Hyperthreading vs. Non-HT for S5 Work
19 Jun 2006 12:42:47 UTC
Topic 191411
(moderation:
With the albert4.37 akos optimized you got more output disabling HT. With the new S5 application does that still hold true? ( Still have days of 4.37 til I get there) Any comparisons HT enabled/disabled anyone can give please? Would like to see examples of same frequencies done on the same host. Thank-you in advance for your help.
Hyperthreading vs. Non-HT for S5 Work
)
Im seeing a ~25% improvement with HT enabled.
Gone from 81 minute crunch times(HT off) to 123 minute crunch times(HT on).
My understanding was that
)
My understanding was that with HT enabled you, of course, took longer to crunch a single unit, but you still crunched more units/hour. You also got a dual benefit if you happened to be crunching, say, SETI and Einstein at the same time since the data/memory paths were different for each and didn't bump heads with one another.
Those who don’t build must burn. It’s as old as history and juvenile delinquents.
Ray Bradbury - Fahrenheit 451
RE: My understanding was
)
That became not true rather abruptly at one step of akosf's improvements, though it certainly was true of the distributed S4 Einstein ap, and has been true right along for most if not all SETI aps.
Yes indeed, That was strongly true even for the most hyperthreading challenged releases of akosf's aps.