Message boards :
Number crunching :
New Faster BOINC 4.13 client for LINUX available now
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · Next
Author | Message |
---|---|
Ned Slider Send message Joined: 12 Oct 01 Posts: 668 Credit: 4,375,315 RAC: 0 |
> Number of CPUs: 1 > > 461 double precision MIPS (Whetstone) per CPU > 1378 integer MIPS (Dhrystone) per CPU > > Number of CPUs: 1 > 941 double precision MIPS (Whetstone) per CPU > 1894 integer MIPS (Dhrystone) per CPU > > Ned these are the before and after benchmarks.... > > A nice improvement Ned....I'm running FC2, > Intel 1.7Ghz processor, 256RAM. > > Ned this might be a stupid question, obvisiously > you're much more computer savy than me, just an end user > here hooked on Linux, do these numbers indicate to you > a significant improvement? I know just looking at them > they are, but can you give me insight as to what I can > expect performance wise? > > Thanks for the optimized client....btw, using i686 arch. > > Best Regards, > > Joe > Nice job Joe :) I'm afraid I don't have any data for Intel chips (only AMD) so it's difficult for me to accurately say how your figures compare, but it certainly looks like you have a good improvement over the standard downloaded client. What processor do you have - I would guess it is a PIII from the speed. If so, try using the P3 client rather than the generic i686 as it will be even faster. The i686 is actually the slowest of the optimisations but will run on any i686 compatible hardware (ie a Pentium Pro or Pentium 2 onwards). To get a more accurate judge of your performance, try trawling through other people's results pages until you find a similar speed pentium chip to yours and your scores should be about the same or slightly better than someone with the same chip using windows 2000 or XP. If your benchmarks are better than an equivelant windows box then you're doing well. Also, have a browse at other people's benchmark scores before and after in the thread below for any with similar numbers to yours: http://forums.pcper.com/showthread.php?t=354308 Ned *** My Guide to Compiling Optimised BOINC and SETI Clients *** *** Download Optimised BOINC and SETI Clients for Linux Here *** |
Roberto Virga Send message Joined: 29 May 99 Posts: 32 Credit: 41,362 RAC: 0 |
> What processor do you have - I would guess it is a PIII from the speed. Doubtful. AFAIK, PIIIs never went higher than 1.4 GHz. At 1.7 GHz the only two possibilities are either an old P4, or a P-M. The benchmark numbers look too low for a P-M, so it's definitely a P4. - Roberto |
Ned Slider Send message Joined: 12 Oct 01 Posts: 668 Credit: 4,375,315 RAC: 0 |
> > What processor do you have - I would guess it is a PIII from the speed. > > Doubtful. AFAIK, PIIIs never went higher than 1.4 GHz. At 1.7 GHz the only two > possibilities are either an old P4, or a P-M. The benchmark numbers look too > low for a P-M, so it's definitely a P4. > > - Roberto > > Thanks Roberto, You're almost certainly more knowledgable than me when it comes to Intel processors, so in that case Joe should try the P4 client :) If you do use an incompatible client (not for the correct architecture) it should just give a seg fault and fail to even run the benchmarks. Just throw it away and try another. Ned *** My Guide to Compiling Optimised BOINC and SETI Clients *** *** Download Optimised BOINC and SETI Clients for Linux Here *** |
mahdia Send message Joined: 3 Apr 99 Posts: 12 Credit: 1,776,582 RAC: 0 |
Great Work, Ned ! I increase the perfomance of my old P3 (1GHz) from 521/1095 to 906/1387 using -march=pentium3 . Especially the build-your-own-boinc-client-guide helped a lot. Greetings from Berlin, Frank |
Ned Slider Send message Joined: 12 Oct 01 Posts: 668 Credit: 4,375,315 RAC: 0 |
> Great Work, Ned ! > > I increase the perfomance of my old P3 (1GHz) from 521/1095 to 906/1387 using > -march=pentium3 . > > Especially the build-your-own-boinc-client-guide helped a lot. > > Greetings from Berlin, Frank > > > Try my P3 client too - it might be even faster. Ned *** My Guide to Compiling Optimised BOINC and SETI Clients *** *** Download Optimised BOINC and SETI Clients for Linux Here *** |
Darth Dogbytes™ Send message Joined: 30 Jul 03 Posts: 7512 Credit: 2,021,148 RAC: 0 |
It would be nice if there was an optimized or 64bit version for Darwin. If there is I missed it. Account frozen... |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
> It would be nice if there was an optimized or 64bit version for Darwin. If > there is I missed it. There is one made by TeamNN ... When I tried it, I did not get a significant increase in speed. Other participants have/had noted improved performance with those builds. I suppose I need to start shopping around and make a list of the places that people can get improved / optimized programs. |
Ned Slider Send message Joined: 12 Oct 01 Posts: 668 Credit: 4,375,315 RAC: 0 |
> > It would be nice if there was an optimized or 64bit version for Darwin. > If > > there is I missed it. > > There is one made by TeamNN ... > > When I tried it, I did not get a significant increase in speed. Other > participants have/had noted improved performance with those builds. > > I suppose I need to start shopping around and make a list of the places that > people can get improved / optimized programs. > Hi Paul, Here you are, I'll start you off on your list for your FAQ. If anyone knows of any more, please feel free to add to the list. I intend to add links to some of these sites to my own site for some of the clients that I don't offer. There's no point in us duplicating each other's efforts :) For Mac OS X and Solaris: http://boinc.berkeley.edu/download_other.php For Linux: http://www.pperry.f2s.com/ http://boinc.us.tt/ Also Mpemba Effect at overclockers UK has this site too: http://www.selman.demon.co.uk/linux/boinc/ SPARC64: Keep an eye on Raithmir's thread here to see where he posts them once finished: http://setiweb.ssl.berkeley.edu/forum_thread.php?id=7210 Update, see here: http://www.kulthea.net/boinc/ And if anyone would like to donate me a copy of MSVC 7 I'll do some optimised Windows clients, but at present I only have MS visual studio 6 and version 7 is needed to compile :) To be honest, I'm really surprised no one has compiled any for windows yet (that I've found at least). I did do a compile using cygwin and the linux souce for Windows that worked, but it wasn't exactly what I'd call optimised! Edit: I see SUN is going to give away Solaris 10 free for x86 (without support) so I might be tempted to do an install and compile some clients for Solaris too at some point. Ned *** My Guide to Compiling Optimised BOINC and SETI Clients *** *** Download Optimised BOINC and SETI Clients for Linux Here *** |
xi3piscium Send message Joined: 17 Aug 99 Posts: 287 Credit: 26,674 RAC: 0 |
|
xi3piscium Send message Joined: 17 Aug 99 Posts: 287 Credit: 26,674 RAC: 0 |
|
xi3piscium Send message Joined: 17 Aug 99 Posts: 287 Credit: 26,674 RAC: 0 |
Ned, 2004-12-16 08:34:42 [---] Benchmark results: 2004-12-16 08:34:42 [---] Number of CPUs: 1 2004-12-16 08:34:42 [---] 939 double precision MIPS (Whetstone) per CPU 2004-12-16 08:34:42 [---] 2284 integer MIPS (Dhrystone) per CPU results with the P3 client, Whetstone went from 941 to 939, but Dhrystone made a big leap !! Thanks for your suggests. Ned Joe |
xi3piscium Send message Joined: 17 Aug 99 Posts: 287 Credit: 26,674 RAC: 0 |
> Ned, > > 2004-12-16 08:34:42 [---] Benchmark results: > 2004-12-16 08:34:42 [---] Number of CPUs: 1 > 2004-12-16 08:34:42 [---] 939 double precision MIPS (Whetstone) per CPU > 2004-12-16 08:34:42 [---] 2284 integer MIPS (Dhrystone) per CPU > > results with the P3 client, Whetstone went from 941 to 939, but > Dhrystone made a big leap !! > > Thanks for your suggests. Ned > > Joe Ned...the cpu is a celeron, is that a P3 class...or should I go back to the i686 arch? |
Benher Send message Joined: 25 Jul 99 Posts: 517 Credit: 465,152 RAC: 0 |
> > > > Joe > Ned...the cpu is a celeron, is that a P3 class...or > should I go back to the i686 arch? > Celeron...oh how shall I name thee? Intel really messed things up calling everyting a Celeron. There are Celeron's using Pentium II circuitry (called a core), ones with Pentium III and yet more using Pentium IV cores. And each pentium also has different cores within each Pentium Class. Anyone confused yet? If not, there are also Celeron D and Celeron M versions ;) It is probably possible to figure out what core the Celeron is using by comparing the Mhz speed rating and the ammount of Level II Cache memory. FSB Mhz L2 Cache Core 66 300-500+ 128 P2 "Mendocino" 100 850-1200 128 P3 "Coppermine" 400 1600-2800 128 P4 "Willamette" 533 2400-3060 256 P4 "Prescott" |
Ned Slider Send message Joined: 12 Oct 01 Posts: 668 Credit: 4,375,315 RAC: 0 |
> > Ned, > > > > 2004-12-16 08:34:42 [---] Benchmark results: > > 2004-12-16 08:34:42 [---] Number of CPUs: 1 > > 2004-12-16 08:34:42 [---] 939 double precision MIPS (Whetstone) per > CPU > > 2004-12-16 08:34:42 [---] 2284 integer MIPS (Dhrystone) per CPU > > > > results with the P3 client, Whetstone went from 941 to 939, but > > Dhrystone made a big leap !! > > > > Thanks for your suggests. Ned > > > > Joe > Ned...the cpu is a celeron, is that a P3 class...or > should I go back to the i686 arch? > Like Ben has illustrated above (and he knows FAR more about Intel processor architecture than I do), it's very difficult for me to tell exactly what processor you have. If it's a PIII or P4 core, then the PIII or P4 client will be the best (fastest), and if it's a PII then the i686 is the client you need. If you're unsure what core processor you have, try them all and see which ones run and which give errors. Normally, when you try and run the benchmarks they either run of fail with a segfault error immediately if it's the wrong processor architucture. I've also seen in the past a little shareware utility that will run from a dos boot floppy called cpuinfo or something like that, that will detect your processor make, model and speed. Try a quick google search and you might find it or something similar. If I can hunt it down maybe I'll put a download on my site to help people correctly identify their processor. Ned *** My Guide to Compiling Optimised BOINC and SETI Clients *** *** Download Optimised BOINC and SETI Clients for Linux Here *** |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
Ned, Thanks ... I started my list ... It is in the FAQ in the Optimize/performance section at the end. Tell me if I can do this part better ... as always, I can use all the help I can get ... :) |
Ned Slider Send message Joined: 12 Oct 01 Posts: 668 Credit: 4,375,315 RAC: 0 |
> Ned, > > Thanks ... I started my list ... It is in the FAQ in the Optimize/performance > section at the end. > > Tell me if I can do this part better ... as always, I can use all the help I > can get ... :) > You're welcome Paul :) I'll try to stop by your site again when I get a spare moment and take a look at what you've added recently. Hope you enjoyed your recent holiday, and it's good to have you back with us :) Regards, Ned *** My Guide to Compiling Optimised BOINC and SETI Clients *** *** Download Optimised BOINC and SETI Clients for Linux Here *** |
xi3piscium Send message Joined: 17 Aug 99 Posts: 287 Credit: 26,674 RAC: 0 |
Ned I was able to run the benchmarks using the P3, but when a wu finished, my linux box crashed. I disable any screensaver, and just go to blank screen in 1 minute, but can not unblank it...so I went back to using the i686 arch, I will try the P4 at Roberto's suggestion. Maybe Roberto can get some idea looking at my benchmarks Something very strange: [yusuf@localhost seti]$ ./boinc_4.13_pentium4-pc-linux-gnu -return_results_immediately 2004-12-18 06:35:53 [---] Starting BOINC client version 4.13 for i686-pc-linux-gnu though I'm running the optimzed core, it still says running the core for i686, is that normal Ned? did the same when I tried the P3 client. 2004-12-18 06:30:36 [---] Benchmark results: 2004-12-18 06:30:36 [---] Number of CPUs: 1 2004-12-18 06:30:36 [---] 917 double precision MIPS (Whetstone) per CPU 2004-12-18 06:30:36 [---] 2161 integer MIPS (Dhrystone) per CPU above benchmarks for the P4 client Am now totaly confused as to whether I'm actually running i686 or P4 |
Ned Slider Send message Joined: 12 Oct 01 Posts: 668 Credit: 4,375,315 RAC: 0 |
> Ned > > I was able to run the benchmarks using the P3, but when a wu finished, my > linux box crashed. I disable any screensaver, and just go to blank screen > in 1 minute, but can not unblank it...so I went back to using the i686 arch, > I will try the P4 at Roberto's suggestion. > Maybe Roberto can get some idea looking at my benchmarks > > Something very strange: > [yusuf@localhost seti]$ ./boinc_4.13_pentium4-pc-linux-gnu > -return_results_immediately > 2004-12-18 06:35:53 [---] Starting BOINC client version 4.13 for > i686-pc-linux-gnu > though I'm running the optimzed core, it still says running the core for > i686, > is that normal Ned? did the same when I tried the P3 client. > 2004-12-18 06:30:36 [---] Benchmark results: > 2004-12-18 06:30:36 [---] Number of CPUs: 1 > 2004-12-18 06:30:36 [---] 917 double precision MIPS (Whetstone) per CPU > 2004-12-18 06:30:36 [---] 2161 integer MIPS (Dhrystone) per CPU > above benchmarks for the P4 client > Am now totaly confused as to whether I'm actually running i686 or P4 > All my clients say "Starting BOINC client version 4.13 for i686-pc-linux-gnu" regardless of which core it's optimised for - that's perfectly normal. I rename the individual binary files for clarity, so you are running the P4 client. Ned *** My Guide to Compiling Optimised BOINC and SETI Clients *** *** Download Optimised BOINC and SETI Clients for Linux Here *** |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
> I'll try to stop by your site again when I get a spare moment and take a look > at what you've added recently. Alas, not much ... I have been ploughing through the things that piled up while I was away. Worse, I was still over-come with travel shock ... It seems to take about a week for me to get over it (sigh) but I hope to be slogging away at it soon ... > Hope you enjoyed your recent holiday, and it's good to have you back with us > :) Yeah, it was neat visiting my daughter. I got to see a historic fishing village, the USS Nautalis, and a few other things ... At least I know what my parameters are, and how I can travel. At least I proved that I can travel, and with the help of many Amtrack employees, not get lost in the process. For all of that, it is good to be back though ... |
xi3piscium Send message Joined: 17 Aug 99 Posts: 287 Credit: 26,674 RAC: 0 |
Strange prob. is occuring, I can't seem trace it, when I'm running KBS to monitor boinc, my system is freezing up, I have diasbaled the screensaver just setting it to go to blank screen after 1 min. After some time I am unable to unblank the screen, and have to reboot. At first I thought it was the optimised clients I was using, but this time I was running the core client from S@H website. Has anyone else experienced the same, is there a batch of wu's that could be causing the prob.? Here is a my cpu info : [yusuf@localhost yusuf]$ cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 15 model : 1 model name : Intel(R) Celeron(R) CPU 1.70GHz stepping : 3 cpu MHz : 1699.897 cache size : 128 KB fdiv_bug : no hlt_bug : no f00f_bug : no coma_bug : no fpu : yes fpu_exception : yes cpuid level : 2 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm bogomips : 3366.91 Not sure which of Ned's clients to run, have tried them all, but now I am convinced it's not the core client. I am running KBS under KDE 3.2, maybe that has something to do with it, because I downloaded KBS for FC3, but am running FC2, also not sure how to completely uninstall KBS, to search for an earlier release. I will go back to running the optimised boinc core for P3 or P4 arch, without running KBS and see what happens. In the mean time any suggests. or advice would be greatly appreciated. Regards, Joe From China: Merry Christmas and Happy Crunching New Year greetings to all. Sorry for the long post... |
©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.