Message boards :
Number crunching :
SSE2
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 9 · Next
Author | Message |
---|---|
nemesis Send message Joined: 12 Oct 99 Posts: 1408 Credit: 35,074,350 RAC: 0 |
darn it, i'm always too late for the alcohol.... thanks for the great work, JD! |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
Hi all, P4 Northwood 2,6 GHz with the new SSE2 app look at my wingman, he runs the standard app. 9,886.92 to 29,973.44 seconds, what a progress for a 72 credit unit regards heinz |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
The SSE2 app makes a huge difference on most of my old machines but on the 2.6 Celeron it has cut 6000 seconds...the slowest got the biggest boost. Thanks everyone that worked on this and Thank you Phud for complaining.... Official Abuser of Boinc Buttons... And no good credit hound! |
Mumps [MM] Send message Joined: 11 Feb 08 Posts: 4454 Credit: 100,893,853 RAC: 30 |
Well, I've grabbed the SSE2 flavor for my middle-aged crunchers and it's doing a bang-up job. I've just passed hiamps in over-all RAC! :-) So that only leaves one really ancient P3 based 4-way cruncher stuck with SSE flavored Chicken soup. Here's to hoping the Whale soup eventually hits the shelves for that venerable ancient too. :-) |
Eirik Send message Joined: 25 Mar 01 Posts: 45 Credit: 2,173,371 RAC: 0 |
I just grabbed SSE2 for this one since it is still crunching its kinda hard for me to tell if the AK made any difference from it. But hopefully it will :-D |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
Well, I've grabbed the SSE2 flavor for my middle-aged crunchers and it's doing a bang-up job. I've just passed hiamps in over-all RAC! :-) Hmmmm, I smell an upgrade comming......Have a great day! Official Abuser of Boinc Buttons... And no good credit hound! |
JDWhale Send message Joined: 6 Apr 99 Posts: 921 Credit: 21,935,817 RAC: 3 |
I just grabbed SSE2 for this one since it is still crunching its kinda hard for me to tell if the AK made any difference from it. I do hope that you meant the SSE2 client for this hostid=3762048. Your Q6600 would be better served running SSSE3x client. Kind regards, JDWhale |
Karsten Vinding Send message Joined: 18 May 99 Posts: 239 Credit: 25,201,931 RAC: 11 |
OK. I have been testing the SSE2 client against the SSE3 client on my AMD Phenom, and my AMD Athlon64-X2, with the knabench tool. I wasn't expecting any speedups, but knabench tells me that the SSE2 version i from 2 - 5,5% faster on the Athlon64, and from -1 - 4% faster on my Phenom. This is a bit of a surprise to me, but I have switched to the SSE2 client on both machines, and will be watching carefully to see if it is indeed faster. |
Eirik Send message Joined: 25 Mar 01 Posts: 45 Credit: 2,173,371 RAC: 0 |
I just grabbed SSE2 for this one since it is still crunching its kinda hard for me to tell if the AK made any difference from it. My bad, i meant the hostid you posted yes. |
Mumps [MM] Send message Joined: 11 Feb 08 Posts: 4454 Credit: 100,893,853 RAC: 30 |
OK. I've noticed something similar, although maybe a bit unrelated as well. I have two crunchers: Xeon Dual-Core/Dual CPU with HT Opteron Dual-Core/Dual CPU Which were previously running the Crunch3r op app. In that configuration, the version on the Opteron was running 7% faster than the version on the Xeon. But since switching over to the AKv8.0, the Xeon has flat out run away from the Opteron. 13% faster and still climbing more rapidly. Just an observation that I felt was worth mentioning. It's most likely primarily driven by the difference in Cache between the processors, but is also somewhat affected by the SSE3 implementations. Regardless of the cause, it seems the AKv8.0 will see a better increase in speed on Intel processors. |
Mumps [MM] Send message Joined: 11 Feb 08 Posts: 4454 Credit: 100,893,853 RAC: 30 |
Well, I've grabbed the SSE2 flavor for my middle-aged crunchers and it's doing a bang-up job. I've just passed hiamps in over-all RAC! :-) Well, it better be a big 'un. 'Cause it looks like this climb won't settle down until well over 16,000 RAC, maybe around 18,000. :-) And in an ever-expanding Arm's race, I may end up having to replace that SSE only host with a Quad-Core sooner than I was contemplating. ;-) (It eats up how much power for a measley 250 RAC?!?!?) Crunch On! |
The Gas Giant Send message Joined: 22 Nov 01 Posts: 1904 Credit: 2,646,654 RAC: 0 |
So my question is..Does anyone yet know if the SSE2 version or the SSE3 version of the AK port faster on a Prescott chip? I'll check it out in a few weeks, but I want my RAC to settle out first. Live long and BOINC! Paul (S@H1 8888) And proud of it! |
_heinz Send message Joined: 25 Feb 05 Posts: 744 Credit: 5,539,270 RAC: 0 |
P4 Northwood 2,66 GHz --------------------- CPU time 15360.40625 stderr out <core_client_version>5.8.11</core_client_version> <![CDATA[ <stderr_txt> Optimized SETI@Home Enhanced application Optimizers: Ben Herndon, Josef Segur, Alex Kan, Simon Zadra Version: Windows SSE2 32-bit based on S@H V5.15 'Noo? No - Ni!' Revision: R-2.4V|xB|FFT:IPP_SSE2|Ben-Joe CPUID: Intel(R) Pentium(R) 4 CPU 2.66GHz Speed: 1 x 2694 MHz Features: MMX SSE SSE2 Work Unit Info WU Credit multi. is: 2.85 WU True angle range: 0.394113 Spikes Pulses Triplets Gaussians Flops 5 0 0 0 22065615942717 </stderr_txt> ]]> Validate state Initial Claimed credit 72.8023185534521 ----------------------------------------------------------- CPU time 10432.66 stderr out <core_client_version>5.8.11</core_client_version> <![CDATA[ <stderr_txt> Windows optimized S@H Enhanced application by Alex Kan Version info: SSE2x (AMD/Intel, Core 2-optimized v8-nographics) V5.13 by Alex Kan SSE2x Win32 Build 44 , Ported by : Jason G, Raistmer, JDWhale CPUID: Intel(R) Pentium(R) 4 CPU 2.66GHz Speed: 1 x 2662 MHz Work Unit Info: ............... Credit multiplier is : 2.85 WU true angle range is : 0.394257 Flopcounter: 22061248049434.602000 Spike count: 5 Pulse count: 1 Triplet count: 0 Gaussian count: 3 called boinc_finish </stderr_txt> ]]> Validate state Valid Claimed credit 72.7879050925926 ------------------------------------------- duration now 67,9% for a 72 credit wu ------------------------------------------- heinz |
JDWhale Send message Joined: 6 Apr 99 Posts: 921 Credit: 21,935,817 RAC: 3 |
I agree, but for testing the new clients I've pulled 2 reserves back into active duty. With outside temps already reaching 30C I'll have to retire both Ginger and MaryAnn very soon... I've identified a couple more SSE2 tweaks I want to try, since MaryAnn is my 2.66GHz Northwood, she will have to stay online a bit longer to run the time trials for evaluation. @Gas Giant - I'll switch Ginger (3.2Ghz Prescott) over to SSE2 clients... I'll do the release client first, but I also want to try two others for comparison. Maybe ~24 hours each, maybe longer if needed for similar ARs. Keep in mind that Jason sprinkles magic dust on his computer and recites ancient incantations while building the releases to give them a little extra power ;-) (So my personal builds would be slightly faster if Jason had build them.) [edit] Wait!!! I already ran those tests on Ginger, about 6 days ago. SSE3 is faster on Prescott. That was before I called MaryAnn back into service.[/edit] Best regards, JDWhale |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
So what does the professor have in it? Too bad they don't show computer names...I bet that would be interesting... Official Abuser of Boinc Buttons... And no good credit hound! |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 65763 Credit: 55,293,173 RAC: 49 |
No comment, It's a 60's TV show for Me. ;) The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
OK. In testing there was around ~8% natural variation in procesing times for the AMD chips with either of the apps running the same test WU. I'd probably attribute that mostly to the state of the cache at any given time, and what context switching is going on, so apparent -1 to 4% difference would be considered 'the same app' on these machines. Full-size WUs may increase the difference one way or another. processing time on the Intel chips doesn't *seem to* exibit the same variation on test WUs , not sure about full-size live WUs though. Either way I'd rely on full size-WUs though, rather than the synthetic shortened test ones, and compare same ARs to figure out which App is really faster on a given machine (if any). The apps may have different performance characteristics at different ARs too. Jason "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
NewtonianRefractor Send message Joined: 19 Sep 04 Posts: 495 Credit: 225,412 RAC: 0 |
OK. I can try to run a test on my Turion64 X2. I currently about 180 Wu analyzed using the SSE3 app, and can switch to SSE2. Here is an Excel Worksheet for what I have so far: http://abrau.durso.googlepages.com/sah_boinc07.xls P.S. I will check in on Friday with a comparison graph... |
Kim Vater Send message Joined: 27 May 99 Posts: 227 Credit: 22,743,307 RAC: 0 |
A quick glance at my two P4 Northwoods and a Pentium-M @1,7Ghz (with AK-V8 SSE2x) shows a performance gain of about 35% over 2.4V ;-) A big thanks to the guys that made it possible :thumbsup: Regards kiva Greetings from Norway Crunch3er & AK-V8 Inside |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
A quick glance at my two P4 Northwoods and a Pentium-M @1,7Ghz (with AK-V8 SSE2x) shows a performance gain of about 35% over 2.4V ;-) Hiya Kim..... Have you noticed that AndyK's stuff has been boinked by the Boinc webpage 'improvements'? "Freedom is just Chaos, with better lighting." Alan Dean Foster |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.