SSE/2 Optimization Guide W/ Pics

[[cTx]Philosophy]

Im at 2.900 with opti 165 and i get 5xxx/16xxx points. Run the benchmark again

im also @ 2.9 and ionly got this
7/6/2006 5:25:56 PM||Running CPU benchmarks
7/6/2006 5:26:55 PM||Benchmark results:
7/6/2006 5:26:55 PM|| Number of CPUs: 2
7/6/2006 5:26:55 PM|| 4955 floating point MIPS (Whetstone) per CPU
7/6/2006 5:26:55 PM|| 11777 integer MIPS (Dhrystone) per CPU
7/6/2006 5:26:55 PM||Finished CPU benchmarks

[@_dud]

are there any mac optimizations?
I have some emac's a a server that could use them.

[David_L6]

I suspended (and "snoozed") the units that were running then exited BOINC. (Did all this starting by clicking on the B on the taskbar - by the clock.)

Downloaded BOINC_5.5.0_SSE2 to my desktop. Unzipped it to C:\program files\BOINC.

Re-started computer and disabled hyperthreading. Ran benchmarks. NO change. Turned off computer. Re-started computer and ran benchmarks. NO change. Re-started computer and enabled hyperthreading. Ran benchmarks. NO change.

CPU is a P4 630 if that matters......

[David_L6]

I think I finally figured it out. Benchmarks improved a lot.

[gelatinousfury]

Holy crap my cpu benchmark scores flew up, very nice!

[largon]

I gather these optimizations don't actually gain any crunching performance because Rosetta simply does not use SSE regarless what client one uses, meaning the optimizations only alter the way credits are granted?

Doesn't this make ranking systems pointless if the optimized clients are credited on different basis?

[ShootStraight]

The long and the short of it is that the benchmark process was/is unfair and didn't accurately reflect the power of a given computer. As such Rosetta devised a new credit scheme that doesn't rely on benchmarks, optimized or not.

As to whether the ranking system is relevant: IMO it is relevant inside a given projects individual scope, however when viewed globally in the aggregate it is entirely irrelevant, as that would imply parity among the separate projects which there really isn't - optimized or no.

-SS

[red]

Why is this thread still sticky if WCG XS is anti optimized?

[ShootStraight]

Why is this thread still sticky if WCG XS is anti optimized?

I reckon this place is pretty much the same as it was left back in July. It would be nice to see this place tidied up a bit - get the graffiti of the walls and such.

-SS

[Movieman]

Why is this thread still sticky if WCG XS is anti optimized?

Maybe because this is the Rosetta section and not the WCG section?
But you have a good point, this thread has served it's purpose long ago.
Thanks for bringing it to our attention.

[red]

Even though Rosetta has its own scoring system I think, I might imagine that people that do multiple projects might stumble upon this and certain projects (like WCG) would punish them.

[mike047]

I believe this optimization is legal to use on the SETI project.

[PyroFire]

why not SSE3 on say an opteron 146