BRP3 Windows CPU App 1.05, opt-out of BRP CPU
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4312
Credit: 250273632
RAC: 35836
10 Jan 2011 16:15:15 UTC
Topic 195568
(moderation:
This is just a renamed 1.04 BRP3 application that now runs under plan class BRP3SSE.
The reason for this release is that it should now be possible to opt-out of the CPU BRP3 App if you wish to only run the CUDA version: Visit your Einstein@Home preferences, edit it for the desired venue and uncheck the Box "Run CPU versions of applications for which GPU versions are available".
This is almost untested yet, feedback welcome!
BM
BM
BRP3 Windows CPU App 1.05, opt-out of BRP CPU
)
Note that with the scheduler update required for this to work we raised the GPU memory requirements for the CUDA application to 300 MB.
BM
BM
Up to now I didn´t find
)
Up to now I didn´t find these check box, whether in the E@H Preferences nor in the BOINC Manager/Extras/Adjustments. May be it requieres more time for the server to change this for each participant.
But since 23:37 UTC last night the predicted crunching time is more realistic and I have so more files in stock as mentioned. Within these 18 BRPS files downloaded since then no non-cuda file was included.
Thanks
Martin
RE: Up to now I didn´t
)
I see "Run CPU versions of applications for which GPU versions are available" in my Einstein@Home preferences. Did you try flushing the browser buffers (ctrl F5 for MSIE)?
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
There's a slight anomaly in
)
There's a slight anomaly in the 'switch' (column view) preferences page, with 'yes' displayed as '0' and 'no' displayed as '1'.
RE: There's a slight
)
Thanks. should be fixed now.
BM
BM
Hallo Gundolf, hallo Richard
)
Hallo Gundolf, hallo Richard !
Thanks for your help. I didn´t see this lowest line, as I awaited this check box a bit upper.
I had, a we say in a free translation from German, I had tomatoos on my eyes.
Kind regards
Martin
Thank you for this, it was
)
Thank you for this, it was getting discouraging watching the CPU pick up a task and crawl on it for 17 hours while knowing full well, the GPU could do it in an hour :x
Now that there is a way to have it NOT assign these tasks to the CPU I will again start using GPU on Einstein.
In closing, again, thank you
Aaron
If god meant for us not to BOINC he'd have made our #$%^%^ shorter!!
Amounts of data releases in
)
Amounts of data releases in one WU for the CPU and CUDA versions of BRP3 is now the same?
And how it relates to the amount of data in ABP2 WUs?
Judging by the records in one of completed BRP3 WUs, it is as before (in ABP2) consists of four pieces of data. But the size of each pieces is much more, like ~35 minutes of observation time vs ~5 minutes?
Hi! Yes, BRP3 units have
)
Hi!
Yes, BRP3 units have longer samples of radio data (what counts is not so much the duration but the number of samples: duration x sampling frequency. The last ABP2 units had 10 sub-units if I remember correctly, the BRP3 units are bundles of 4 sub-units. I think the increase in data content per workunit is reflected by the higher credits per WU.
CU
HB
Yah, you're right. I did not
)
Yah, you're right. I did not pay attention to another sample rate. In the old data was 128 ms / sample, now is 500 ms.
And accordingly the old data packs were 2M samples (central-search data) or 1M samples (anti-center search) each (4 or later 10 data packs per WU).
New packaging of data is 4M samples in each, and 4 such packs in each WU.
Hmm then it turns out my computers process data much more slowly than it did in ABP2 (in terms number of samples / amounts of data per hour, for now not look to credits, although they are also slightly smaller). So the question be is - why?
In the BRP3 algorithms have been changed? Or together with new data (from other telescopes) in the same time search parameter space was increased?
And last question for today - CPU application of BRP3 have only SSE(1) version? (cuz sse2/3/4 not needed for this kind of data?) or my client get wrong version (SSE1 instead of SSE2)