5850) gets more credit here, another single GPU host (HD3850) and the dual GPU host (HD5870 and HD5850) get more credit at DNETC.


using core #0 - dunno why i can't select #3 on DNet.

I guess this has to do with the different version of the dnet-client used over there since the -bench command with the DNETC-version also only tests core #0, #1 and #2 unlike the version used here which also tests core #3
2) Questions and Answers : Windows : 20 tasks per day? (Message 385)
Posted 19 May 2011 by Profile Marty
Post:

Well, it happens to any machine where I abort 20 tasks or more. In this case it is this one:

http://moowrap.net/show_host_detail.php?hostid=118

Not really a surprise if you are aborting WUs. At some projects that brings down the tasks/day limit to prevent generation of huge amounts of invalid/aborted WUs by a single host.
It usually will go up again with valid WUs delivered by that host.
3) Questions and Answers : Windows : 6990? (Message 356)
Posted 18 May 2011 by Profile Marty
Post:
Or you could buy HD6950s and try to mod them into HD6970s (http://www.techpowerup.com/articles/overclocking/vidcard/159) and send back the ones which don't work.
4) Questions and Answers : Windows : Long run times (Message 346)
Posted 17 May 2011 by Profile Marty
Post:
This is crazy... It is taking two hours to complete one WU on a single 5850 card in my i7.

Same problem here.
i7 930 - ati 5850 - win 7 64bit and two hours for a single WU.
Any solution/fix/help/exorcism?

Intel(R) CPU 2160@3.15GHz - ATI HD5850 (Catalyst 11.3) - Win7 x64: runtime around 20 minutes (without dedicated CPUs)
5) Questions and Answers : Windows : Multi-GPU task takes longer than single-gpu task (Message 334)
Posted 16 May 2011 by Profile Marty
Post:
Vista64 with 1xHD5870 and 1xHD5850 (Catalyst 11.5) running on a AMD Phenom II x4 955

After running the -bench command i switched to the recommended core 3, but the load of the cards is still not over 90% on both of them.

dnetc v2.9109-518-GTR-10092921 for ATI Stream on Win32 (WindowsNT 6.0).
Please provide the *entire* version descriptor when submitting bug reports.
The distributed.net bug report pages are at http://bugs.distributed.net/

[May 16 15:23:57 UTC] RC5-72: using core #0 (IL 4-pipe c).
[May 16 15:24:02 UTC] RC5-72: Benchmark for core #0 (IL 4-pipe c)
0.00:00:02.26 [1,944,551,487 keys/sec]
[May 16 15:24:02 UTC] RC5-72: using core #1 (IL 4-pipe c alt).
[May 16 15:24:08 UTC] RC5-72: Benchmark for core #1 (IL 4-pipe c alt)
0.00:00:03.86 [1,128,436,111 keys/sec]
[May 16 15:24:08 UTC] RC5-72: using core #2 (IL 4-pipe 2 threads).
[May 16 15:24:29 UTC] RC5-72: Benchmark for core #2 (IL 4-pipe 2 threads)
0.00:00:17.23 [249,835,586 keys/sec]
[May 16 15:24:29 UTC] RC5-72: using core #3 (IL 4-pipe cs-1).
[May 16 15:24:33 UTC] RC5-72: Benchmark for core #3 (IL 4-pipe cs-1)
0.00:00:02.23 [1,998,692,544 keys/sec]
[May 16 15:24:33 UTC] RC5-72 benchmark summary :
Default core : #0 (IL 4-pipe c)
Fastest core : #3 (IL 4-pipe cs-1)
[May 16 15:24:33 UTC] Core #3 is marginally faster than the default core.
Testing variability might lead to pick one or the other.


Then i suspended the BOINC CPU tasks and the load on both GPUs went up to over 95% but the CPU load for dnetc518-win32-x86-stream.exe went to 50%, meaning it used 2 full cores. This was never the case at DNETC. Something is still not right there.
Also noticed that the 1 packet on each GPU finished within a couple seconds, at least according to the load decreases visible with MSI Afterburner and the percentage display within BOINC ??? Also the GPU load indicates that the single packets might have varying runtimes on the same GPU ???

With the BOINC CPU tasks suspended the runtime also improved, as expected, since the load on the GPUs inreceased.

DNETC used "distributed.net v2.9108-517 client for ATI Stream on Win32" (see bench result below) with the following job.xml file for ATI cards:
<job_desc>
<task>
<application>dnetc_1.31_windows_intelx86__ati14.exe</application>
<command_line>-runoffline -multiok=1 -ckpoint chkpoint -pausefile pause -exitfile exit -inbase in -outbase out -priority 5 -n -1 -runbuffers -l stderr.txt</command_line>
</task>
</job_desc>


dnetc v2.9108-517-GTR-10021520 for ATI Stream on Win32 (WindowsNT 6.0).
Please provide the *entire* version descriptor when submitting bug reports.
The distributed.net bug report pages are at http://bugs.distributed.net/

[May 16 15:38:05 UTC] RC5-72: using core #0 (IL 4-pipe c).
[May 16 15:38:10 UTC] RC5-72: Benchmark for core #0 (IL 4-pipe c)
0.00:00:02.38 [1,835,759,074 keys/sec]
[May 16 15:38:10 UTC] RC5-72: using core #1 (IL 4-pipe c alt).
[May 16 15:38:16 UTC] RC5-72: Benchmark for core #1 (IL 4-pipe c alt)
0.00:00:03.90 [1,119,097,905 keys/sec]
[May 16 15:38:16 UTC] RC5-72: using core #2 (IL 4-pipe 2 threads).
[May 16 15:38:23 UTC] RC5-72: Benchmark for core #2 (IL 4-pipe 2 threads)
0.00:00:03.72 [1,182,694,097 keys/sec]
[May 16 15:38:23 UTC] RC5-72 benchmark summary :
Default core : #0 (IL 4-pipe c)
Fastest core : #0 (IL 4-pipe c)


Is there a reason why you use "distributed.net v2.9109-518" opposed to "distributed.net v2.9108-517" here?
6) Message boards : News : Switched to granting static credit (Message 315)
Posted 15 May 2011 by Profile Marty
Post:
@AriZonaMoon: Only 4 words for you: You don't get it.

For me "credit whores" are not the people accumulating more/much credit. For me cw are people asking for more credit than justified or people who moan if a unfair credit system is corrected.

I e.g. got for one host (WinXP32 AthlonXP 2800) the following credit for a single HD3850:
Completed and validated 15,615.52 273.38 15,719.53
Completed and validated 19,393.97 299.16 19,523.15
Completed and validated 16,514.00 266.19 16,596.65

This is about 1 Cr/s. This is ridiculously high for this card.

Another of my hosts (Vista64 Phenom II 955) got the following credit for a HD5870 and a HD5870:
Completed and validated 837.84 109.70 9,339.50
Completed and validated 829.16 178.01 9,261.85
Completed and validated 923.14 3.62 10,272.77

This is about 11 Cr/s. This is also too high for those 2 cards combined.
7) Message boards : News : Switched to granting static credit (Message 312)
Posted 14 May 2011 by Profile Marty
Post:
Congratulation admins on the right decision to go for a fixed credit system based on work achieved. Much fairer this way.

The previous credit output was ridiculously high. And don't worry about the moaning of the credit whores, they will not be missed or come back, i case you run a stable and reliable project, if the now supposedly higher projects are unstable or down.

If now you could only fix the problem for some multi GPU hosts where only some of the GPUs run at full power. This would be beautiful.
8) Questions and Answers : Windows : Multi-GPU task takes longer than single-gpu task (Message 85)
Posted 4 May 2011 by Profile Marty
Post:

I think the problem with Dnetc was that some work units had an odd number of packets, so one GPU would still be working while the other had nothing to do. Then it took a few workunits for the other one to kick back in again. So a WU with 12 packets would run normal with 2 GPU's, yet one with 11 or 13 packets would cause at least one (if not both) GPU to hang.

The case of the uneven packets explains the idle GPU at the end or the faster GPU idling with even packets.
I see one GPU run 50% through most/all of it's packets while the other one runs at 100%. It all might be connected with the high CPU issue described in the thread. Haven't had time to test this by disabling CPU computation and running Moowrap exclusively.

Edit:
Ok for this kind of credit i can live with idling GPUs :D
14379 12212 3 May 2011 | 12:13:55 UTC 3 May 2011 | 16:18:37 UTC Completed and validated 681.87 62.29 8,590.69 Distributed.net Client v1.00 (ati14)
14183 12017 3 May 2011 | 12:13:55 UTC 4 May 2011 | 6:10:40 UTC Completed and validated 720.32 5.48 9,075.10 Distributed.net Client v1.00 (ati14)
14179 12013 3 May 2011 | 12:13:55 UTC 4 May 2011 | 6:10:40 UTC Completed and validated 697.41 19.08 8,786.55 Distributed.net Client v1.00 (ati14)
9) Questions and Answers : Windows : Multi-GPU task takes longer than single-gpu task (Message 66)
Posted 3 May 2011 by Profile Marty
Post:
Question for project admin:
Assuming this project is built on DNETC sources: Was something changed in the ATI application.
On my multi GPU system (1x 5870 and 1x 5850) it behaves differently than on DNETC. One of the two GPUs only gets back up to 50% GPU usage after it finished the first segment (block of work). Previously both GPUs were running over 90% GPU usage the whole time, except for the end when the faster GPU has to wait for the slower on to finish it's last segment.





 
Copyright © 2011-2021 Moo! Wrapper Project