ABP2 CUDA applications

astrocrab
astrocrab
Joined: 28 Jan 08
Posts: 208
Credit: 429,202,534
RAC: 0

RE: The next batch will

Message 96293 in response to message 96284

Quote:

The next batch will become available within the next 24 hours. Stay tuned.
HB

where are new ABP2 tasks?

14.01.2010 8:40:31 Einstein@Home Requesting new tasks for GPU
14.01.2010 8:40:36 Einstein@Home Scheduler request completed: got 0 new tasks
14.01.2010 8:40:36 Einstein@Home Message from server: No work sent

astrocrab
astrocrab
Joined: 28 Jan 08
Posts: 208
Credit: 429,202,534
RAC: 0

I'm still getting 3.13,

I'm still getting 3.13, ABP1

there are "cpu task - no", gpu "task - yes" set in preferences

Svenie25
Svenie25
Joined: 21 Mar 05
Posts: 139
Credit: 2,436,862
RAC: 0

Looks likes the

Looks likes the validationproblem Windows vs. Linux is still alive:

http://einsteinathome.org/workunit/65824840

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,222
Credit: 137,332,827
RAC: 24,309

From Bernd : The new

From Bernd :

The new generation of ABP2 Apps that should come out today has a couple
of changes to the code:

1. The template bank is written in double precision numbers, but they
are read into single precision variables. Apparently the system routines
of Windows and Linux do a different rounding when doing so. So we now
read them in in double precision and do the rounding explicitly.

2. The resampling code is rewritten to avoid to reach the precision
limit of single precision calculations.

3. For CUDA Apps only: When a client requests GPU work, it is checked
whether the GPU has enough physical ('global') memory to run the App.
However it often happens that this memory isn't available when the Task
is actually started. Previously such tasks would error out (exit code
3), which is by far the most common error of CUDA App on Einstein@home.
David Anderson implemented a way that should avoid this. An App can now
fail temporarily on memory allocation, and will then be started again
later when the memory becomes available. This requires changes to
server, client and app. The server changes are in place now, the new
generation of Apps should feature the app part. It is now up to
participants to upgrade the client, I think 6.10.26+ will do to use this
feature.

There might be additional validation problems between old and new
generation of ABP2 Apps, so I wanted to get as many tasks 'sent' (and
thus consistently assigned to old generation apps) as possible and will
the WUG keep stopped until the new Apps are out and all new workunits
will be calculated by those.

The Linux 1.04 App is meant to fix a library dependency problem of 1.02
that is specific to Linux, it is not a 'new generation' ABP2 App.

and Oliver :

One minor addition to this: the new CUDA releases (all platforms) will
feature dedicated exit codes for all CUDA-related errors (current range:
1001-1013). This way we can get error statistics much easier and focus
on the real issues.

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

Elphidieus
Elphidieus
Joined: 20 Feb 05
Posts: 234
Credit: 15,951,593
RAC: 10,116

Hi guys, Any idea why my

Hi guys,

Any idea why my CUDA workunits just sits there and doing nothing while the CPU's crunching away....? I'm on 6.10.29 with Cuda 2.3 drivers installed, running on Mac OS X Snow Leopard.

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3,516
Credit: 471,303,446
RAC: 52,838

RE: Hi guys, Any idea why

Message 96298 in response to message 96297

Quote:

Hi guys,

Any idea why my CUDA workunits just sits there and doing nothing while the CPU's crunching away....? I'm on 6.10.29 with Cuda 2.3 drivers installed, running on Mac OS X Snow Leopard.

Hi,

Most likely (as it is the default) your account is set to "do NOT use GPU when user is active", that is, if BOINC considers the Mac to be "in use", BOINC will not allow the GPU app to run. I think this is mostly done by watching for keyboard and mouse activity, and if there is none for a certain time, the Mac is "idle" as far as BOINC is concerned.

So just to test this, you might want to temporarily configure your account in the web preferences to use the GPU even if the user is active. Will it run GPU tasks now?

Are there any informative messages in the message tab ?

Greeting to Malaysia (I've been there a couple of months last year).

CU
HB

Elphidieus
Elphidieus
Joined: 20 Feb 05
Posts: 234
Credit: 15,951,593
RAC: 10,116

RE: Hi, Most likely (as

Message 96299 in response to message 96298

Quote:


Hi,

Most likely (as it is the default) your account is set to "do NOT use GPU when user is active", that is, if BOINC considers the Mac to be "in use", BOINC will not allow the GPU app to run. I think this is mostly done by watching for keyboard and mouse activity, and if there is none for a certain time, the Mac is "idle" as far as BOINC is concerned.

So just to test this, you might want to temporarily configure your account in the web preferences to use the GPU even if the user is active. Will it run GPU tasks now?

Are there any informative messages in the message tab ?

Greeting to Malaysia (I've been there a couple of months last year).

CU
HB

Hi Bikeman,

Well I've got all GPU related settings enabled, including using GPU when user is active or when computer is in use. I've even tried letting my Mac idle and display-sleep for over 10 minutes. Nothing happens. There was also no error messages of any kind, except for the following:

Sun Jan 17 20:11:16 2010 NVIDIA GPU 0: GeForce GT 120 (driver version unknown, CUDA version 2030, compute capability 1.1, 512MB, 90 GFLOPS peak)

Driver version unknown...?

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3,516
Credit: 471,303,446
RAC: 52,838

Strange. Another possibility

Message 96300 in response to message 96299

Strange. Another possibility is that BOINC core client thinks that the S5R6 tasks should be given priority because of their deadline (remember, the ABP2 tasks will also need a full CPU core in addition to the GPU).

The "driver version unknown" is quite common and shouldn't be a problem.

Anyway, I'm afraid that the "90 GFlops" CUDA card would not do very well anyway, it could even diminish throughput. depending on the CPU type.

CU
H-B

Elphidieus
Elphidieus
Joined: 20 Feb 05
Posts: 234
Credit: 15,951,593
RAC: 10,116

RE: Strange. Another

Message 96301 in response to message 96300

Quote:

Strange. Another possibility is that BOINC core client thinks that the S5R6 tasks should be given priority because of their deadline (remember, the ABP2 tasks will also need a full CPU core in addition to the GPU).

The "driver version unknown" is quite common and shouldn't be a problem.

CU
H-B

But that should not be a problem because the S5R6 was not there when the ABP2 workunits completed their downloads. It didn't even start when ABP2 was the only workunit on the cache... So now it's even more strange...

GPV67
GPV67
Joined: 26 Mar 05
Posts: 10
Credit: 17,901,559
RAC: 301

RE: RE: Strange. Another

Message 96302 in response to message 96301

Quote:
Quote:

Strange. Another possibility is that BOINC core client thinks that the S5R6 tasks should be given priority because of their deadline (remember, the ABP2 tasks will also need a full CPU core in addition to the GPU).

The "driver version unknown" is quite common and shouldn't be a problem.

CU
H-B

But that should not be a problem because the S5R6 was not there when the ABP2 workunits completed their downloads. It didn't even start when ABP2 was the only workunit on the cache... So now it's even more strange...


Hi,
I have the same behavior on my MacPro 2008 (8 core + GeForce 8800GT) with STSP (5.07) units.
Units are successfully downloaded, but never computed.
There was no issue with STSP 5.03 application.
Note: Issue with both BOINC manager 6.10.28 and 6.10.29

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.