Message boards :
Number crunching :
optimized app and quad core
Message board moderation
Author | Message |
---|---|
Steve (Andoh) Anderson Send message Joined: 20 Nov 00 Posts: 13 Credit: 87,650,099 RAC: 0 |
Is there an optimized app that works with a Kentsfield quad core and XP? I've gone through the optimized sticky but could not find a reference that mentions this combination directly. I use one for my core 2 duo's but don't want to hose up my quad machine. Andoh |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Is there an optimized app that works with a Kentsfield quad core and XP? I've gone through the optimized sticky but could not find a reference that mentions this combination directly. I use one for my core 2 duo's but don't want to hose up my quad machine. 32 bit or 64 bit? Do you want graphics (screensaver) or none? "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
Is there an optimized app that works with a Kentsfield quad core and XP? I've gone through the optimized sticky but could not find a reference that mentions this combination directly. I use one for my core 2 duo's but don't want to hose up my quad machine. Doesn't matter how many cores you've got - the acid test is what CPU instructions each/every core can use (they should all be the same, AFAIK). I think Kentsfield is just another of the SSSE3 Core 2 architecture, but check with CPU-Z to reassure yourself. Then use the SSSE3 (or whatever) from the optimiser of your choice. |
Steve (Andoh) Anderson Send message Joined: 20 Nov 00 Posts: 13 Credit: 87,650,099 RAC: 0 |
msattler's, your question's answered mine. Thanks for the explination Richard. It worked. Andoh |
JDWhale Send message Joined: 6 Apr 99 Posts: 921 Credit: 21,935,817 RAC: 3 |
Excuse me, but while on the Core2 Quad subject... On this particular processor, Kentsfield, can anyone attest that 2.4 ssse3 flavor is indeed faster and therefore a tastier treat than 2.4 sse3 running on XP 32bit? There might be differences in performance between the Duo's and Quads because of the extra bandwidth required to feed 4 cores on the same size bus. Also, I realize that with the WU Credit Multiplier pending, Crunch3r's build of 2.4V is poised to become the de facto standard, but for right now XP users have a choice between 2.4 or 2.4V. I am partial to 2.4 because of the cache and memory speed printed in the results. While maybe not terribly accurate, I feel it can be used as an indicator of BIOS configuration issues and is therefore somewhat useful. Sorry for that semi-extraneous setup, now I'll get to my second question: Have any of you an opinion on which among the flavors of 2.4 and 2.4V is faster overall on XP 32bit? Please include which of the Core2 chips you are using when responding so that we can reference with Intel tables regarding L2 sizes, frequency, etc. TIA |
JLDun Send message Joined: 21 Apr 06 Posts: 573 Credit: 196,101 RAC: 0 |
I've run both, myself, (XP-32bit-Celeron 1.4GHz-SSE2-nGFX), and they were about the same for the WU's I was running 'at the time'. |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
Excuse me, but while on the Core2 Quad subject... I ran a test a little while ago of SSE3 vs SSSE3. Results were as shown here: IIRC these were run on the same machine one after the other. Conclusion was that there is little difference in performance overall. As an aside, when I upgraded from a Dual core (E6400) to Quad (Q6600 - Kentsfield) I had to up the o/c from 2.8G to 3.1G to get equivalent performance per core. I assume this was caused by FSB contention but L2 Cache could also have had an impact. F. |
David Send message Joined: 19 May 99 Posts: 411 Credit: 1,426,457 RAC: 0 |
As an aside, when I upgraded from a Dual core (E6400) to Quad (Q6600 - Kentsfield) I had to up the o/c from 2.8G to 3.1G to get equivalent performance per core. I assume this was caused by FSB contention but L2 Cache could also have had an impact. Wow. I have 2 Q6600 & they run well, but my E6300 dual core runs very slow in comparison. For some reason it wont overclock (I think its the motherboard), and only runs 1.5 gig of 667 ram as there is no dual channel on that board. I have seen some E6300/6400 chips that scream along, but mine takes basically 2x as long as the Q6600 did when at stock speed and I thought that was about right given the 2.4Ghz Vs 1.86Ghz. The current run of "25 min" WU's on the 3Ghz Q6600 take roughly 60 to 70 mins on the E6300. All run the Crunch3r's SSSE3 no graphics 2.4V build |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
As an aside, when I upgraded from a Dual core (E6400) to Quad (Q6600 - Kentsfield) I had to up the o/c from 2.8G to 3.1G to get equivalent performance per core. I assume this was caused by FSB contention but L2 Cache could also have had an impact. My comparison was like-for-like. I took the E6400 (2.14GHz) off my Asus board and replaced it with the Q6600 (2.4GHz). I am running Corsair PC2-6400 (400MHz) RAM in dual-channel. F. |
JDWhale Send message Joined: 6 Apr 99 Posts: 921 Credit: 21,935,817 RAC: 3 |
Thanks, Fred W, both for these tests and for all the presentations you make to help us understand some of the stats and bring them into focus. So, if I'm reading this chart and data correctly, it looks like marginal improvement running Crunch3r's old 2.4 ssse3 build rather than the 2.4V ssse3 build. Assuming the app refered to as "new" was the 2.4V app. The newer app seems to make the "sweet" WUs slightly less sweet, but also makes the "sour" WUs slightly less sour. "Sweet" refers to the higher credit/CPU second while "sour" refers to lower credit/CPU sec WUs. I realize that you refered to Angle Range(AR) vs. CPU Time in your presentation, please excuse me for expressing my conclusion based on "current" requested credit running newer BOINC versions crunching those ARs. Boinc OnOn... |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
Interpretation not quite correct, and my apologies for not clarifying the chart. The red and blue columns are the 2.4V SSE3 and SSSE3 apps running on the same hardware. The "new" refers to a different (same spec) host. If you look at the figures relating to the red and blue columns, the mean value at most AR's for either column falls within the max and min values of the other. So, if we could wait for an infinite number of readings, I would expect there to be little or no difference in the mean values plotted. I.e. don't expect any significant improvement if you change from one to the other. F. |
JDWhale Send message Joined: 6 Apr 99 Posts: 921 Credit: 21,935,817 RAC: 3 |
Thanks for clarifying that.... Rather it seems that the "new" host in these tests running the SSSE3 2.4V app reflect my prior "sweet and sour"(*) conclusions when compared to the prior host. (*) Not to be confused with #6 at nearby Chinese takeout, "Sweet and Sour Chicken";-). |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
Ahh!! You too have a nearby Chinese - prefer the King Prawn myself ;) F. |
©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.