s@h on GPU CUDA now?!

Message boards : Number crunching : s@h on GPU CUDA now?!
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · Next

AuthorMessage
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 9952
Credit: 7,399,343
RAC: 77
United Kingdom
Message 840060 - Posted: 15 Dec 2008, 0:44:25 UTC - in response to Message 840032.  
Last modified: 15 Dec 2008, 0:45:14 UTC

Is the Windows side debugged enough to be worth adding a whole new rat's nest into the fun?

A help or confused hindrance?

Only if you are a 100% certified masochist.

... I can be provided that the effort is not wasted...

Before I descend into Beta, how far along the Windows debug are you?

(Or do you already have your Linux experts group ready and primed?...)

Cheers,
Martin
See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 840060 · Report as offensive
Profile Byron S Goodgame
Volunteer tester
Avatar

Send message
Joined: 16 Jan 06
Posts: 1145
Credit: 3,936,993
RAC: 0
United States
Message 840064 - Posted: 15 Dec 2008, 0:57:42 UTC - in response to Message 840060.  

Is the Windows side debugged enough to be worth adding a whole new rat's nest into the fun?

A help or confused hindrance?

Only if you are a 100% certified masochist.

... I can be provided that the effort is not wasted...

Before I descend into Beta, how far along the Windows debug are you?

(Or do you already have your Linux experts group ready and primed?...)

Cheers,
Martin

It's coming along but no where near ready for prime time, though like WinterKnight eluded too, it's just perfect for you average masochistic Beta tester.
ID: 840064 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 13249
Credit: 160,562,010
RAC: 233,162
United Kingdom
Message 840276 - Posted: 15 Dec 2008, 19:26:14 UTC

There's a report that the Beta SETI@home Enhanced message board has been enabled for new testers without any Credit/RAC to post directly. Those of you already testing the Windows version may want to join in the discussions there.
ID: 840276 · Report as offensive
MarkJ Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 08
Posts: 1130
Credit: 80,663,642
RAC: 763
Australia
Message 840803 - Posted: 17 Dec 2008, 11:34:28 UTC - in response to Message 839070.  

Let the developers try to put out a stable 6.4 client first, before requesting enhancements to the textual information.


I guess BOINC 6.4.5 must be stable then, seeing as its now the release version for Windows and Linux :)
BOINC blog
ID: 840803 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 9952
Credit: 7,399,343
RAC: 77
United Kingdom
Message 840810 - Posted: 17 Dec 2008, 11:58:47 UTC - in response to Message 840803.  

I guess BOINC 6.4.5 must be stable then, seeing as its now the release version for Windows and Linux :)

That's a nice jump forward.

Any Beta people around to say how the GPU debug is progressing?

Pre- or Post- Christmas this year? ;-)


Happy fast crunchin',
Martin

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 840810 · Report as offensive
Profile skildude
Avatar

Send message
Joined: 4 Oct 00
Posts: 9533
Credit: 47,098,778
RAC: 30,008
Yemen
Message 840838 - Posted: 17 Dec 2008, 14:46:25 UTC

side question about 6.4.5 . If I'm running an ATI GPU is there any real reason to install 6.4.5?
In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope
ID: 840838 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 14893
Credit: 4,148,785
RAC: 1,216
Netherlands
Message 840840 - Posted: 17 Dec 2008, 14:49:13 UTC - in response to Message 840838.  
Last modified: 17 Dec 2008, 14:50:48 UTC

Only if you want your work fetch to be severely broken. ;-)

But no, there is no and there won't be for any time support for ATI in BOINC.
getting conflicting answers here from the developers, one saying he is in contact with the AMD/ATI developers, while the other says no-one of AMD/ATI has contacted them...

Therefore I run FAH on my ATI card. :-)

It's that 6.3.23's BOINC Manager is broken, else I would return to that version.
I'll wait for the next 6.5 alpha to come along then...
ID: 840840 · Report as offensive
Profile skildude
Avatar

Send message
Joined: 4 Oct 00
Posts: 9533
Credit: 47,098,778
RAC: 30,008
Yemen
Message 840844 - Posted: 17 Dec 2008, 14:55:51 UTC - in response to Message 840840.  

OK so I'll avoid the current 6.4.5 BOINC since it really doesnt help anything other than users of NVIDIA CUda GPU's

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope
ID: 840844 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 9952
Credit: 7,399,343
RAC: 77
United Kingdom
Message 840869 - Posted: 17 Dec 2008, 15:56:10 UTC - in response to Message 840840.  

Only if you want your work fetch to be severely broken. ;-) ...

OK... I'll bite...

What's the continuing broken bit?

Happy crunchin',
Martin

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 840869 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 13249
Credit: 160,562,010
RAC: 233,162
United Kingdom
Message 840870 - Posted: 17 Dec 2008, 16:00:21 UTC - in response to Message 840869.  

Only if you want your work fetch to be severely broken. ;-) ...

OK... I'll bite...

What's the continuing broken bit?

DCF, for one.
ID: 840870 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 9952
Credit: 7,399,343
RAC: 77
United Kingdom
Message 841152 - Posted: 17 Dec 2008, 23:01:58 UTC - in response to Message 840870.  

... What's the continuing broken bit?

DCF, for one.

I can imagine that difficult one staying broken for a while yet. Is there still not the 'argument' about the confusing amalgam of work cache and connect interval?

Part of the problem is that most of the time it works well enough...

Is not more of the problem that we need a "Full Manual Control Override" function?

Add another GUI tab for 'advanced parameters' tweaks?

Happy crunchin',
Martin

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 841152 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 9952
Credit: 7,399,343
RAC: 77
United Kingdom
Message 841153 - Posted: 17 Dec 2008, 23:04:33 UTC

Looks like there has been a good push from nVidia!

Great stuff if Boinc, s@h and e@h are getting some help with all of this.


Happy fast crunchin',
Martin


See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 841153 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 14893
Credit: 4,148,785
RAC: 1,216
Netherlands
Message 841159 - Posted: 17 Dec 2008, 23:07:36 UTC - in response to Message 841153.  

Great stuff if Boinc, s@h and e@h are getting some help with all of this.

Not 'if' but "that". It's a collaboration between Nvidia, BOINC, SAH, EAH and GPUGrid.
ID: 841159 · Report as offensive
Profile Sir Ulli
Volunteer tester
Avatar

Send message
Joined: 21 Oct 99
Posts: 2246
Credit: 6,136,250
RAC: 0
Germany
Message 841169 - Posted: 17 Dec 2008, 23:27:08 UTC

what is if you are using an opt Client with an app.xml

you dont get CUDA WUS, i think...

Greetings from Germany NRW
Ulli


ID: 841169 · Report as offensive
Profile Geek@Play
Volunteer tester
Avatar

Send message
Joined: 31 Jul 01
Posts: 2467
Credit: 86,146,931
RAC: 0
United States
Message 841172 - Posted: 17 Dec 2008, 23:34:32 UTC
Last modified: 17 Dec 2008, 23:39:09 UTC

Could an app_info file be constructed so as to allow 3 cores of a quad CPU to continue to crunch using the AK_V8 application as version 6.03 and use the 4th core with the Cuda 6.05 application?

Perhaps on second thought it's best to crunch AP on three cores and MB on one core using Cuda???

Can anyone supply the file name of the Cuda app?
Boinc....Boinc....Boinc....Boinc....
ID: 841172 · Report as offensive
Profile arkayn
Volunteer tester
Avatar

Send message
Joined: 14 May 99
Posts: 4401
Credit: 54,992,452
RAC: 66
United States
Message 841216 - Posted: 18 Dec 2008, 0:44:21 UTC - in response to Message 841172.  

setiathome_6.05_windows_intelx86__cuda.exe

ID: 841216 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 13249
Credit: 160,562,010
RAC: 233,162
United Kingdom
Message 841220 - Posted: 18 Dec 2008, 0:55:33 UTC - in response to Message 841216.  

ID: 841220 · Report as offensive
Zoran Kirsic

Send message
Joined: 22 May 99
Posts: 34
Credit: 102,258
RAC: 0
Croatia
Message 850322 - Posted: 7 Jan 2009, 1:56:43 UTC

is this a problem or what? why cuda errors!??

http://setiathome.berkeley.edu/result.php?resultid=1115022816



Task ID 1115022816
Name 24no08aa.31472.16432.10.8.42_0
Workunit 391762832
Created 7 Jan 2009 0:32:41 UTC
Sent 7 Jan 2009 1:36:23 UTC
Received 7 Jan 2009 1:42:04 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 4234493
Report deadline 14 Jan 2009 1:36:23 UTC
CPU time 46.35938
stderr out

<core_client_version>6.4.5</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_CUDA: Found 1 CUDA device(s):
Device 1 : GeForce 8600 GT
totalGlobalMem = 268107776
sharedMemPerBlock = 16384
regsPerBlock = 8192
warpSize = 32
memPitch = 262144
maxThreadsPerBlock = 512
clockRate = 1188000
totalConstMem = 65536
major = 1
minor = 1
textureAlignment = 256
deviceOverlap = 1
multiProcessorCount = 4
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce 8600 GT is okay
SETI@home using CUDA accelerated device GeForce 8600 GT
Rise priority modification by Raistmer based on rev380 of SETI@home sources
Priority of worker thread rised successfully
setiathome_enhanced 6.02 Visual Studio/Microsoft C++
libboinc: 6.3.22

Work Unit Info:
...............
WU true angle range is : 2.729029
Optimal function choices:
-----------------------------------------------------
name
-----------------------------------------------------
v_BaseLineSmooth (no other)
v_GetPowerSpectrum 0.00044 0.00000
v_ChirpData 0.01754 0.00000
v_Transpose4 0.01142 0.00000
FPU opt folding 0.00237 0.00000
Cuda error 'cudaMemcpyToSymbol(cudaAcc_GaussFit_settings, (void*) &settings, sizeof(settings))' in file 'd:/BTR/seticuda/Berkeley_rep/client/cuda/cudaAcc_gaussfit.cu' in line 454 : invalid device symbol.
Cuda error 'cudaMemcpyToSymbol(cudaAcc_PulseFind_settings, (void*) &PulseFind_settings, sizeof(PulseFind_settings))' in file 'd:/BTR/seticuda/Berkeley_rep/client/cuda/cudaAcc_pulsefind.cu' in line 669 : invalid device symbol.
Cuda error 'cudaAcc_CalcChirpData_kernel2' in file 'd:/BTR/seticuda/Berkeley_rep/client/cuda/cudaAcc_CalcChirpData.cu' in line 106 : invalid device function .
Cuda error 'cudaAcc_GetPowerSpectrum_kernel' in file 'd:/BTR/seticuda/Berkeley_rep/client/cuda/cudaAcc_PowerSpectrum.cu' in line 56 : invalid device function .
Cuda error 'cudaAcc_summax_kernel<8>' in file 'd:/BTR/seticuda/Berkeley_rep/client/cuda/cudaAcc_summax.cu' in line 156 : invalid device function .
SETI@Home Informational message -9 result_overflow
NOTE: The number of results detected exceeds the storage space allocated.

Flopcounter: 331093600.000000

Spike count: 30
Pulse count: 0
Triplet count: 0
Gaussian count: 0
called boinc_finish

</stderr_txt>
]]>

Validate state Initial
Claimed credit 0.0146733565192404
Granted credit 0
application version 6.05
ID: 850322 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6115
Credit: 101,074,330
RAC: 51,680
Russia
Message 850327 - Posted: 7 Jan 2009, 2:02:56 UTC - in response to Message 850322.  

Yes, it's problem but probably not with your setup but with app itself.
VHAR tasks result in overflow time to time with different CUDA errors.
ID: 850327 · Report as offensive
Zoran Kirsic

Send message
Joined: 22 May 99
Posts: 34
Credit: 102,258
RAC: 0
Croatia
Message 850330 - Posted: 7 Jan 2009, 2:06:56 UTC

can I fix it somehow?
ID: 850330 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · Next

Message boards : Number crunching : s@h on GPU CUDA now?!


 
©2019 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.