I'm waisting resources by running 1 CPU+GPU task, cause even 4 at once on an GTX480, is no problem, even on an 470, which I tried. GPU load 80% (highest), avg 70%.
(A supid parse error in app_info.xml, lost some WU, but Resent Option turned on, got them back.....)
This happened some time ago, have make an 'exceptable app_info.xml file, but sometimes is nor quite clear what this parse error can be?
When running SETI MB WU's, also 2 at a time, the GPU load is =>98%
Card's also run hot, doing 2 MB per card and 4 on CPU.
I assume this Arecibo Binary Pulsar Search (STSP) v3.11 (ABP2cuda23), is the latest version and I seem to run Arecibo Binary Pulsar Search (STSP) v3.08 .
This one is hard to compair, cause 2, completely different card-architecture
are used, 200 series and 400 FERMI series, but I don't know the precise difference between these two applications.
This task is
(partly) done on 3 different O.S.ses, unfortunatly with no success for Linux and
Darwin. The last 2 on CPU only.
And a bit OFF TOPIC, too. Question is which version, v3.11(CUDA 2.3) or
v3.08(CUDA 3.1) , is the most efficient?
Again, the type of card's used, the 200 or 400 series, does a better job, when running CUDA 2.3 or 3.0/3.1? I don't know if the version numbers, have anything todo with
the CUDA version, used.
Could it be possible that 'older' or 200 series NVidia, run faster on CUDA 2.3
(difference in speed-up, between 2.2 and 2.3, are known) and 400 series or FERMI's do better with CUDA 3.1?
But I know, the difference, between CUDA 2.2 and 2.3, in SETI's
MB (Multi Beam), is quite big in speed up. Difference between 2.3 and 3.1, isn't
that big, or even neglectable.
Gaining performance of GPU's, on Einstein's CUDA tasks, leaves one
option, running multiple tasks, at a time and one has to make sure, there's enough memory available and the card is 'Up to it'[
During the large block of "only SETI WUs" above those WUs there were probably only 2 SETIs running at any one time.
You're right, for quite some time, EAH tasks were just waiting for their turn.
Don't be misled by large block of "only SETI WUs" above: they are there due to SAH servers being offline for some time so my machine didn't have opportunity to report them back, while EAH tasks have been reported as they went.
So if there is someone out there who can create an app_info.xml , PLEASE HELP !
Well, actually it is pretty easy. The structure of the app_info.xml is documented here: http://boinc.berkeley.edu/wiki/Anonymous_platform and everything else you need you'll find in the existing client_state.xml in your BOINC data directory (which has the same structure, only some additional information).
It might take a bit longer to create a file for Einstein@home than for other projects, cause more applications and files are involved here, but it is not more difficult.
For a start, this is the part you'll need for APB2 CUDA:
Thanks for starter.
I finally created an app_info.xml that works for me (Mac OS X). The structure of the xml file was not the problem. It was determining which tags I needed and the correct values.
Naam p2030_53696_28007_0118_G203.36+00.90.N_1.dm_610_15
Werkeenheid 84284590
Aangemaakt 13 Nov 2010 11:56:27 UTC
Verzonden 13 Nov 2010 22:48:46 UTC
Received 18 Nov 2010 1:59:50 UTC
Server status Over
Uitkomst Fout tijdens controle (32:00100000)
Client status Gereed
Exit status 0 (0x0)
Computer ID 3311912
Report deadline 27 Nov 2010 22:48:46 UTC
Loop tijd 14,242.87
CPU time 13,007.59
Validatie status Ongeldig
Claimed credit 99.10
Granted credit 0.00
programma versie Arecibo Binary Pulsar Search (STSP) v3.08
Haven't seen this one , header above, is a mystery, to me?!
[ADDED] Are those WU's validated by 'a script or by hand'?
I'm waisting resources by
)
I'm waisting resources by running 1 CPU+GPU task, cause even 4 at once on an GTX480, is no problem, even on an 470, which I tried. GPU load 80% (highest), avg 70%.
(A supid parse error in app_info.xml, lost some WU, but Resent Option turned on, got them back.....)
This happened some time ago, have make an 'exceptable app_info.xml file, but sometimes is nor quite clear what this parse error can be?
When running SETI MB WU's, also 2 at a time, the GPU load is =>98%
Card's also run hot, doing 2 MB per card and 4 on CPU.
Also noticed 2 app.'s are
)
Also noticed 2 app.'s are used, or can be used.
I assume this Arecibo Binary Pulsar Search (STSP) v3.11 (ABP2cuda23), is the latest version and I seem to run Arecibo Binary Pulsar Search (STSP) v3.08 .
This one is hard to compair, cause 2, completely different card-architecture
are used, 200 series and 400 FERMI series, but I don't know the precise difference between these two applications.
This task is
(partly) done on 3 different O.S.ses, unfortunatly with no success for Linux and
Darwin. The last 2 on CPU only.
And a bit OFF TOPIC, too. Question is which version, v3.11(CUDA 2.3) or
v3.08(CUDA 3.1) , is the most efficient?
Again, the type of card's used, the 200 or 400 series, does a better job, when running CUDA 2.3 or 3.0/3.1?
I don't know if the version numbers, have anything todo with
the CUDA version, used.
Could it be possible that 'older' or 200 series NVidia, run faster on CUDA 2.3
(difference in speed-up, between 2.2 and 2.3, are known) and 400 series or FERMI's do better with CUDA 3.1?
But I know, the difference, between CUDA 2.2 and 2.3, in SETI's
MB (Multi Beam), is quite big in speed up. Difference between 2.3 and 3.1, isn't
that big, or even neglectable.
Gaining performance of GPU's, on Einstein's CUDA tasks, leaves one
option, running multiple tasks, at a time and one has to make sure, there's enough memory available and the card is 'Up to it'[
Hi! Only the x.11 series
)
Hi!
Only the x.11 series apps are CUDA apps, the x.08 apps only use the CPU. So you definitely can't compare their runtimes.
CU
HB
RE: During the large block
)
You're right, for quite some time, EAH tasks were just waiting for their turn.
Don't be misled by large block of "only SETI WUs" above: they are there due to SAH servers being offline for some time so my machine didn't have opportunity to report them back, while EAH tasks have been reported as they went.
Metod ...
Why does your Boinc screen
)
Why does your Boinc screen have additional columns,CPU% ,History Tab.?I am sort of new at this, are you using a different version of Boinc ?
Thanks in advance,
Oldtroupador
RE: Why does your Boinc
)
He is running a second party manager called BOINC TASKS You can find it here.. http://www.efmer.eu/boinc/boinc_tasks/
PROUD MEMBER OF Team Starfire World BOINC
RE: RE: So if there is
)
Thanks for starter.
I finally created an app_info.xml that works for me (Mac OS X). The structure of the xml file was not the problem. It was determining which tags I needed and the correct values.
Ted
Can't 'see' any error?
)
Can't 'see' any error?
RE: Can't 'see' any error?
)
I've got one of those too (WU 87678429).
RE: RE: Can't 'see' any
)
Naam p2030_53696_28007_0118_G203.36+00.90.N_1.dm_610_15
Werkeenheid 84284590
Aangemaakt 13 Nov 2010 11:56:27 UTC
Verzonden 13 Nov 2010 22:48:46 UTC
Received 18 Nov 2010 1:59:50 UTC
Server status Over
Uitkomst Fout tijdens controle (32:00100000)
Client status Gereed
Exit status 0 (0x0)
Computer ID 3311912
Report deadline 27 Nov 2010 22:48:46 UTC
Loop tijd 14,242.87
CPU time 13,007.59
Validatie status Ongeldig
Claimed credit 99.10
Granted credit 0.00
programma versie Arecibo Binary Pulsar Search (STSP) v3.08
Haven't seen this one , header above, is a mystery, to me?!
[ADDED]
Are those WU's validated by 'a script or by hand'?