Black hole for tasks

th3
th3
Joined: 24 Aug 06
Posts: 208
Credit: 2208434
RAC: 0

RE: People are pushing

Message 74738 in response to message 74732

Quote:
People are pushing engineering sample penryns to >4.5gigs on air. I don't have a conroe, so I'm not sure what that'd equate to in terms of runtime per WU.


Impressive. I checked xtremesystems and one guy there has broken 10 sec SuperPI 1M on aircooling, ~4.6GHz with ~1.8GHz DDR3 at tight timings, and thats on a quad. Maybe its not too far fetched with 5GHz for dualcoe Penryns then. And, below 10s with "only" 4.6GHz, that says something about the new radix divider and the bigger cache compared to Conroe, very nice improvement in integer performance. SSE4 can do something in the not too distant future also.
- in addition, forgot that there are some smaller S5R3 WUs, iirc there was some at around 150 credits, could end up somewhere below 3 hours on Penryn with the right cooling. Still its only 10 WUs per day. Lets assume SSE4 will make miracles for E@H when new compilers are out, then maybe Penryns could reach 16 WUs per CPU/day. And there could be smaller WUs coming?

Jim Bailey
Jim Bailey
Joined: 31 Aug 05
Posts: 91
Credit: 1452829
RAC: 0

Here's another "black hole":

Here's another "black hole": comp ID 111883, 7 WU's downloaded today, 2 WU's done in Oct., 396 WU issued to this one box, and it is a p4 @ 1.8Ghz!!!! Hello? Hello?

Brian Silvers
Brian Silvers
Joined: 26 Aug 05
Posts: 772
Credit: 282700
RAC: 0

RE: The only way to avoid

Message 74740 in response to message 74737

Quote:

The only way to avoid that is if your running 5.10.something (I forget the exact build number) which has 221 Auto Abort Redundant Results capability, the project has 221 functionality enabled, and the result had never been started at all.

While I know this is an old message from you, FWIW, the version that supports that functionality is actually 5.8.17. Seeing as how I tried to install 5.10.28 the other day and it blew up on me, I'm remaining back on 5.8.16, which doesn't support it... I'm holding off updating my BOINC client until such time as either LHC implements the aborts (their IR=5, MQ=3), or Einstein ups IR to 3 in response to this kind of inquiry... Fortunately the deadlines are fairly tight here, so unless you have a bunch of no replies or computation errors by wingmen, you won't build up huge amounts of pending (relative size of credit awarded taken into account)...

Brian

Comment viewing options

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