| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Retired Forums
Forum: UD Windows Agent Support [Read Only]
Thread: Processor Speed Oddity |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 14
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I have 2 simillar PCs running WCG agent under Windows.
1st: P4 Prescott 3.0 FSB 800 1GB RAM (DDR 400) ASUS P4GD1 (intel 915) Processor Performance: 173 2nd: Celeron 2.8Ggz FSB 400 1Gb RAM (DDR 266) ASUS P4P800SE (Intel 865PE) Processor Performance: 189 I can't understand why second system shows better performance. Scores above are the same for several weeks. I've also checked P4 processor temperature, and run throttling monitoring soft. All OK. I've also seen in this forum, that different P4 3Ggz processors shows different Processor Performance rating sometimes 172 sometimes around 200 for example: Thanks for RT's effort. From Windflowers Device #01 Device Name...........Device-01 Processor Score.......178 Model......................Pentinum 4 3.2GHz Prescott FSB800 OverClock...............None Other......................Air Cooling Motherboard............ECS Device #02 Device Name...........Device-02 Processor Score.......195 Model......................Pentinum 4 3.0GHz Northwood FSB800 OverClock...............None Other......................Air Cooling Motherboard............MagicPro What's the secret? Why Prescott is so bad? |
||
|
|
depriens
Senior Cruncher The Netherlands Joined: Jul 29, 2005 Post Count: 350 Status: Offline Project Badges: 
|
The benchmark in the UD software runs every time a new workunit starts. So it depends on how much the cpu is used at that very moment. Not too long ago my P4 2.66 scored just al little over 120 when I was stressing my cpu at the same time a new workunit was downloaded. The next time I made sure that the system was idle during the downloading of a new workunit and than my processor scores approx. 180.
----------------------------------------So it just depends on how stressed your system is at the moment a new workunit starts.   |
||
|
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges:
|
I have 2 simillar PCs running WCG agent under Windows. 1st: P4 Prescott 3.0 FSB 800 1GB RAM (DDR 400) ASUS P4GD1 (intel 915) Processor Performance: 173 2nd: Celeron 2.8Ggz FSB 400 1Gb RAM (DDR 266) ASUS P4P800SE (Intel 865PE) Processor Performance: 189 What's the secret? Why Prescott is so bad? It must just be an Intel thing. I was a bit curious about those numbers myself, and what an Intel equivalent to this would be: An AMD64 3400 2.2GHz and an AMD64 4000 2.4GHz here are both scoring 238 and holding their own. (better floating point unit) Robert Taylor has collected many WCG processor scores here: http://www.unofficialworldcommunitygrid.com/ Check several entries, but in glancing at them, I see several for the P4 3.0GHz in the 195-225 range.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 6 times, last edit by retsof at Dec 14, 2005 9:36:19 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
retsof It must just be an Intel thing. I was a bit curious about those numbers myself, and what an Intel equivalent to this would be: An AMD64 3400 2.2GHz and an AMD64 4000 2.4GHz here are both scoring 238 and holding their own. (better floating point unit) Robert Taylor has collected many WCG processor scores here: http://www.unofficialworldcommunitygrid.com/ Check several entries, but in glancing at them, I see several for the P4 3.0GHz in the 195-225 range. Thanks for this link, Retsof And I've found there that Prescott scores are really low: 178 P4 3,2 Prescott Windflowers WCG 187 P4 3,4 Prescott bruce boytler WCG 170 P4 3 Prescott malice4you GridPg14 And it's unusual thing when processor with old design and less frequency outperfom newer one. All other benchmarks shows significant superiority of P4 Prescott. Looks like WCG agent has a bug in it's benchmark code. I'll check the time needed to process the same workunit on that both systems(yes, it's possible). I'm sure that P4 will finish it much faster. to depriens: depriens The benchmark in the UD software runs every time a new workunit starts. So it depends on how much the cpu is used at that very moment. Not too long ago my P4 2.66 scored just al little over 120 when I was stressing my cpu at the same time a new workunit was downloaded. The next time I made sure that the system was idle during the downloading of a new workunit and than my processor scores approx. 180. So it just depends on how stressed your system is at the moment a new workunit starts. I'm new to this forum and do not know how to add stats to posting signature, but I'm not new to crunching (my current rank is 11). I've mentioned that scores are about the same for weeks. And there are no processor intensive tasks there. |
||
|
|
depriens
Senior Cruncher The Netherlands Joined: Jul 29, 2005 Post Count: 350 Status: Offline Project Badges:
|
depriens The benchmark in the UD software runs every time a new workunit starts. So it depends on how much the cpu is used at that very moment. Not too long ago my P4 2.66 scored just al little over 120 when I was stressing my cpu at the same time a new workunit was downloaded. The next time I made sure that the system was idle during the downloading of a new workunit and than my processor scores approx. 180. So it just depends on how stressed your system is at the moment a new workunit starts. I'm new to this forum and do not know how to add stats to posting signature, but I'm not new to crunching (my current rank is 11). I've mentioned that scores are about the same for weeks. And there are no processor intensive tasks there. OK, I'm sorry. I missed the 'several weeks' sentence.  Rank 11, that's quite impressive! Keep up the good work! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
And it's unusual thing when processor with old design and less frequency outperfom newer one. Keeping the machine quiet when sending in workunits is good advice when checking a processor benchmark. All other benchmarks shows significant superiority of P4 Prescott. Looks like WCG agent has a bug in it's benchmark code. I'll check the time needed to process the same workunit on that both systems(yes, it's possible). I'm sure that P4 will finish it much faster. The benchmark code may be from the original grid.org days around 2000 since it is part of the UD interface. It has to guess for newer machines. That's not a high priority for update. They'd rather get boinc, new projects, etc. etc. working. to depriens If you are not on a team, go to the team area and join one or find one. There's no statistics penalty, since both member numbers and team numbers come from the same runtime. Any of them could use the help. If you have any team questions, ask in the team area, not here because of forum rules.I'm new to this forum and do not know how to add stats to posting signature, but I'm not new to crunching (my current rank is 11). I've mentioned that scores are about the same for weeks. And there are no processor intensive tasks there. It looks as if you found the way. It will be automatically updated each day. Hmmm. I see you have at least 20 computers crunching... excellent  work.[Edit 7 times, last edit by Former Member at Dec 15, 2005 3:28:43 PM] |
||
|
|
Alther
Former World Community Grid Tech United States of America Joined: Sep 30, 2004 Post Count: 414 Status: Offline Project Badges: 
|
The benchmark code may be from the original grid.org days around 2000 since it is part of the UD interface. It has to guess for newer machines. The benchmark is a standard benchmarking test, and is not processor dependent. It measures integer and floating point performance since that is the most important measurement in typcial grid applications. There is no "guessing" involved. What you do see with newer processors is that the Agent doesn't have a nice, pretty description for them (like Intel Pentium 4..). So it just uses the manufacturer ID...but that is just descriptive, it has nothing at all to do with the processor score.
Rick Alther
Former World Community Grid Developer |
||
|
|
depriens
Senior Cruncher The Netherlands Joined: Jul 29, 2005 Post Count: 350 Status: Offline Project Badges:
|
to depriens If you are not on a team, go to the team area and join one or find one. There's no statistics penalty, since both member numbers and team numbers come from the same runtime. Any of them could use the help. If you have any team questions, ask in the team area, not here because of forum rules.I'm not sure if you're asking me, but I am on a team already, but I'm thinking of starting a new team when I have recruited some friends of mine...  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
to depriens If you are not on a team, go to the team area and join one or find one. There's no statistics penalty, since both member numbers and team numbers come from the same runtime. Any of them could use the help. If you have any team questions, ask in the team area, not here because of forum rules.I'm not sure if you're asking me, but I am on a team already, but I'm thinking of starting a new team when I have recruited some friends of mine... |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Alther
The benchmark code may be from the original grid.org days around 2000 since it is part of the UD interface. It has to guess for newer machines. The benchmark is a standard benchmarking test, and is not processor dependent. It measures integer and floating point performance since that is the most important measurement in typcial grid applications. There is no "guessing" involved. What you do see with newer processors is that the Agent doesn't have a nice, pretty description for them (like Intel Pentium 4..). So it just uses the manufacturer ID...but that is just descriptive, it has nothing at all to do with the processor score. Dear Alther, I think you've not understand my question. Celeron 2.8 can't outperform P4 3.0 Prescott all the time. In any benchmark. If UD benchmark shows such result, then it's a bug there. I've run dozen of other benchmarks, and P4 3.0 scored 8-60% more than Celeron 2.8 finally, I've switched CPUs on that machines and got the same oddity in UD benchmark again. Now I'm running the same workunit on 2 machines (cloned via hack, sorry) after 30 min of work time, Celeron 2.8 has finished 4% and P4 Prescott has finished 7% Flange Keeping the machine quiet when sending in workunits is good advice when checking a processor benchmark. Sure Thing. I've specially stopped all user processes and nobody working there during my tests). |
||
|
|
|