Message boards :
Number crunching :
Does anyone have a perfect "Cobblestone" machine?
Message board moderation
Author | Message |
---|---|
Benher Send message Joined: 25 Jul 99 Posts: 517 Credit: 465,152 RAC: 0 |
Cobblestone machine benchmarks would be with both numbers as close to 1000 as possible. FP: 1000 Int: 1000 Anyone got one? Provide a link to its info page. |
Chilean Send message Joined: 6 Apr 03 Posts: 498 Credit: 3,200,504 RAC: 0 |
|
JAF Send message Joined: 9 Aug 00 Posts: 289 Credit: 168,721 RAC: 0 |
> But isnt that the speed of the PC ? thats what i ahve always thought > I think Benher is working on "the average Seti work unit" and maybe a machine that a real even integer/floating point benchmark would somehow help him. I'm just guessing, so maybe he will enlighten us (me). His threads have been quite interesting and it's the sort of thing I like to see in Number Crunching (there's too many topics here that belong in the "questions" forum, and yes, I'm a violator too)! <img src='http://www.boincsynergy.com/images/stats/comb-912.jpg'> |
Hans Dorn Send message Joined: 3 Apr 99 Posts: 2262 Credit: 26,448,570 RAC: 0 |
> Cobblestone machine benchmarks would be with both numbers as close to 1000 as > possible. > > FP: 1000 > Int: 1000 > > Anyone got one? Provide a link to its info page. > Hi Ben! All my (quite different) machines are lower in FP than in INT performance. This might have to do with the the different operand sizes of the integer (4 Byte) and the FP (8 bytes) operands used in the benchmarks. You might try compiling a client with a single precision FP benchmark. Regads Hans |
Hans Dorn Send message Joined: 3 Apr 99 Posts: 2262 Credit: 26,448,570 RAC: 0 |
Hi Ben, after some fiddling around with grep I came across this one: http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=179062 Regards Hans EDIT: Some more, with more results: http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=449428 http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=439768 http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=412509 http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=401437 A 2.4Gig Xeon running on a MP board seems to be what you're looking for |
Everette Dobbins Send message Joined: 13 Jan 00 Posts: 291 Credit: 22,594,655 RAC: 0 |
> > But isnt that the speed of the PC ? thats what i ahve always thought > > > I think Benher is working on "the average Seti work unit" and maybe a machine > that a real even integer/floating point benchmark would somehow help him. > > I'm just guessing, so maybe he will enlighten us (me). His threads have been > quite interesting and it's the sort of thing I like to see in Number Crunching > (there's too many topics here that belong in the "questions" forum, and yes, > I'm a violator too)! > > I dout that he will enlighten you. The cobblstone is becoming such a flop as far as crediting goes no one is going to want to take responsibillity for coming up with the Idea. I hope he come through for you. |
Iztok s52d (and friends) Send message Joined: 12 Jan 01 Posts: 136 Credit: 393,469,375 RAC: 116 |
> Hi Ben, > > after some fiddling around with grep I came across this one: > > http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=179062 > > > Regards Hans > > EDIT: Some more, with more results: > > http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=449428 > http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=439768 > http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=412509 > http://setiweb.ssl.berkeley.edu/show_host_detail.php?hostid=401437 > > A 2.4Gig Xeon running on a MP board seems to be what you're looking for > As I have one: (double 2.4 Xeon, HT on, linux) 2004-10-08 10:38:48 [---] 623 integer MIPS (Dhrystone) per CPU 2004-10-13 10:39:51 [---] 404 double precision MIPS (Whetstone) per CPU 2004-10-13 10:39:51 [---] 619 integer MIPS (Dhrystone) per CPU 2004-10-14 23:16:12 [---] 462 double precision MIPS (Whetstone) per CPU 2004-10-14 23:16:12 [---] 745 integer MIPS (Dhrystone) per CPU 2004-10-18 08:41:58 [---] 1135 double precision MIPS (Whetstone) per CPU 2004-10-18 08:41:58 [---] 656 integer MIPS (Dhrystone) per CPU 2004-10-20 19:13:37 [---] 1353 double precision MIPS (Whetstone) per CPU 2004-10-20 19:13:37 [---] 1156 integer MIPS (Dhrystone) per CPU 2004-10-25 19:14:39 [---] 1300 double precision MIPS (Whetstone) per CPU 2004-10-25 19:14:39 [---] 1111 integer MIPS (Dhrystone) per CPU 2004-10-30 19:15:41 [---] 1252 double precision MIPS (Whetstone) per CPU 2004-10-30 19:15:41 [---] 1369 integer MIPS (Dhrystone) per CPU 2004-11-04 18:16:44 [---] 1424 double precision MIPS (Whetstone) per CPU 2004-11-04 18:16:44 [---] 1431 integer MIPS (Dhrystone) per CPU 2004-11-09 18:17:46 [---] 1317 double precision MIPS (Whetstone) per CPU 2004-11-09 18:17:46 [---] 1310 integer MIPS (Dhrystone) per CPU 2004-11-14 18:18:48 [---] 1264 double precision MIPS (Whetstone) per CPU 2004-11-14 18:18:48 [---] 1232 integer MIPS (Dhrystone) per CPU 2004-11-19 18:19:50 [---] 1290 double precision MIPS (Whetstone) per CPU 2004-11-19 18:19:50 [---] 1622 integer MIPS (Dhrystone) per CPU 2004-11-24 18:20:51 [---] 1401 double precision MIPS (Whetstone) per CPU 2004-11-24 18:20:51 [---] 1595 integer MIPS (Dhrystone) per CPU 2004-11-29 18:21:53 [---] 1433 double precision MIPS (Whetstone) per CPU 2004-11-29 18:21:53 [---] 1588 integer MIPS (Dhrystone) per CPU 2004-12-04 18:22:56 [---] 1452 double precision MIPS (Whetstone) per CPU 2004-12-04 18:22:56 [---] 1277 integer MIPS (Dhrystone) per CPU 2004-12-09 18:23:58 [---] 1322 double precision MIPS (Whetstone) per CPU 2004-12-09 18:23:58 [---] 1209 integer MIPS (Dhrystone) per CPU 2004-12-14 18:25:01 [---] 1321 double precision MIPS (Whetstone) per CPU 2004-12-14 18:25:01 [---] 1131 integer MIPS (Dhrystone) per CPU 2004-12-19 18:26:04 [---] 1279 double precision MIPS (Whetstone) per CPU 2004-12-19 18:26:04 [---] 1342 integer MIPS (Dhrystone) per CPU 2004-12-24 18:27:06 [---] 1327 double precision MIPS (Whetstone) per CPU 2004-12-24 18:27:06 [---] 1246 integer MIPS (Dhrystone) per CPU 2004-12-29 18:28:08 [---] 1296 double precision MIPS (Whetstone) per CPU 2004-12-29 18:28:08 [---] 1163 integer MIPS (Dhrystone) per CPU 2005-01-03 18:29:11 [---] 1422 double precision MIPS (Whetstone) per CPU 2005-01-03 18:29:11 [---] 1288 integer MIPS (Dhrystone) per CPU 2005-01-08 18:30:13 [---] 1296 double precision MIPS (Whetstone) per CPU 2005-01-08 18:30:13 [---] 1196 integer MIPS (Dhrystone) per CPU 2005-01-13 18:31:15 [---] 1423 double precision MIPS (Whetstone) per CPU 2005-01-13 18:31:15 [---] 1365 integer MIPS (Dhrystone) per CPU 2005-01-18 18:32:17 [---] 1357 double precision MIPS (Whetstone) per CPU 2005-01-18 18:32:17 [---] 1308 integer MIPS (Dhrystone) per CPU 2005-01-23 18:33:20 [---] 1211 double precision MIPS (Whetstone) per CPU 2005-01-23 18:33:20 [---] 1228 integer MIPS (Dhrystone) per CPU 2005-01-28 18:34:22 [---] 1311 double precision MIPS (Whetstone) per CPU 2005-01-28 18:34:22 [---] 1418 integer MIPS (Dhrystone) per CPU 2005-01-31 11:52:39 [---] 1373 double precision MIPS (Whetstone) per CPU 2005-01-31 11:52:39 [---] 1630 integer MIPS (Dhrystone) per CPU 2005-02-05 11:53:42 [---] 1414 double precision MIPS (Whetstone) per CPU 2005-02-05 11:53:42 [---] 1648 integer MIPS (Dhrystone) per CPU 2005-02-10 11:54:44 [---] 1420 double precision MIPS (Whetstone) per CPU 2005-02-10 11:54:44 [---] 1563 integer MIPS (Dhrystone) per CPU 2005-02-15 11:55:47 [---] 1119 double precision MIPS (Whetstone) per CPU 2005-02-15 11:55:47 [---] 952 integer MIPS (Dhrystone) per CPU You can see original boinc, slightly optimized boinc, Boinc compiled with Intel C. I guess machine was pretty loaded three days ago to get almost 1000/1000 ;-) Kernel was upgraded a month ago. Frankly, I've been a bit upset by Wintendo boxes out-cobblering me, after considering adding "--mips-factor 1.2" option to multiply everything with 1.2 (sin), my collegue managed to compile local version. To comment: >The cobblstone is becoming such a flop as far as crediting goes no one is going >to want to take responsibillity for coming up with the Idea. It is good idea to do fairly benchmark. If we have to be pedantic, some filtering/averaging can do great. Maybe in version 6.3 ;-) This is proper way to smoth sinners (see above). Simpy by comparing RAC on different boxes I can get feeling on speed and non-boinc load. By checking work done in last few days I already discovered problem on one PC. Boinc is a kind of SMTP as well. Enjoy, Iztok |
©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.