Gravitational units time and BRP3 units time
24 Jan 2011 19:39:47 UTC
Topic 195601
(moderation:
Hello!
In last week my computer receive (mainly) BRP3 units. Server status page confirm, that it does not only my host situation:
[pre]Computing BRP3 S5GC1HF S5GC1 ABP2 in DB
CPU weeks
(from successful tasks last week) 29,100 11,298 16 0 40,414
CPU PFLOP
(from successful tasks last week) 40,925 14,999 16 0 55,940
Granted credit / 100
(from successful tasks last week) 1,242,645 499,044 344 0 1,742,033[/pre]
Why? Thank you!
Gravitational units time and BRP3 units time
)
BM said in other thread, that they are trying to reach a balance between these two projects, that will give the best results. And btw BRP3 works 7 times faster then S51GCHF. That is why we do more BRP3 and less S5GC. It is normal.
I think right now we are at a
)
I think right now we are at a ratio of around 10% GW to 90% BRP. But only Bernd knows...
In the last days only a few 1000 GW-WU finished compaired to ca. 50000-60000 when BRP hassend started yet.
May be that's true. But we
)
May be that's true. But we have only to crunch and wait how it will go. Currently we have all the chances to finish both even before summer begins.
Looking at the numbers, I
)
Looking at the numbers, I found that both runs will finish obviously in the same time. This way it can some day say us "No jobs currently available for your computer".
I guess project developers have some hidden cards on their hands for this situation, do they?
RE: Looking at the numbers,
)
The current ETA for BRP3 is 126 days, vs 60 for S5GC1HF. Unless BRP3 gets another major speedup in the nVidia app, or an ATI app is launched they won't be done anywhere near the same time.
RE: ......or an ATI app is
)
SETI will likely discover ET before that happens.
RE: RE: ......or an ATI
)
Especially now that SETI has a (third-party, Windows) ATI app.
It would be interesting to
)
It would be interesting to see how many on this project could make use of an ATI app.
Count me as 1.
Richard Haselgrove
)
But Raistmer, the developer of the ATI SETI aps at Lunatics, has been quite vocal about limitations of and bugs in the AMD-supplied development environment. Also his two aps won't run usefully on all models of ATI cards.
For example, his current MB ap (r177) won't run on my HD 4550 512MB card at all, and his current AP ap (r505) slows to something between a dead halt and many, many times slower than single instance if double instance is attempted on it.
I have plenty of respect for Raistmer, but these results are far from an argument that scarce project resources would most gainfully be spent on ATI aps for Einstein at the moment.
Notwithstanding my reservations, were such an effort made, of course, I'd happily volunteer my HD 4550 host for beta test, and would certainly add it to my Einstein effort upon release.