I have an Athlon 64 3000 winchester running at 2.93 ghz. Stable and 50C load. The server keeps sending me work units in the 8 hour range. Is this to be expected? I have a suspicion that the benchmark program needs a serious update to include the 64 winnies. Any other winnie drivers care to comment? Got power? God bless PG&E.
> I see most of your systems are running under Linux. Linux does Einstein VERY
> poorly. Check out this thread as a way to possibly boost your throughput:
>
> http://einsteinathome.org/node/188222
Thanks for the advice.
Through my lurking around here last weekend, i've already learned of the performance penalties inherent with einstein's Linux client. One can only hope they either open the source up or improve the bottlenecks.
As for the zero credit thing, i've not encountered that at all. All uploads have been successful and i've received credit for them up to this point (803.39 since starting on Saturday with 120.89 pending). The only thing I will not do is resort to running Windows. Not a zealot, mind you. I just find Windows to be a tad too limited environment-wise. Besides, years of using vi have rendered me totally useless in VisualStudio :D
If I get motivated enough, I may do the wine trick. I'm not overly concerned though. I imagine a better einstein client for Linux will emerge eventually.
AMD Athlon64 3400+ running Windows XP Professional SP2
10/03/2005 20:55:48 Running CPU benchmarks
10/03/2005 20:55:48 Suspending computation and network activity - running CPU benchmarks
10/03/2005 20:56:47 Benchmark results:
10/03/2005 20:56:47 Number of CPUs: 1
10/03/2005 20:56:47 2108 double precision MIPS (Whetstone) per CPU
10/03/2005 20:56:47 3908 integer MIPS (Dhrystone) per CPU
10/03/2005 20:56:47 Finished CPU benchmarks
This machine completes an Einstein WU in roughly 5 hours and 30 minutes.
Northwood 3.2 @ 3.7GHz, Einstein runs ~9.5 hours
1 GB ram, XP Pro SP2
Measured floating point speed 1595.02 million ops/sec
Measured integer speed 2434.84 million ops/sec
those benchmarks are for each virtual processor.
To shed some light on the Pentium 4 design.
The main reason Intel introduced Hyperthreading Technology was to try and help correct the problem of miscalculations. If I remember correctly the 130nm core (Northwood) has a 21 staged pipeline, and the 90nm (Prescot) has a 31 staged pipeline. If a miscalculation occurs at the end of the pipeline, the cpu has to clear the pipeline and restart the calculations of for the command over again. This is where Hyperthreading improves performance in that it has another command to be calculated ready to go down the pipeline, thus minimzing the ill effects of miscalculation. AMD does not have this problem because their processors are very short pipelined cpus and do not suffer as much from miscalculations. In fact one of the reasons Intel's Prescot cpus runs so hot is because of the longer pipeline. Anandtech has a good article about the Prescot and the major flaws in the design. http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2343
Lets see:
P4 3.2 Prescott HT OC'd to only 3.38
Asus P4C Deluxe E, 1 gig Corsair DDR
Seti@ 2:45 per WU per Thread
Einstein@ 7:50 per WU per Thread
Set to change apps every 60 mins.
So, I am getting about 500 credits per day average for Seti and a little over 100 credits per day average for Einstein. (But I only started this week, credits should go up to 200 or so, I think)
I'm thinking about using my P4b 2.66 for Einstein as well, should be about 9 hrs per I think.
> > Are the semprons so good or are the P4`s so bad ?
> >
> > now that I oc`d the 2200+ to 1800mhz, my time is 6:34 hrs per WU for
> einstein
> > and 3.75 for S@H
>
> Well, don't underestimate the P4 HT:
>
> 3.75h. for S@H means you are doing some 6.4WU's a day.
>
> With my P4 HT , I'm doing real time some 10 to 11 WU's a day ;)
>
>
Hi,
I just started running the client this morning arond 5:00AM local
I have two machines, but only one runing it full time.
Benchmark results
2257 double precision MIPS (Whetstone) per CPU
4189 Interger MIPS (Dhrystone) per CPU
1st WU was 05:28:48
2nd WU was 05:20:19
I have an Athlon 64 3000
)
I have an Athlon 64 3000 winchester running at 2.93 ghz. Stable and 50C load. The server keeps sending me work units in the 8 hour range. Is this to be expected? I have a suspicion that the benchmark program needs a serious update to include the 64 winnies. Any other winnie drivers care to comment? Got power? God bless PG&E.
> I see most of your systems
)
> I see most of your systems are running under Linux. Linux does Einstein VERY
> poorly. Check out this thread as a way to possibly boost your throughput:
>
> http://einsteinathome.org/node/188222
Thanks for the advice.
Through my lurking around here last weekend, i've already learned of the performance penalties inherent with einstein's Linux client. One can only hope they either open the source up or improve the bottlenecks.
As for the zero credit thing, i've not encountered that at all. All uploads have been successful and i've received credit for them up to this point (803.39 since starting on Saturday with 120.89 pending). The only thing I will not do is resort to running Windows. Not a zealot, mind you. I just find Windows to be a tad too limited environment-wise. Besides, years of using vi have rendered me totally useless in VisualStudio :D
If I get motivated enough, I may do the wine trick. I'm not overly concerned though. I imagine a better einstein client for Linux will emerge eventually.
Prescot MaxPost Working
)
Prescot MaxPost Working 4.200MHz Full Load 24/7 Watercolled
Takes 9 hours more or less for 2 WU'S Einsteins Project
Windows XP Pro, SP2
2005-03-10 21:33:40 [---] Running CPU benchmarks
2005-03-10 21:33:40 [---] Suspending computation and network activity - running CPU benchmarks
2005-03-10 21:34:39 [---] Benchmark results:
2005-03-10 21:34:39 [---] Number of CPUs: 2
2005-03-10 21:34:39 [---] 1844 double precision MIPS (Whetstone) per CPU
2005-03-10 21:34:39 [---] 2172 integer MIPS (Dhrystone) per CPU
2005-03-10 21:34:39 [---] Finished CPU benchmarks
----------------------------
http://www.boincsynergy.com/stats/teams.php?team=1290&project=eah
----------------------------
AMD Athlon64 3400+ running
)
AMD Athlon64 3400+ running Windows XP Professional SP2
10/03/2005 20:55:48 Running CPU benchmarks
10/03/2005 20:55:48 Suspending computation and network activity - running CPU benchmarks
10/03/2005 20:56:47 Benchmark results:
10/03/2005 20:56:47 Number of CPUs: 1
10/03/2005 20:56:47 2108 double precision MIPS (Whetstone) per CPU
10/03/2005 20:56:47 3908 integer MIPS (Dhrystone) per CPU
10/03/2005 20:56:47 Finished CPU benchmarks
This machine completes an Einstein WU in roughly 5 hours and 30 minutes.
Northwood 3.2 @ 3.7GHz,
)
Northwood 3.2 @ 3.7GHz, Einstein runs ~9.5 hours
1 GB ram, XP Pro SP2
Measured floating point speed 1595.02 million ops/sec
Measured integer speed 2434.84 million ops/sec
those benchmarks are for each virtual processor.
To shed some light on the Pentium 4 design.
The main reason Intel introduced Hyperthreading Technology was to try and help correct the problem of miscalculations. If I remember correctly the 130nm core (Northwood) has a 21 staged pipeline, and the 90nm (Prescot) has a 31 staged pipeline. If a miscalculation occurs at the end of the pipeline, the cpu has to clear the pipeline and restart the calculations of for the command over again. This is where Hyperthreading improves performance in that it has another command to be calculated ready to go down the pipeline, thus minimzing the ill effects of miscalculation. AMD does not have this problem because their processors are very short pipelined cpus and do not suffer as much from miscalculations. In fact one of the reasons Intel's Prescot cpus runs so hot is because of the longer pipeline. Anandtech has a good article about the Prescot and the major flaws in the design.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2343
Athlon XP 2000+ (Palomino),
)
Athlon XP 2000+ (Palomino), 768 MB @ MSI K7Master MS6341 (AMD 761)
Win 2000 (all Updates/Servicepacks).
E@H: 8h
S@H: 5,5h
It could be a littler bit faster (about 30 Minutes) if i would not run some additional Software in the background.
#k.
--
"In zweifelhaften F
Lets see: P4 3.2 Prescott HT
)
Lets see:
P4 3.2 Prescott HT OC'd to only 3.38
Asus P4C Deluxe E, 1 gig Corsair DDR
Seti@ 2:45 per WU per Thread
Einstein@ 7:50 per WU per Thread
Set to change apps every 60 mins.
So, I am getting about 500 credits per day average for Seti and a little over 100 credits per day average for Einstein. (But I only started this week, credits should go up to 200 or so, I think)
I'm thinking about using my P4b 2.66 for Einstein as well, should be about 9 hrs per I think.
> > Are the semprons so good or are the P4`s so bad ?
> >
> > now that I oc`d the 2200+ to 1800mhz, my time is 6:34 hrs per WU for
> einstein
> > and 3.75 for S@H
>
> Well, don't underestimate the P4 HT:
>
> 3.75h. for S@H means you are doing some 6.4WU's a day.
>
> With my P4 HT , I'm doing real time some 10 to 11 WU's a day ;)
>
>
http://www.boincstats.com/signature/user_63355.gif
Hi, I just started running
)
Hi,
I just started running the client this morning arond 5:00AM local
I have two machines, but only one runing it full time.
Benchmark results
2257 double precision MIPS (Whetstone) per CPU
4189 Interger MIPS (Dhrystone) per CPU
1st WU was 05:28:48
2nd WU was 05:20:19
The Night Menace, My Einstein@Home Water Cooled PC
I think my WU's will all be
)
I think my WU's will all be in the 5 hour range.
My current stats
The Night Menace, My Einstein@Home Water Cooled PC
Athlon64 3200+ 1GB Corsair
)
Athlon64 3200+
1GB Corsair Value Select
Asus A8N-SLI-deluxe
2*Gainward GF6600GT GOlden sample
--- - 2005-03-17 06:28:33 - 1897 double precision MIPS (Whetstone) per CPU
--- - 2005-03-17 06:28:33 - 5217 integer MIPS (Dhrystone) per CPU
with 32bit windowsXP SP1