Message boards :
Number crunching :
my benchmarks dropped when switching from 32 to 64 bit windows 7
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
William Kendrick Send message Joined: 25 Dec 08 Posts: 46 Credit: 180,614 RAC: 0 |
The BOINC core client is what does the benchmarks, and it does come in both 32 and 64 bit Windows versions. If you're still running the 32 bit version, the benchmarks have probably been affected by running under WOW64. OTOH, if you did a 64 bit BOINC install after switching to 64 bit Windows, the benchmarks are different at least partially because it's a different compile of the sources.Joe thank you joe. boincmagr.exe does not have a *32 bit behind it, so i have to assume it is indeed 64 bit boinc/seti. every one of you who commented, thank you very much. i apologize for my inexperience, and what a great community! some of the bigger crunchers even came in here to chime in... I REALLY APPRECIATE THAT :) here's what i have surmised from this: the 64 bit instruction set is bigger and uses more ram(DUH); i only have 4 gigs of ram. my cpu is overclocked to 3.8@1.5 volts. i only have a 212 core gtx 260 dualfrozr ocv2, and by upgrading it seems to have opened it up a bit. overall, i have found 64 bit windows to be desireable, especially from a mixed setup such as mine, where i do encoding, video editing, and video gaming, on the same rig, so if you're looking into hopping over, dont hesitate! except the price, of corse ;) my whetstone lowered because of the increased instruction set; thats what i think anyway, and my dhrystone increased almost 2000 because of the same instruction set; opening a bottleneck that previously had been on it on a 32 bit windows 7. this is all conjecture of course; i'm hypothesizing, at best, and at worst, wildly guessing. perhaps someone has more insight, perhaps not, either way THANK YOU GUYS FOR HELPING; its nice to belong to a polite group of individuals, i always get stuck with the trolls, lol any other tips info or hypotheses is greatly appreciated. i have been humbled by those of you in the know thanks again william kendrick |
Saaby900T Send message Joined: 24 Dec 10 Posts: 76 Credit: 4,971,171 RAC: 0 |
get some more RAM and see what the numbers are. I highly doubt it will change but ehh worth a shot. |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
no no, i left it there for posterity, i did see your post... i just thought i'd be funny to respond to it lol But you called -BeNt- a troll, a term that I find is tossed around a little too much these days, and is considered to be inflamatory. I'm not certain that very many people here on these forums could spot a real internet troll type personality. It seems we label anyone we don't like or says things we disagree with to be trolls, but there's a major difference between someone having a different opinion and a troll looking to get a rise out of people. |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
BOINC/seti doesn't use that much ram. seti uses about 32 Mb/core. GPU WU's use about 100Mb/WU 4Gb of ram is more than adequate for most people. In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Here are the results of running the BOINC benchmarks. All 4 versions were run on the same machine. I ran the benchmarks 5 times with consistent results between each run. Each run I saw a change of less than 100 for each value. The x86 version of BOINC generated consistent results in Windows 7 x64 & Windows 7 x86. From the numbers you can see the different versions & platforms of BOINC give different results for the benchmark. Starting BOINC client version 6.10.43 for windows_x86_64 To run the benchmarks I have program folder for each version of BOINC in folders like this: D:\BOINC Apps\Boinc 6.10.43 exes\x64\ D:\BOINC Apps\Boinc 6.10.43 exes\x86\ D:\BOINC Apps\Boinc 6.10.58 exes\x64\ D:\BOINC Apps\Boinc 6.10.58 exes\x86\ I ran them separately without being attached to a project to avoid any speculation of any project influencing the results. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
-BeNt- Send message Joined: 17 Oct 99 Posts: 1234 Credit: 10,116,112 RAC: 0 |
Interesting, so basically, as far as your computer is concerned anyways, there is no negligible difference in reporting from the 32 or 64bit versions between the two latest boinc clients. Guess it just must be 'one of those things', I would like to know what caused his obvious drop in numbers though. Traveling through space at ~67,000mph! |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Interesting, so basically, as far as your computer is concerned anyways, there is no negligible difference in reporting from the 32 or 64bit versions between the two latest boinc clients. Guess it just must be 'one of those things', I would like to know what caused his obvious drop in numbers though. MIPS/Whetstone is consistently lower with x64, but not very much on the CPU I was using. The item that caused the lower benchmark does seem to be just the BOINC app. Which is no biggie. It is just a rough benchmark to guesstimate how much work the computer can do. EDIT: Also these benchmarks were done with SP1. I forgot to include that before. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
William Kendrick Send message Joined: 25 Dec 08 Posts: 46 Credit: 180,614 RAC: 0 |
no no, i left it there for posterity, i did see your post... i just thought i'd be funny to respond to it lol i think your eyes should re read what i said. and, since youw brought it up, i said that he'd make a good one... i never said he was one... love how you twisted my words... too bad its posted above us for posterity, and inflammatory comments like yours are against the rules, too. i was commenting on his thoroughness. often, the best trolls are the ones who are thorough, thats all. dont read too far into comments i post, they are made lightly. |
William Kendrick Send message Joined: 25 Dec 08 Posts: 46 Credit: 180,614 RAC: 0 |
Here are the results of running the BOINC benchmarks. All 4 versions were run on the same machine. I ran the benchmarks 5 times with consistent results between each run. Each run I saw a change of less than 100 for each value. The x86 version of BOINC generated consistent results in Windows 7 x64 & Windows 7 x86. each and every time its ran on a x64, the dhrystone is significantly higher. same as i found.. i ran it on two different machines. both times the numbers were consistant with those i posted earlier... i dont get it... |
-BeNt- Send message Joined: 17 Oct 99 Posts: 1234 Credit: 10,116,112 RAC: 0 |
It's not the way people read it though. I decided to semi-ignore it though, glad I wasn't the only one with either train of thought. Traveling through space at ~67,000mph! |
William Kendrick Send message Joined: 25 Dec 08 Posts: 46 Credit: 180,614 RAC: 0 |
i totally understand. but no worries. the statement was meant, in a weird way, to be a compliment to your thoroughness, nothing more. hope i cleared that up. |
William Kendrick Send message Joined: 25 Dec 08 Posts: 46 Credit: 180,614 RAC: 0 |
i really am bad with words. i told my doc one time i was antisocial, and he laughed at me and told me i was a social 'noob'(my word, not his), but that antisocial behavior typically is associated with psychotic tendencies, and that if was truly anti social i'd have to be locked up, lol i'm sorry i keep coming off like a dumb@$$, but i'm not doing it on purpose; i literally am bad with words, saying and acting in whatever fashion i want. i am changing this, thanks for being patient with me :) so what i gather is that i had an unoptimized version of seti 64 bit, and my dhrystone was showing proper numbers, while my whetstone was showing a bit of a decrease. this is now corrected, and my whetstone jumped from 3200 to 3800 or so, equal with my cpu clock speed. i have to assume it was a mixture of what everyone here has told me to begin with, as well as me... reading too much into nothing. |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
I will only state that I did not twist any of your words, I simply read them as I saw them. And I will also say that my comments were in no way "inflamatory", unless speaking up on a untouched subject is now considered inflamatory. |
-BeNt- Send message Joined: 17 Oct 99 Posts: 1234 Credit: 10,116,112 RAC: 0 |
There is no issue, lets not make one please. ;) I understand what he meant, it was aimed at me if you want to call it that. No need to expound on what has already been discussed. I took it the same way you did but expected he meant his best intentions. Which according to him he did, so there is no issue. Back to the issue, remember your whetstone won't match your cpu speed. The whetstone is the measure of your floating point performance. The Dhrystone shows your integer performance alone, both measure in MIPS(Million Instructions Per Second). For instance my q9650 is a quad core running at 3GHz. Floating Point = 4168.57 MIPS Integer Speed = 8488.96 MIPS e8400 running at 3Ghz Floating Point = 3048.57 MIPS Integer Speed = 4434.72 MIPS I'm not entirely sure what effects that number besides making the processor faster. However the only difference, besides the obvious 2 extra cores, between these two processors is the cache size, and the speed of the ram in the system 1066 vs 800, which I'm sure may have a little to do with it. This is a link to the Boinc Wiki that explains and has some addition linked material that explains what each part measures and why. And how it matters to your 'score' etc. Would be interesting to see how different changes in a system effect it. I know when I OC my system to 4GHz the floating point goes up a good bit but the integer speed stays about the same on the quad. Never tested it on the dual core to know. Traveling through space at ~67,000mph! |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Here is a good example to show how the benchmark doesn't really matter a great deal. Compared to Bent's E8400 mine spit out about twice on the integer, but I'm sure they would both crunch the same work at the same rate. Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz [Family 6 Model 23 Stepping 6] Benchmark results: Number of CPUs: 2 3062 floating point MIPS (Whetstone) per CPU 8872 integer MIPS (Dhrystone) per CPU SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
William Kendrick Send message Joined: 25 Dec 08 Posts: 46 Credit: 180,614 RAC: 0 |
There is no issue, lets not make one please. ;) I understand what he meant, it was aimed at me if you want to call it that. No need to expound on what has already been discussed. I took it the same way you did but expected he meant his best intentions. Which according to him he did, so there is no issue. i'm no longer responding to him, lol. thanks for the info, reading it now. |
William Kendrick Send message Joined: 25 Dec 08 Posts: 46 Credit: 180,614 RAC: 0 |
Here is a good example to show how the benchmark doesn't really matter a great deal. Compared to Bent's E8400 mine spit out about twice on the integer, but I'm sure they would both crunch the same work at the same rate. WOW. enough said i think. hal you've been a big help figuring this all out. thanks man. all the tests you ran for me, really really are appreciated. final benchmark Number of CPUs:6 3822 floating point MIPS (Whetstone) per CPU 10203 integer MIPS (dhrystone) per CPU and i take it to mean that hal has stated that this BM is just a... diagram, the same thing that EVERYONE here has said. thanks hal, bent, and everyone else too :) i'm glad i'm here :) |
-BeNt- Send message Joined: 17 Oct 99 Posts: 1234 Credit: 10,116,112 RAC: 0 |
No problem most here like to help when possible especially when it's an unexplored item like you had. Besides the true test is how much work you are actually getting done! So start crunching! Traveling through space at ~67,000mph! |
©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.