Well, I`m running an OC`d sempron 2200+ @ 1710Mhz with 256Mb 2700DDR kingston on a Asrock k7s41gx...

Setispeed : aprox. 4.25 Hrs/WU

Einsteinspeed : aprox. 7.5 Hrs/WU

Lookslike Sempron kicks P4 ass..

Since I`m lookin` for an upgrade, what setup will be a major improvement ? In speed that is.. I was thinking that I`d be better off adding a "cheapo' Sempron 3100+ system, instead of going for 64bit AMD`s or the HT P4`s.. I mean how fast is a the top of the line anyway? I bet none go under the 4 Hr/wu for Einstein..

In that light, I`d get the same speed if I got a cheapo asrock and a new sempron .. that would be a lot cheaper, right?
____________

--- - 2005-02-27 13:15:47 - Suspending computation and network activity - running CPU benchmarks
--- - 2005-02-27 13:16:48 - Benchmark results:
--- - 2005-02-27 13:16:48 - Number of CPUs: 1
--- - 2005-02-27 13:16:48 - 1592 double precision MIPS (Whetstone) per CPU
--- - 2005-02-27 13:16:48 - 3855 integer MIPS (Dhrystone) per CPU
--- - 2005-02-27 13:16:48 - Finished CPU benchmarks


is that any good?
____________

i got an athlon XP 2200+ at 1900 Mhz and 768 MB DDR at 266 Mhz on an Asrock K7VT2 and i takes about 3.5 hours for a Seti@Home and about 7 hours for an Einstein@Home WU.
____________

Just tried to crank it up to 210FSB, but it keeps freezing up just as windows is starting.. damn..

I got it stable at 1800mhz, cpu running a bit hot.. 53-55 C ..Cpuidle keeps it in check, but this costs cycles..

benchmark at 1800hz

--- - 2005-02-27 13:50:47 - Benchmark results:
--- - 2005-02-27 13:50:47 - Number of CPUs: 1
--- - 2005-02-27 13:50:47 - 1679 double precision MIPS (Whetstone) per CPU
--- - 2005-02-27 13:50:47 - 4044 integer MIPS (Dhrystone) per CPU
--- - 2005-02-27 13:50:47 - Finished CPU benchmarks

might even set it a bit lower, I`m running it at the limit it seems..
____________

I guess the first thing to do is get a better cooler.. or get me one of them watercooling kits.. maybe it will hold at 1900mhz..

Right now I`m using the boxed sempron, so not really suitable for high oc`s
____________

HW: P4 2,67@2,7 / 512MB@135MHz / 1384 Whetstones / 4189 Dhrystones
SW: WinXP Pro SP2 / BOINC 4.19 / Einstein 4.79

WU duration: something around 8,75h
____________
Greetings, Santas little helper

P4 HT 2.4@2.88GHz, bus speed 960MHz, 512MB RAM@160MHz. CPU temp 44°C/111°F with room temp around the 20°C/68°F.

Running on Win XP Pro SP2. Boinc v4.19 and Einstein 4.79.

Crunching 2 WU's at the same time. CPU time per WU between 11:10 and 12:20h.

Extremly cheap ASRock AMD K7S41GX

Float: 1525.48
Integer: 3700.26

Einstein: 7:57:29
Seti: 4:11:17

____________
Es grüßt:

AMD Barton XP2400-M @2.1ghz 512k L2
512mb PC3200 ram

E@H: 6 hr
S@H: 2.5-2.75 hr
P@H: 55 min


____________

*I still know CRAP when I see it.

as I suspected... even the cheapest AMD, the sempron 2200+ I have is kicking the butts off most Intel P4 systems.. now how funny is that ?

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
____________

> as I suspected... even the cheapest AMD, the sempron 2200+ I have is kicking
> the butts off most Intel P4 systems.. now how funny is that ?
>
> 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
>

I was always under the assumption that sempron's were worse than an Athlon XP. But my Athlon XP 2400+ takes almost 12 hours for a WU.
____________
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell

> 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 2.4@2.88GHz, I'm doing real time some 10 to 11 WU's a day ;)

____________
Greetings from Belgium
Thierry

> >
>
> I was always under the assumption that sempron's were worse than an Athlon XP.
> But my Athlon XP 2400+ takes almost 12 hours for a WU.
>

Yes, Athlon XP's are normaly faster... and as i said i need 7 hours for a e@h WU with my 2200+... something doesnt seem to be alright with yours
____________

>
> > >
> >
> > I was always under the assumption that sempron's were worse than an
> Athlon XP.
> > But my Athlon XP 2400+ takes almost 12 hours for a WU.
> >
>
> Yes, Athlon XP's are normaly faster... and as i said i need 7 hours for a e@h
> WU with my 2200+... something doesnt seem to be alright with yours
>

Hmmm, well I have several 2400+ machines, and they are all around this speed. I wonder what it could be?
____________
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell

> I was always under the assumption that sempron's were worse than an Athlon XP.
> But my Athlon XP 2400+ takes almost 12 hours for a WU.

That's because you are running Linux. The Einstein client performs very badly in Linux. In SETI, there is little, if any, difference between Windows and Linux.



____________
Be lucky,

Neil

I'm running a XP3200, Asus A7N8X delux MB, 1.5 gig Corsair DDR 400 ram, nothing overclocked. My Einstein times are 5:50 to 6:00 hours, seti Boinc 2:30, Mfold 1.25 55 minutes average.
____________

AMD64 3200+ with Win ME: SETI 2.5 to 3 hrs, Einstein 4 to 5 hrs, Predictor 40 to 50 mins. LHC 4 seconds to 2 hrs.
____________

> > I was always under the assumption that sempron's were worse than an
> Athlon XP.
> > But my Athlon XP 2400+ takes almost 12 hours for a WU.
>
> That's because you are running Linux. The Einstein client performs very badly
> in Linux. In SETI, there is little, if any, difference between Windows and
> Linux.


Why is this? You would think in Linux they could make it even better than for Windows.
____________
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell

okay...

after several crashes I have settled for 1710Mhz with a 190 FSB and a Vcore of 1.67 Volts..

I cranked it up to 1900 Mhz for a few minutes at 210fsb.. it crashed.. at 60C it shut down

I stand corrected for the Einstein WU time.. it takes me 7hrs36mins to complete.. But that`s at 1710mhz, once my watercooler is hooked up, I can run stable at 1850mhz or so, that would amount to aprox. 30mins speedup..


Basicly, all indications are that AMD beats the Intel P4 at any test..

I`ll never buy intel again..
____________

> Why is this? You would think in Linux they could make it even better than for
> Windows.

Compiling for Linux is a lot more complicated than compiling for Windows because of the far more diverse range of platforms the compiler can target for. They simply haven't found the "right" compiler optimisations yet. Actually, it's even more complicated than that because the libraries linked (either dynamically or statically) need to be optimised too to have any real effect.

It's a shame the source code has not been released - otherwise some of the people out here with more time to spare may have cracked it by now. ;)


____________
Be lucky,

Neil

> Compiling for Linux is a lot more complicated than compiling for Windows
> because of the far more diverse range of platforms the compiler can target
> for. They simply haven't found the "right" compiler optimisations yet.
> Actually, it's even more complicated than that because the libraries linked
> (either dynamically or statically) need to be optimised too to have any real
> effect.

This still seems odd though. Because if the code is well-written, it should perform roughly the same as it does on windows, since both programs are basically spending most of their time doing Floating Point calculations, which is a CPU problem not a compiler problem.

I do agree they should release the source code. I think their arguments for not releasing it are kind of weak, so what if someone optimizes their client to get more points? Big deal, we are here to find gravity waves, not be the person with the most points!
____________
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell

> This still seems odd though. Because if the code is well-written, it should
> perform roughly the same as it does on windows, since both programs are
> basically spending most of their time doing Floating Point calculations, which
> is a CPU problem not a compiler problem.

Well I did find some optimized BOINC clients at the following, via another thread:
http://boinc.us.tt/
http://www.pperry.f2s.com/downloads.htm

Using these new clients, almost doubled all my benchmarks(notice my P4 3.06GHz HT matches very closely with my AMD Athlon 1.2 GHz)

*Intel P4 3.06 GHz HT
567.51 million ops/sec
1331.32 million ops/sec

1512.76 million ops/sec
2344.25 million ops/sec

*Intel P4 2.0 GHz
543.37 million ops/sec
1565.7 million ops/sec

1096.71 million ops/sec
2570 million ops/sec

*Intel P4 1.5 GHz
407.42 million ops/sec
1158.59 million ops/sec

829.67 million ops/sec
1915.24 million ops/sec

*Intel Celeron 2.70 GHz
705.88 million ops/sec
2246.1 million ops/sec

1551.89 million ops/sec
3232.92 million ops/sec

*AMD XP 2400+
1069 million ops/sec
2462.66 million ops/sec

2108.88 million ops/sec
4195.66 million ops/sec

*AMD XP 2000+
891.09 million ops/sec
2050.3 million ops/sec

1754.39 million ops/sec
3491.33 million ops/sec

*AMD Athlon 1.3 GHz
694.1 million ops/sec
1541.09 million ops/sec

1354.62 million ops/sec
2706.22 million ops/sec

*AMD Athlon 1.2 GHz
608.25 million ops/sec
1209.16 million ops/sec

1242.12 million ops/sec
2480.67 million ops/sec


The only thing is, I doubt the optimized boinc client has much effect on the actually einstein stuff, since that is a separate program. But it just shows, an optimized einstein client for linux would greatly improve results more than likely.
____________
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell

>
> This still seems odd though. Because if the code is well-written, it should
> perform roughly the same as it does on windows, since both programs are
> basically spending most of their time doing Floating Point calculations, which
> is a CPU problem not a compiler problem.
>
I checked the GCC compiler manual about this; There a several compiler flags available for FPC, with no flags set you will loose speed for sure.


____________
John,

> I checked the GCC compiler manual about this; There a several compiler flags
> available for FPC, with no flags set you will loose speed for sure.

Well when you compile with gcc you would think it would optimize as much as it can for the 686 arch. I realize none of us run that arch anymore, but at the same time, and therefore using the athlon-xp or pentium4 arch would give better FP results.

The only thing is though, for windows to work on different archs, there is no way it can be optimized for an athlon-xp or penitum4 either.

So I wonder what is really going on.
____________
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell

> The only thing is though, for windows to work on different archs, there is no
> way it can be optimized for an athlon-xp or penitum4 either.

I have no indept knowledge about windows/intel but could imagine a mechanisme that at installation time code and Libs are just better selected and matched to the specific arch then is done with "general" linux distro's as most of the people run.
Further more the tight relation between Microsoft and Intel might have given them an advantage anyhow ??

Any specialist out here ??



____________
John,

Running at my default speed for P4 2.8 (Prescott) and HT running,
I complete:

seti unit in about 3 hr 10 min
Einstein in about 12-13 hr
ProteinProdictor in about 2 hr
small LHC unit in abour an hr 10 min (w/ full # of turns)
large LHC unit in about 10-11 hr (w/ full # of turns)
and Climate, well, 4-5 weeks :Þ

With HT off, Seti takes about 1 hr 50 min, or 3 hr 40 min for 2 units
I have not tried disabling HT and running the others.

____________

> This still seems odd though. Because if the code is well-written, it should
> perform roughly the same as it does on windows, since both programs are
> basically spending most of their time doing Floating Point calculations, which
> is a CPU problem not a compiler problem.

Even the best-written code will turn into garbage if compiled with a poor compiler. Most of the calculations will be handled by code in the maths libraries so optimising those would provide the greatest improvement.


____________
Be lucky,

Neil

HT P4 3.2GHz, 1Gig RAM, WinXP - SP2, BOINC 4.24 :

-Whetstone 1359
-Drystone 1720

E@H: v4.79, 11 WU's, avg 12.3 hours
LHC: v4.64, 9 WU's, avg 1.5 hours
PP@H: v4.22, 74 WU's, avg 1.9 hours
S@H: v4.09, 31 WU's, avg 2.9 hours

Extra notes:
E@H avg time per WU with v4.72 was 11.8 hours
Also when comparing chips do not underestimate the advantage of HT.
____________

team.

Catch your own wave...

> Well when you compile with gcc you would think it would optimize as much as it
> can for the 686 arch. I realize none of us run that arch anymore, but at the
> same time, and therefore using the athlon-xp or pentium4 arch would give
> better FP results.
>
> The only thing is though, for windows to work on different archs, there is no
> way it can be optimized for an athlon-xp or penitum4 either.

What you are forgetting is that those are only sub-categories of the x86 architecture. GCC is designed to produce code for PowerPC, Sparc, etc., etc.....

There are also trade-offs to consider. Sometimes, optimising for speed will produce larger executables - not always desireable. The default optimisations in GCC are a reasonable compromise to ensure stable executeables of acceptable size and with acceptable speed in most situations. It is the programmers prerogative to adjust those compromises for a given situation. A rich set of compiler flags is available for that purpose.

However, as I touched upon previously, another factor is the optimisation of the library routines with which the executable is linked. Linking can be done either statically or dynamically.

With dynamic linking, library routines are provided by the end-users system and loaded into memory as and when required and so the programmer has no control over them. The advantages of dynamic linking are a smaller executable and bug-fixes/enhancements to the libraries will automatically be available to the executable. It does mean, though, that the end-user will often need to have a particular version of the libraries available.

With static linking, the library routines are built in to the executable at compile time. This does produce much larger executables and requires a re-compile if anything changes in the libraries. In return, it gives the programmer more control over the end result and removes any dependancy requirements on the end-users system.

To achieve optimum performance in the Einstein app, it would be necessary to re-compile the libraries used with optimisation for each of the platforms supported and to also compile the app with similar optimisations and then to link statically. Finding the best optimisations for each supported architecture could, in itself, take a great deal of time and experimentation.

Releasing the source code would allow users to do this experimentation and produce optimised apps for a number of architectures as has been the case with the BOINC client. In fact, I always compile my own BOINC client and will, when I get around to it, have another go at the SETI app. My earlier attempts failed to get that to compile at all due to errors in the source code tarball.


____________
Be lucky,

Neil

Intel Prescott 3.2GHz HT on. 1GB PC3200 RAM. Running XP SP2.
Benchmarks using BOINC Manager V4.24 (real benchmarking)
Measured floating point speed 1339.55 million ops/sec
Measured integer speed 1709.98 million ops/sec

Einstein WU's take between 38,000 and 42,000 seconds each.
LHC million rotation WU's 45,000 to 47,000 seconds
Protein Pred 5,700 to 8,000 seconds
Seti 8,000 to 16,000 seconds
Climate 2.5 to 2.8 sec /time step

____________
Paul

I don't think the CPU is the limiting factor. However your PC2700 RAM might be unless you are able to run it non-sync with the FSB. I can't look at it right now but I've got a Sempron 2200 box at home with an Asrock K7S41 but the non-GX version. The K7S41GX is rated to DDR333 whilst the K7S41 is rated to DDR400. The non-GX version is slightly dearer than the GX but still very much a budget board. I'm absolutely sure that I've got the home system running at 200mHz FSB but using PC3200 RAM.

The only difference between the K7S41GX and the K7S41 seems to be the onboard graphics. Are you using that or an external graphics card? In any case I think the problem you are having is more likely to be related to the lack of a PCI/AGP lock on these boards. It's not really documented in the motherboard manual (same for both MBs) but at certain FSBs the PCI and AGP busses will be in spec at 33 and 66 mHz and at other values of FSB they will be quite out of spec. When I was playing around with mine, I'm sure that it was locking up at 180-190mHz and OK at 200mHz or something like that. It was explained to me at the time that choosing FSBs of 166 or 200 would have the PCI/AGP busses close to spec whilst mid-range values like 180-190 would be quite out of spec and give problems. Have you tried 200mHz?

Also, using improved air cooling is a lot easier than water cooling. Just lapping the heatsink, using Arctic Silver and putting on a bigger faster fan can give you much better cooling. My cpu temp as measured by motherboard monitor is around 55C (summer - no aircon) and seems to be quite OK. It has been into the low 60s in a heatwave without locking up whilst running Seti. I try to keep below 55C and I'm sure it's OK at that level

Gary,

you are right, the diff between the gx and non gx version of the asrock MB is the fsb of 333 and 400mhz. However, if you raise the fsb to 200mhz, it will run at 400Mhz for the DDR, even if it is a pc2700 memorycard. The kingston seems to have no problem with that. If I try to run it at 210fsb, the system locks up just after the 'welcome" screen of windows XP. I guess It has a big deal to doi with the DDR ability to handle the 400+ fsb settings on a full load. It runs okay at 200mhz and CPU-Z tells me my mem is running at 400mhz at that setting. Now, the problem is that I get a sharp rise in temp as soon as boinc starts computing. CPU-Idle is kicking in all the time, it is set at 55C max. I do not dare to go above that for long crunching sessions. Sooo, the higher Clockspeed is surely faster, but the tempmanagers slow it down. Therefore I choose the 190FSB, it keeps the temp around 54C, without the loss of cycles involved with colling the cpu. it is faster.. the boinc does not count the realtime cpu usage, but the actual time the cpu was used for calculations.. the cpu idle time at higher freq is lost time, and therefore slower..

I`ll go get me some of that silverpaste to get a better thermal contact, maybe it will help..
____________

I run about 48 to 59 depending on WU and OC settings. 200 - 210MHz FSB. That's with stock cooling and 4 case fans. (3 in and 1 exhaust) That 60 to 80mm converter sounds good though, wonder if there would be any advantage to mounting an 80 mm over a 60 mm fan for a forced air effect.
____________

Hi -
I'm running a new AMD Sempron 3100 - here's the info from my machine profile.

CPU type AuthenticAMD AMD Sempron(tm) Processor 3100+
Operating System Microsoft Windows XP Home Edition, Service Pack 2, (05.01.2600.00)
Memory 510.73 MB
Measured floating point speed 1670.89 million ops/sec
Measured integer speed 4574.31 million ops/sec

My time per work unit is about 7.5 hours.



____________

> Hi -
> I'm running a new AMD Sempron 3100 - here's the info from my machine profile.
>
>
> CPU type AuthenticAMD AMD Sempron(tm) Processor 3100+
> Operating System Microsoft Windows XP Home Edition, Service Pack 2,
> (05.01.2600.00)
> Memory 510.73 MB
> Measured floating point speed 1670.89 million ops/sec
> Measured integer speed 4574.31 million ops/sec
>
> My time per work unit is about 7.5 hours.
>
>
Hmm. My Athlon XP 2600 does much better:
Measured floating point speed 1850.6 million ops/sec
Measured integer speed 4442.28 million ops/sec
About 6 hrs 20+ mins per WU.
____________


Seti Classic Final Total: 11446 WU.

> Hmm. My Athlon XP 2600 does much better:
> Measured floating point speed 1850.6 million ops/sec
> Measured integer speed 4442.28 million ops/sec
> About 6 hrs 20+ mins per WU.

AMD model numbers are very confusing in a situation like this because they try to combine cache size, bus speed and processor speed in one number. A Sempron only has 256KB of L2 cache, but the socket 754 version has the DDR memory control on chip reducing memory latency and only runs at 1.8Ghz. An XP 2600+ is either 2.133Ghz, 512KB L2 cache with a 133Mhz(266DDR) or 2.133Ghz, 256KB L2 cache with a 166Mhz(333MhzDDR) bus. Those numbers seems reasonable for a 1.8Ghz AMD chip since lower latency/faster memory access isn't as important for E@H.

3.2 GHz P4 HT, 2GB DDR2 SDRAM @ 533MHz, Win XP SP2

CPU's: 2
Whet: 1377
Dhry: 2613

My times are skewed because I typically play Battle for Middle Earth while SETI and Einstein are running (database work keeping me from activating my transferred account for SETI). SETI has been finishing under 4 hours on this comp. With my interruptions, Einstein is finishing in about 10-12 hours.

I'm no expert so I must ask: how does it stack up and could it do better?
____________

AMD Barton XP2400-M @2.09ghz 512k L2
512mb PC3200 ram
Whet: 1960.26
Dhry: 4710.78

E@H: 6 hr
S@H: 2.5-2.75 hr.
P@H: 55 min.
CPDN: a really long time


____________

*I still know CRAP when I see it.

The Athlon numeric co-processor is twice as fast as the P4 co-processor, clock cycle for clock cycle. Also, the higher speed P4s will clock throttle under 100% CPU load unles they have very good cooling.

My Athlon XP 2700:

FPU: 2025.32 million ops/sec
Integer: 4919.23 million ops/sec

Time for a WU: 5:30 to 6:00 hours


____________

> 3.2 GHz P4 HT, 2GB DDR2 SDRAM @ 533MHz, Win XP SP2
>
> CPU's: 2
> Whet: 1377
> Dhry: 2613
>
> My times are skewed because I typically play Battle for Middle Earth while
> SETI and Einstein are running (database work keeping me from activating my
> transferred account for SETI). SETI has been finishing under 4 hours on this
> comp. With my interruptions, Einstein is finishing in about 10-12 hours.
>
> I'm no expert so I must ask: how does it stack up and could it do better?
>

I have a similar setup but only half the RAM. Times are fairly consistent though, so RAM seems to have only a minor effect on processing power. Or at least 1GB is no constraint to processing speed.
____________

team.

Catch your own wave...

> The Athlon numeric co-processor is twice as fast as the P4 co-processor, clock
> cycle for clock cycle. Also, the higher speed P4s will clock throttle under
> 100% CPU load unles they have very good cooling.
>
I'm not so sure about that Evan...I've optimized seti@home client so I have some knowledge here.

It seems to me that the design of the Intel FPU makes it practically impossible to write totally optimal code in C, and only partly so in Assembly language.

However, as has been shown, Intel can get more performance if it does Hypterthreading (ie it can output more work units in a 24 hour period). It is still only one CPU core, but pretending to be two. The reason is all of the various rules required by the Intel FPU to keep it chugging along.

Two problems with Intel are latency and throughput.

Example:

Throughput - You can't execute a floating point addition more often than every other CPU cycle. Same for a multiply, or most other operations.

Latency - A Pentium IV floating point multiply will finish computing 5 cycles after it begins. If the next instruction in the program tries to make use of this multipy result, it has to twiddle its thumbs for 4 cycles to act on the result.

This causes problems because many code sequences involve maybe 5 multiplies followed by 4 additions on the same set of numbers. This isn't so easy to interleave (multiply | add | multiply | add | etc.).

In truth if everything could be sequenced correctly the FPU in an intel can begin one and complete another FPU instruction every CPU cycle. Even with the pipelining of AMD and multiple execution units I think one completion every cycle isn't possible there.

Hyperthreading comes closer to achieving this on Intel as the CPU scheduler has two different program threads to work with, and can find those alternating instructions and latency delayed instructions can be dealt with. So while thread #1 is waiting for the results of a multipy, thread #2 can begin an addition or two.

____________

What you say seems to support my contention. Also, the real world benchmarks support it as well. I haven't programmed the P4 in assembler so I will take your word for it. The P4 does have a very long pipeline and if the execution prediction fails the flush takes a long time. Of course, that is where hand optimization comes in.

The clock throttling problem is very real. I sell computers for a living and I see machines from various name brand manufacturers that have suboptimal cooling. When pushed it is amazing how slow a P4 can be. It will cut the clock rate in half, at least, in order to maintain safe temperatures.
____________

When I started in 1999 I got a Proteva with the PIII 500 when it 1st came out and then started running seti classic in 2000 and after a year of the slow times (14hrs to 18hrs) I bought a Systemax with a AMD 1700+ XP with 512.

Then for some reason when the wife made me go store shopping I picked up a HP with XP and a 2.5 P4 w/512ram.

It ran seti classics at ave. of 4:20 and Einsteins at ave. of 9.5hrs.

Then last year about this time I got a Systemax with XP Pro the AMD 64 3200+ and 1gig ram and a 200gig HD.

It has a seti classic ave. of 2:20 and an Eisnstein ave. of 5:45


AMD is the one for me.

And XP Pro (Bill Gates is my neighbor : )






-Samson Ben Yoseph-
____________

JAHMAGIC,

Just wondering, do you work at Fermilab?
____________
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell

> > as I suspected... even the cheapest AMD, the sempron 2200+ I have is
> kicking
> > the butts off most Intel P4 systems.. now how funny is that ?
> >
> > 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
> >
>
> I was always under the assumption that sempron's were worse than an Athlon XP.
> But my Athlon XP 2400+ takes almost 12 hours for a WU.
>

Something else is holding your system back. My Athlon XP 1900+ takes about 8 hours for a E@H WU and about 3 hr 45 min to 3 hr 50 min typically for a SETI WU.

Are you running Linux on it or something?

BTW, my CPU wasn't OCed, so is running at 1.6 GHz
____________

Duron 1100/ASUS A7V333/128 RAM PC2700 SAMSUNG
E@H WU - little less than 12 hours.
SETI@home Classic - 5 hours 30 minutes per WU (average).
____________

Athlon Mobile 2600+ @ 2400 mhz / Abit NF7-S rev 2
E@H = 5hrs 20min
____________

[oops]

Ah, this might work:
[pre]
AppName Min Max Avg WU Count
mfoldB125 4059.515625 9951.40625 7238.0598221144 956
mfoldB120 4.2589850 15765.953125 7100.4262898567 382
setiathome 22.90625 30884.96875 13010.96382848 2641
einstein 28794.828125 51789.125 39240.910247093 43
sixtrack 2.8125 63085.390625 2671.8234310456 3115
hadsm3 1793074.375 2325737.25 2041320.1964286 14
[pre]

____________

> I have no indept knowledge about windows/intel but could imagine a mechanisme
> that at installation time code and Libs are just better selected and matched
> to the specific arch

I don't believe this. The dlls are all the same for all x86 archs.

> then is done with "general" linux distro's as most of the
> people run.

Like on Windows the general Linux distros have general 586/686 optimized packages and libs. On Gentoo however the system is completely compiled on installation time and can and is in general be tuned at last for the processor and mostly with more optimization flags on.

regards

martin

____________

> JAHMAGIC,
>
> Just wondering, do you work at Fermilab?
>




Yes, but I have been retired for about 8 years now and that was over 10 years ago.


Back when Fermilab's founder, the late physicist Robert Wilson was still alive.

(he is still there at the complex)

It is quite a place......6,800-acre complex that looks like a National Park!


Of course with particle accelerators under the ground.


http://www.fnal.gov/



I imagine since you are at Illinois Institute of Technology that you may have been to Batavia.




____________

> I imagine since you are at Illinois Institute of Technology that you may have
> been to Batavia.

I actually have a co-op there every other semester. It definately is a very cool place. I just hope they get the new linear accelerator they are vying for, so they don't die out after the LHC gets completed at CERN.



____________
such things just should not be writ so please destroy this if you wish to live 'tis better in ignorance to dwell than to go screaming into the abyss worse than hell

P4 1.6 Ghz 15 hours per WU with between 75-95 claimed credit.

____________

AMD Athlon XP-M 2600+ 6h45min per WU
____________

HP Pavillion a220n
Athlon XP 2600+ at 2.083 MHz, locked.
Asus A7N8X-VM 512MB SDRAM

wu times=6:20 to 6:45.

For only 2.083, I think it's a great time.


____________

I just started einstein@home this weekend as seti@home seems to be searching for electricians as opposed to aliens :D

I haven't used boinc nearly long enough yet to compile good averages.
Here's what I have come up with so far...

AMD Athlon XP 2600+ ( @ 2.25 GHz ) Linux
seti@home: ~02h:30m
einstein@home: ~10h:20m

-----------------------------------
AMD Sempron 2200+ (1.5 GHz) Linux
seti@home: ~04h:40m
einstein@home: ~15h:10m

-----------------------------------
AMD Athlon 2800+ (2.0 GHz) Linux
seti@home: ~03h:40m
einstein@home: ~10h:50m

-----------------------------------
Intel Pentium 4 (1.80 GHz) Linux
seti@home ~05h:30m
einstein@home ~17h:05m

-----------------------------------
Intel Pentium 4 (1.60 GHz) Linux
seti@home ~07h:00m
einstein@home ~19h:24m

-----------------------------------
AMD Athlon K6/2 (700 MHz) Linux
seti@home ~10h:05m
einstein@home ~24h:10m

-----------------------------------
Intel Pentium 4 (1.40 GHz) Linux
seti@home ~07h:45m
einstein@home ~23h:05m

-----------------------------------
Intel Pentium 4 (1.70 Ghz) Linux
seti@home ~06h:05m
einstein@home ~18h:10m

-----------------------------------
Intel Pentium III (450 MHz) Linux
seti@home ~15h:25m
einstein@home ~48h:03m (estimated)

-----------------------------------
Intel Pentium III Coppermine (1.0 GHz) Linux
seti@home ~09h:05m
einstein@home ~24h:16m (estimated)

-----------------------------------
AMD K6 3D (450 MHz) Linux
seti@home ~10h:42m (One very small WU)
einstein@home ~102h:19m (estimated)

-----------------------------------
Intel Pentium III (866 MHz)
seti@home ~09h:15m
einstein@home ~39h:50m (estimated)



____________

> I just started einstein@home this weekend as seti@home seems to be searching
> for electricians as opposed to aliens :D
>
> I haven't used boinc nearly long enough yet to compile good averages.
> Here's what I have come up with so far...
>
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://einstein.phys.uwm.edu/forum_thread.php?id=1208
____________


Seti Classic Final Total: 11446 WU.

Same AMD XP 2400+ computer takes 22000 seconds / packet with windows 2000 and 38000 seconds with linux. Even worst thing is that einstein gives zero credit at half of all ready made linux packets (not overclocked). So einstein program has been removed from all of my linux computers until...

____________

<strong>AMD Athlon64 3700+ (Win2000)</strong>
FPU : 2267.72 million ops/sec
INT: 6284.15 million ops/sec
Avg : 18.9k sec

<strong>AMD Athlon64 3200+ (Linux)</strong>
FPU : 2134.64 million ops/sec
INT : 4893.04 million ops/sec
Avg : 38.7k sec

<strong>AMD Athlon64 3000+ (Linux)</strong>
FPU : 2138 million ops/sec
INT : 4878.45 million ops/sec
Avg : 37.7k sec

<strong>AMD AthlonXP 3000+ (Win2000)</strong>
FPU : 2002.54 million ops/sec
INT : 4820.02 million ops/sec
Avg : 21.6k sec

<strong>AMD AthlonXP 3000+ (Linux)</strong>
FPU : 2257.9 million ops/sec
INT : 4770.85 million ops/sec
Avg : 35.8k sec

<strong>AMD AthlonXP 2800+ (Linux)</strong>
FPU : 2191.91 million ops/sec
INT : 4340.49 million ops/sec
Avg : 36.4k sec

<strong>Dual AMD AthlonMP 2800+ (Linux)</strong>
FPU : 2252.69 million ops/sec
INT : 4458.22 million ops/sec
Avg : 35.8k sec

<strong>AMD AthlonXP 2700+ (Linux)</strong>
FPU : 2269.82 million ops/sec
INT : 4813.92 million ops/sec
Avg : 35.7k sec

<strong>AMD AthlonXP 2600+ (Linux)</strong>
FPU : 2181.68 million ops/sec
INT : 4538.67 million ops/sec
Avg : 37.1k sec

<strong>AMD AthlonXP 2500+ (Linux)</strong>
FPU : 1933.56 million ops/sec
INT : 3829.77 million ops/sec
Avg : 41.2k sec

<strong>AMD AthlonXP 2400+ (Linux)</strong>
FPU : 2102.69 million ops/sec
INT : 4149.54 million ops/sec
Avg : 38.2k sec

<strong>AMD AthlonXP 2200+ (Linux)</strong>
FPU : 1886.7 million ops/sec
INT : 3986.49 million ops/sec
Avg : 42.9k sec

<strong>AMD AthlonXP 2100+ (Linux)</strong>
FPU : 1820.19 million ops/sec
INT : 3648.92 million ops/sec
Avg : 45.4k sec

<strong>Dual AMD AthlonMP 2000+ (Linux)</strong>
FPU : 1759.1 million ops/sec
INT : 3483.2 million ops/sec
Avg : 48k sec

<strong>AMD AthlonXP-M 1800+ (WinXP)</strong>
FPU : 1389.65 million ops/sec
INT : 3383.78 million ops/sec
Avg : 30.9k sec
--------------------------------------------
From those numbers it becomes clear that :
- the E@H Linux Client is <strong>absolutely terrible</strong> in terms of performance (adding to the 0 Credits/Invalidation bug)
- it loves raw clock rates
- it does not take advantage of larger (512k+) L2 Caches

(Performance numbers based on still relatively few WorkUnits, therefor subject to certain variation)
____________
Scientific Network : 44800 MHz - 77824 MB - 1970 GB

AMD Barton 2500+, 32bit, Fedora Core 3, average einstein wu over 21 results, 42k seconds.

AMD64 2600+, 64bit, Fedora Core 3 x86_64, average enstein uw over 11 results 78k seconds.

Something is majorly wrong with their code running on AMD64 in 64bit linux.

Duke out


____________

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://einstein.phys.uwm.edu/forum_thread.php?id=1208

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 3.0E0@4.510 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 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
____________

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 Fällen entscheide man sich für das Richtige"(K.Kraus) =
support your local anti-geek team!

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 2.4@2.88GHz, 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






____________
<p>The
Night Menace, My Einstein@Home Water Cooled PC

I think my WU's will all be in the 5 hour range.

My current stats

____________
<p>The
Night Menace, My Einstein@Home Water Cooled PC

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

AMD Sempron 2500+ (10.5x167, 1gb PC3200 DDR, DFI KT600-AL)
Windows XP SP2

[---] 1594 double precision MIPS (Whetstone) per CPU
[---] 2616 integer MIPS (Dhrystone) per CPU

Average WU time: 7 hrs 31 min

----------------------

AMD Athlon 1300 (133 MHz FSB, 512mb PC133 SDRAM, IWill KK266plus)
Windows XP SP2

[---] 1203 double precision MIPS (Whetstone) per CPU
[---] 2082 integer MIPS (Dhrystone) per CPU

Average WU time: 9 hrs 48 min


____________


------------------
If I love truth, then why don't I love correction? Perhaps Jack Nicholson was right - maybe I really can't handle the truth!

Processor XP2500 lightly overclocked.
OS is linux Fedora Core 2
Client is windows 4.19 running under wine.


1989620 528440 14 Mar 2005 13:53:06 UTC 17 Mar 2005 16:06:34 UTC Over Success Done 21,441.83 84.74 pending
1989615 528439 14 Mar 2005 13:53:00 UTC 17 Mar 2005 10:08:59 UTC Over Success Done 21,445.03 84.76 pending
1989611 528438 14 Mar 2005 13:52:54 UTC 17 Mar 2005 4:11:13 UTC Over Success Done 21,439.92 84.74 pending
1989607 528437 14 Mar 2005 13:52:48 UTC 16 Mar 2005 22:13:47 UTC Over Success Done 21,440.34 84.74 100.32
1989603 528436 14 Mar 2005 13:52:42 UTC 16 Mar 2005 16:16:15 UTC Over Success Done 21,435.01 84.72 87.19
1930090 515153 12 Mar 2005 20:34:51 UTC 16 Mar 2005 10:18:46 UTC Over Success Done 21,468.50 84.85 pending
1930086 515152 12 Mar 2005 20:34:45 UTC 16 Mar 2005 4:20:32 UTC Over Success Done 21,441.09 84.74 99.09
1930082 515151 12 Mar 2005 20:34:39 UTC 15 Mar 2005 22:21:59 UTC Over Success Done 21,438.30 84.89 pending
1930078 515150 12 Mar 2005 20:34:32 UTC 15 Mar 2005 16:24:30 UTC Over Success Done 21,448.26 84.93 pending
1930074 515149 12 Mar 2005 20:34:26 UTC 15 Mar 2005 10:26:50 UTC Over Success Done 21,462.54 84.98 84.98
1930070 515148 12 Mar 2005 20:34:20 UTC 15 Mar 2005 4:28:47 UTC Over Success Done 21,437.70 84.89 99.38

____________

--- - 2005-03-17 17:08:51 - Benchmark results:
--- - 2005-03-17 17:08:51 - Number of CPUs: 1
--- - 2005-03-17 17:08:51 - 2033 double precision MIPS (Whetstone) per CPU
--- - 2005-03-17 17:08:51 - 4909 integer MIPS (Dhrystone) per CPU
--- - 2005-03-17 17:08:51 - Finished CPU benchmarks

It's an overclocked Athlon XP 2600+ running about 2192 Mhz. Computer has 1 Gig of DDR RAM.
____________
<BR><BR>
Join BOINC@California!

Here is what FX55 numbers look like

3/18/2005 4:15:51 AM|| 2486 double precision MIPS (Whetstone) per CPU
3/18/2005 4:15:51 AM|| 4641 integer MIPS (Dhrystone) per CPU

The FX53 was
2257 double precision MIPS (Whetstone) per CPU
4189 Interger MIPS (Dhrystone) per CPU


WU's will be in the 18k+ second area.

After 20 completed WU's the FX53's average is 19294.396 seconds
or 5:21.57 per WU.

____________
<p>The
Night Menace, My Einstein@Home Water Cooled PC

AMD64 FX-53

Measured floating point speed 2278.32 million ops/sec
Measured integer speed 4235.6 million ops/sec

Last 3 completed results:
2036285 538572 15 Mar 2005 22:56:16 UTC 18 Mar 2005 18:56:55 UTC Over Success Done 18,964.64 71.49 96.59
1963219 434764 14 Mar 2005 23:31:48 UTC 17 Mar 2005 19:43:31 UTC Over Success Done 18,968.55 69.33 95.42
1879830 503869 14 Mar 2005 7:47:21 UTC 16 Mar 2005 4:29:56 UTC Over Success Done 18,960.94 69.30 pending

AMD64 FX-55

2168891 568240 19 Mar 2005 17:09:14 UTC 20 Mar 2005 8:33:51 UTC Over Success Done 18,412.19 75.95 pending

2123188 557911 18 Mar 2005 10:35:33 UTC 19 Mar 2005 14:13:53 UTC Over Success Done 18,202.58 68.53 81.59

2107933 554465 18 Mar 2005 5:21:08 UTC 18 Mar 2005 12:53:46 UTC Over Success Done 18,164.02 68.39 70.08


____________
<p>The
Night Menace, My Einstein@Home Water Cooled PC

AMD64 3400+ Newcastle

Benchmark results:
Number of CPUs: 1
2561 double MIPS (Whetstone) per CPU
6991 integer MIPS (Dhrystone) per CPU
Finished CPU benchmarks

2111885 555363 19 Mar 2005 23:13:51 UTC 20 Mar 2005 8:48:34 UTC Over Success Done 16,812.55 92.59 pending
2105486 521654 19 Mar 2005 18:33:36 UTC 20 Mar 2005 4:08:22 UTC Over Success Done 16,921.39 93.19 93.19
2099553 552594 19 Mar 2005 12:37:39 UTC 19 Mar 2005 23:13:51 UTC Over Success Done 16,853.52 92.82 pending
2087313 549884 19 Mar 2005 7:59:55 UTC 19 Mar 2005 17:08:24 UTC Over Success Done 16,746.73 92.23 pending
2061694 544206 19 Mar 2005 2:49:38 UTC 19 Mar 2005 12:37:39 UTC Over Success Done 16,662.17 90.98 90.98
2061687 544204 19 Mar 2005 0:38:51 UTC 19 Mar 2005 7:59:55 UTC Over Success Done 16,817.94 91.83 pending

AMD64 3400+ Newcastle 16800 seconds. Very nice!
What is the clock speed?
____________

AMD64 3400+ Newcastle 16800 seconds. Very nice!
What is the clock speed? Quote


AMD64 3400+ @ 2650 mhz
10x265 HTT


____________

AMD64 3400+ @ 2650 mhz
10x265 HTT

That is interesting. I was under the impression that the client worked strictly on MIPS independant of the memory bus. I raised my FX to 2650 13x203 HTT to see how it compares with your figures. I'l post them tomorrow...

____________
<p>The
Night Menace, My Einstein@Home Water Cooled PC

Note how I cut off about 3000 seconds just by overclocking it to 200 extra Mhz. Well, it's not "just," more like 10% on my Athlon XP-2600 (which strangely reports as Athlon XP-2400). My goal is to get sub-20,000 second numbers on this old setup, as I'm too cheap (think college student level of poorness ;)) to get a 64-bit system like some of you. Maybe I should build another XP 2600+ system. ;)

19 Mar 2005 3:04:10 UTC 20 Mar 2005 8:16:26 UTC Over Success Done 22,187.50 89.13 78.62
18 Mar 2005 20:51:58 UTC 19 Mar 2005 22:17:23 UTC Over Success Done 22,085.27 88.72 77.90
18 Mar 2005 5:06:59 UTC 19 Mar 2005 3:04:10 UTC Over Success Done 21,608.12 86.80 74.57
17 Mar 2005 20:38:24 UTC 18 Mar 2005 20:51:58 UTC Over Success Done 22,549.13 90.59 pending
17 Mar 2005 2:41:53 UTC 18 Mar 2005 5:06:52 UTC Over Success Done 23,785.30 95.55 86.68
16 Mar 2005 0:58:32 UTC 17 Mar 2005 20:38:24 UTC Over Success Done 25,509.21 91.44 79.10
15 Mar 2005 4:11:02 UTC 17 Mar 2005 2:41:53 UTC Over Success Done 25,201.47 90.33 103.40
14 Mar 2005 6:26:06 UTC 16 Mar 2005 0:58:32 UTC Over Success Done 25,554.57 91.60 94.68

____________
<BR><BR>
Join BOINC@California!

Dual 2.5Ghz PowerPC, OS 10.3.4 & 2G ram, runnung E@H.

2005-03-20 20:41:32 [---] Benchmark results:
2005-03-20 20:41:32 [---] Number of CPUs: 2
2005-03-20 20:41:32 [---] 1396 double precision MIPS (Whetstone) per CPU
2005-03-20 20:41:32 [---] 4591 integer MIPS (Dhrystone) per CPU
2005-03-20 20:41:32 [---] Finished CPU benchmarks

AMD ATHLON 2500XP+ Barton core
512 meg of ram not oc'ed slowest wu 25,597 seconds
windows xp home sp2 fastest wu 25,097 seconds 16 results


AMD ATHLON
128 meg of ram
windows xp pro sp2
not oc'd 42000 seconds on average
____________

loose -> lose
dependancy -> dependency
littler -> little
independant -> independent
____________

"My other computer is a virus farm."

> loose -> lose
> dependancy -> dependency
> littler -> little
> independant -> independent
>

lysdexia -> dyslexia ;)
____________
<BR><BR>
Join BOINC@California!

Fastest WU to be completed by my 2 systems so far is from my Duron, at 26,500 seconds. No, my Duron isn't overclocked and runs at 1.8GHZ.
____________

Ok I have several WU's that were done last night and some interesting insights.

FX53 13x203=2639Mhz

2207550 81562 21 Mar 2005 1:32:50 UTC 21 Mar 2005 10:58:16 UTC Over Success Done ...16,869.13... 70.56 pending

2187045 572354 20 Mar 2005 20:50:56 UTC 21 Mar 2005 6:15:30 UTC Over Success Done ..16,876.17... 70.59 pending

2186978 572338 20 Mar 2005 10:29:02 UTC 20 Mar 2005 23:16:56 UTC Over Success Done ...16,975.91... 71.00 pending

What I can conclude from this is MIPS is everything.
The FX53 with dual on die memory controllers and 1 mb of l2 cache has really no effect on WU performance and with that front side bus doesn't either.

Interesting enough the benchmark for the client is flawed

AAdjuster's
Benchmark results:
Number of CPUs: 1
2561 double MIPS (Whetstone) per CPU
6991 integer MIPS (Dhrystone) per CPU
Finished CPU benchmarks

The AAdjuster's Dhrystone benchmarks are out of wack.

The FX53's benchmarks at 2639 MHz
Measured floating point speed 2530.23 million ops/sec
Measured integer speed 4697.29 million ops/sec

These results are clearly at odds with each other.
Another interesting thing is that AAduster's credits are inflated even though the time is nearly the identical..
My claimed credts all range in the 72 area where all of his/hers are in the 90's
very strange indeed.





____________
<p>The
Night Menace, My Einstein@Home Water Cooled PC

> Interesting enough the benchmark for the client is flawed
>
> AAdjuster's
> Benchmark results:
> Number of CPUs: 1
> 2561 double MIPS (Whetstone) per CPU
> 6991 integer MIPS (Dhrystone) per CPU
> Finished CPU benchmarks
>
> The AAdjuster's Dhrystone benchmarks are out of wack.
>
> The FX53's benchmarks at 2639 MHz
> Measured floating point speed 2530.23 million ops/sec
> Measured integer speed 4697.29 million ops/sec
>
> These results are clearly at odds with each other.
> Another interesting thing is that AAduster's credits are inflated even though
> the time is nearly the identical..
> My claimed credts all range in the 72 area where all of his/hers are in the
> 90's
> very strange indeed.
>

I'd guess you're using 4.25 and he's using 4.19

Dominique

[edit]
Using 4.25 I get around 65. With 4.19 I used to get around 85.


____________

*I still know CRAP when I see it.

Single AMD Opteron 144
MSI Master-F dual Opteron MB
1GB ECC PC2700
320GB RAID5 3Ware SATA RAID
Win 2K Pro running file shares

3181 Integer MIPS
1713 Double MIPS

Try beating that with a $250 chip from Intel.
I can't imagine what a dual 250 would do, much less an 8-way 852. *drool*

<BR>

My laptop doesn't seem to do a good job. Approx 13 hours to complete one WU, compared to my Duron that takes 8 hours. My laptop has a mobile Athlon XP 1900+, but I think it's because of the 386MB SDR memory that's causing it to slow down.
____________

> Single AMD Opteron 144
> MSI Master-F dual Opteron MB
> 1GB ECC PC2700
> 320GB RAID5 3Ware SATA RAID
> Win 2K Pro running file shares
>
> 3181 Integer MIPS
> 1713 Double MIPS
>
> Try beating that with a $250 chip from Intel.

On E@H, it can be done, no sweat. I have a single 3.0 GHz P4 478 Prescott running at about 3.4 GHz. It cost $189.00 in the retail box. It routinely turns in results in 41,200 sec. Sounds slow? But it is hyperthreading & does two units from one CPU in that same 41,200 sec. So on a per-unit basis, it does 41200/2=20600 sec per unit. For most tasks, hyperthreading is not as valuable, but for E@H it exactly doubles the throughput. This computer is called "yeongw2k" if you want to check. Running with old, slow PC2100 memory too.

ADDMP (I like your Opteron, though. Like to have one myself.)

____________

> > Single AMD Opteron 144
> > MSI Master-F dual Opteron MB
> > 1GB ECC PC2700
> > 320GB RAID5 3Ware SATA RAID
> > Win 2K Pro running file shares
> >
> > 3181 Integer MIPS
> > 1713 Double MIPS
> >
> > Try beating that with a $250 chip from Intel.
>
> On E@H, it can be done, no sweat. I have a single 3.0 GHz P4 478 Prescott
> running at about 3.4 GHz. It cost $189.00 in the retail box. It routinely
> turns in results in 41,200 sec. Sounds slow? But it is hyperthreading &
> does two units from one CPU in that same 41,200 sec. So on a per-unit basis,
> it does 41200/2=20600 sec per unit. For most tasks, hyperthreading is not as
> valuable, but for E@H it exactly doubles the throughput. This computer is
> called "yeongw2k" if you want to check. Running with old, slow PC2100 memory

I should have said that I had looked up how PickCoder's Opteron 144 was doing on E@H & it was running 24800 sec/unit compared to 20600 sec/unit for the 3 GHz P4 with hyperthreading.

And I should not have said that hyperthreading doubled the throughput. I have never run it with hyperthreading turned off & expect it would look quite bad.

[This is correct-the-blunders night.]

ADDMP
____________

> And I should not have said that hyperthreading doubled the throughput. I have
> never run it with hyperthreading turned off & expect it would look quite
> bad.

It does not look that bad. HT does not double the performance. It does give a boost to throughput between 10% to as high as 70% more work done per unit time.

We did fairly extensive tests during the late stages of the Beta test program with HT on and off and it does surprisingly well at improving the total production. And I say surprising in that our usage of the processors is very atypical ... we heavily use the math portions and do not have a very "standard" mix of instruction streams.

I mean, my G5 does more work because of the two CPUs than does any of my HT systems, but it is a cheap way to "see" two CPUs.
____________

> > Single AMD Opteron 144
> > MSI Master-F dual Opteron MB
> > 1GB ECC PC2700
> > 320GB RAID5 3Ware SATA RAID
> > Win 2K Pro running file shares
> >
> > 3181 Integer MIPS
> > 1713 Double MIPS
> >
> > Try beating that with a $250 chip from Intel.
>
> On E@H, it can be done, no sweat. I have a single 3.0 GHz P4 478 Prescott
> running at about 3.4 GHz. It cost $189.00 in the retail box. It routinely
> turns in results in 41,200 sec. Sounds slow? But it is hyperthreading &
> does two units from one CPU in that same 41,200 sec. So on a per-unit basis,
> it does 41200/2=20600 sec per unit. For most tasks, hyperthreading is not as
> valuable, but for E@H it exactly doubles the throughput. This computer is
> called "yeongw2k" if you want to check. Running with old, slow PC2100 memory
> too.
>
> ADDMP (I like your Opteron, though. Like to have one myself.)
>

I just loaded a 3.0Ghz Prescott here and notice that too. You're not really comparing apples to apples, though. That's 2 WU running at a time. I've yet to find a way to run more than one WU, for a project, on a single CPU. When I run E@H, it slightly raises the temp on the chip. I was surprised after a few days of continuous crunching. Task Mgr shows 99% usage, but I know it's not even touching the available CPU bandwidth. The machine doesn't even blink with E@H running. I can run defrag(NTFS) and any number of other applications and they barely reflect, not that I can tell, that E@H is using "99%" of the CPU. I'm eager to find out if multiple WU on a 64-bit capable CPU would reduce the WU time or not. Anyone got any ideas?

Glen
____________

22,000 sec. +/- 500 sec.

AMD 2400+ Mobile at 2.08 ghz on a Biostar M7NCG 400 MB with 512m RAM

$77 for the CPU - $63 for the MB - $76 RAM


____________

*I still know CRAP when I see it.

PickCoder wrote:

> I've yet to
> find a way to run more than one WU, for a project, on a single CPU.>
> Glen

I was interested in your reply, but I do not quite get the sentence above.

I started up BOINC-E@H on the hyperthreading P4 & it immediately began running two WUs of E@H simultaneously. It is still running two WUs simultaneously.

That is what it does on a real-dual-CPU machine too.

Are you saying that is different from your experience?

ADDMP
____________

Average time about 29 500 seconds.
1536 Whetstone
2633 Dhrystone
PC "custom-built" by a small firm.
AMD Athlon XP 2000+ (1.7 GHz)
512 MB RAM (2x256)
I don't know if it matters, but the graphics card is GeForce FX 5500 (128/128). No problems with the screen-saver or the "show graphics" option.
No errors up to now. (I'm running only Einstein and SETI.)
Have fun!
Vladimir

Oh, BTW - BOINC 4.25 splashed over 4.19 (I forgot to uninstall it).
____________

The H/T fakes the O/S into thinking there's two cores and two sets of registers. The CPU can only execute one instruction at a time, so it has to switch between the instruction streams, much like thread execution under an O/S. Seeing the CPU throughput double on a P4 H/T makes me wonder how much I could squeeze out of a single Opteron. I would like to be able to run multiple computation threads on a single CPU, to see how much bandwidth the Opteron really is capable of. Would 2 BOINC threads kill the Opteron's performance or would it run both threads with the same WU time(or less) as a single thread? Unfortunately, I've yet to find a way to make BOINC run multiple WU threads when a single CPU is found. The CPU temp goes up a small amount, but the heat sink isn't even warm to the touch so I know it's not running @ max operating temp. I think the Opteron is being underused when running BOINC. Taskmgr shows 98-99% usage, but I seriously doubt the CPU is really loaded @ 99%. I've considered loading something like VMWare on it and running BOINC in several VME's. I'm not willing to buy it, though, just to test multiple threads. I'll check to see if there's a temp demo that'll lemme try it.

Glen

> I was interested in your reply, but I do not quite get the sentence above.
>
> I started up BOINC-E@H on the hyperthreading P4 & it immediately began
> running two WUs of E@H simultaneously. It is still running two WUs
> simultaneously.
>
> That is what it does on a real-dual-CPU machine too.
>
> Are you saying that is different from your experience?
>
> ADDMP
>
____________

> ... I think the Opteron is being underused when running BOINC. Taskmgr shows
> 98-99% usage, but I seriously doubt the CPU is really loaded @ 99%. I've
> considered loading something like VMWare on it and running BOINC in several
> VME's. I'm not willing to buy it, though, just to test multiple threads. I'll
> check to see if there's a temp demo that'll lemme try it.

Why do you think that 99% CPU usage is not high enough?

If you judge it by the fact that if you start anything else up and it does not seem "slow" that there is remaining CPU cycles; well, it is not a true understanding.

BOINC and most other DC programs run at a low priority so they will "naturally" fade into the background and your computer will not seem to be burdened with any work at all. If you do manage to force it to run more than one thread, all you will accomplish is to slow down the processing ...
____________

PickCoder wrote:

> The H/T fakes the O/S into thinking there's two cores and two sets of
> registers. The CPU can only execute one instruction at a time, so it has to
> switch between the instruction streams,

No, I don't think that's quite right. As I understand it, the P4 has parallel computation capabilities for SOME PARTS of the calculation. It can actually do more than one calculation at the exact same time. Hyperthreading was devised by Intel do make more efficient use of these duplicated sections of the CPU.

That's entirely different from time-slicing, which has been around much longer, & only produces the illusion of two or more actions happening at the same time, by switching rapidly from one to the other. Time-slicing does not increase through-put. It somewhat decreases through-put because there is some overhead in repeatedly stopping one calculation & starting up the other.

The MMX & 3DNOW & SSE & SSE2 & SSE3 sections also allow more than one calculation in the same clock cycle, but the E@H programers apparently haven't gotten around to putting them to work. But in Folding@Home, they greatly increase the throughput.

I don't know why AMD hasn't developed hyperthreading yet. Maybe someone else knows. Maybe Intel holds some patents.

But both Intel & AMD are said to be working on single chips that will contain two complete CPUs for late this year or early next year. That goes further than the P4 which only has duplication for some parts. Maybe AMD believes it has more to gain by concentrating its design force on this further development. I have seen reports that AMD might beat Intel into production.

ADDMP
____________

Hello everybody,

With my system, Einstein seems to always be around 6 1/2 hours.


einstein 4.79 H1_1308.9_1309.2_0.1_Tl03est02_4 06:32:10 100.00%
einstein 4.79 H1_1308.9_1309.0_0.1_Tl63est02_1 06:27:42 100.00%
einstein 4.79 H1J308.9_1309,1_0.1_Tl6_Test02_1 06:28:06 100.00%
einstein 4.79 HC1308.9_1309,2_0, CTl6_Test02_0 06:29:14 100.00%
einstein 4.79 HC1308.9_1309,3_0. CTl6_Test02_0 06:27:08 100.00%
einstein 4.79 H1J308.9_1309.4_0.1_Tl6_Test02_0 06:31:56 100.00%
einstein 4.79 H1J308.9_1309,0_0. CTl7_Test02_0 00:17:59 4.51% 06:21:14
mfoldB1254.24 t0227E_1_36878_3 02:18:49 100.00%
mfoldB1254,24 t0227E_1_42830_3 02:19:39 100.00%
mfoldB1254.24 t0227E_1_43000_2 02:16:15 100.00%
mfoldB1254.24 t0227E_1_43004_0 02:20:11 100.00%
mfoldB1254,24 t0227E_1_42820_3 02:19:01 100.00%
mfoldB1254,24 t0227E_1_34853_1 02:16:46 100.00%
mfoldB1254.24 t0227E_1_42753_3 02:19:35 100.00%
mfoldB1254.24 t0227E_1_42968_2 02:19:40 100.00%
mfoldB1254.24 t0227E_1_42989_1 02:22:19 100.00%
mfoldB1254.24 t0227E_l_42948_2 02:20:05 100.00%
mfoldB1254.24 t0227E_1_42494_5 02:16:22 100.00%
mfoldB1254.24 t0224E_1_93412_1 00:56:29 100.00%
setiathome 4.09 11 ja05ab ,27751.22881,823596,85_1 03:19:04 100.00%
setiathome 4.09 19dc04ab,27343.14545,873586,171_2 03:14:09 100.00%

.....................................................
AMD Athlon XP 2400+ 266 FSB Socket A - Thornton - OEM
Manufacturer AMD - Model AXDC2400DKV3C

CPU Rating 2000MHz
Socket Type SOCKET-A
Bus Frequency 266MHz
Multiplier 15
Core Voltage 1.65v

128 KB L1 Cache Size - 256 KB L2 Cache Size
.....................................................

Benchmarks and stats courtesy pcpitstop.com:

Processor Description:
AMD Athlon XP
Nominal Clock Speed 2000 MHz
Measured Clock Speed 2000 MHz
Speed Rating 5762 (?)

Main Board Description:
VIA Technologies, Inc. KT333CF-8235
BIOS Phoenix Technologies, LTD 6.00 PG 08/25/2003
System Board KT333CF-8235


Memory Configuration:
RAM installed 768 MB
Max RAM module size 32 MB
Memory Type 512+256,T16 DDR (budget ram)
Speed Rating 4955 MB/s

____________

Athlon XP 2600+
512 MB PC133

Einstein@Home WU crunch times average 6 hours, 15 minutes.
____________

>I don't know why AMD hasn't developed hyperthreading yet. Maybe someone else
>knows. Maybe Intel holds some patents.

AMD doesn't need hyperthreading as much as intel. Their pipeline is 5 to 10 instructions when a P4 is around 30 instructions. So on avereage the AMD chip gets 60% less pipeline stalls and 60% less overhead in refilling the pipeline when a stall does happen. In terms of total time wasted (and filled in with HT) this is probably the largest source because even with the best preditions it happens more often than the other sources that HT fills in.

In real numbers intel gains about 30% due to HT normally, AMD would probably gain less than 5%.

There could be patent/copyright issues as well.
____________
BOINC WIKI

BOINCing since 2002/12/8

> But both Intel & AMD are said to be working on single chips that will
> contain two complete CPUs for late this year or early next year. That goes
> further than the P4 which only has duplication for some parts. Maybe AMD
> believes it has more to gain by concentrating its design force on this further
> development. I have seen reports that AMD might beat Intel into production.

IBM is doing the same ... there are rumors that we may see dual core, dual processor systems from Apple this year. Still only a rumor though ...

The Intel dual cores, at least to start, will have HT, so, the dual core with HT will look like 4 CPUs where AMD will have one chip that has two CPUs. In this case, the dual processor AMD system will have 4 physical processors and the Intel Dual will have 4 physical processors, with the appearance of 8 logical processors.
____________

i am getting around 5 hours per WU on my 2.5ghz barton. around 7 hours on my 1.88ghz tbred b.

INTEL P4 3066@3500

Benchmark results:
Number of CPUs: 2
1510 double MIPS (Whetstone) per CPU
1305 integer MIPS (Dhrystone) per CPU
Finished CPU benchmarks

does anyone know if there exists a list like this one ?
http://www.setigermany.de/statistiken/sg/benchmark/benchmark.htm

____________

Certainly not an exact match, but look here:
http://www.boincstats.com/stats/host_cpu_stats.php?pr=einstein&st=0&to=100&or=10
____________

Einstein times:
Power Mac G5 @ 2X2.5 = 3.5 hrs (2 at a time)
iMac G5 @ 2 Ghz = 4.5 hrs
eMac G4 @ 800 Mhz = 13 hrs

SETI times:
Power Mac G5 @ 2X2.5 = 2.5 hrs (2 at a time)
iMac G5 @ 2 Ghz = 3 hrs
eMac G4 @ 800 Mhz = 10.5 hrs
iMac G3 @ 500 Mhz = 15.25 hrs
____________

I takes 6:10 hours to crunch(E@H)with mi AMD Athlon(tm) XP 2800+,S@H takes 1.2h or 2 hours
____________
Son of a BIT!

AMD X2 3800+ at stock speed with old 4x512MB DDR266CL2
About 22k per WU using both cores.
____________

iMac G5 1.6 ghz - 5 hours 41 minutes average
Pentium 4 3.4 ghz - 11 hrs 16 minutes per thread. (2 threads) 5hrs 38 min average
Pentium D840 dual core 0 7 hrs 22 min per core or 3 hrs 41 min average.
AMD 64 X2 4200 - 5 hrs 42 min per core or 2 hrs 51 min average.
Athlon 2400 - 6 hrs 25 min average
Athlon 2800 (previous because Motherboard just died) - 6 hrs 7 min average
(looking for a replacement mobo then will upgrade to Athlon 3200)
iBook 600, 600mhz G3 - about 27 hours average.

____________

Team MacNN - The best Macintosh team ever.

iBook G4 1.33 GHz, 7.7 hours aaverage
PowerBook G4 1.67 GHZ, 6.3 hours average
____________
[/url]

Join Team MacNN

Mac Mini (G4 1.25GHz) - 8 hours, 8 min avg

iBook (G3 900MHz) - all results are gone, haven't used in a while; guessing 20 hours?

AMD 3700+ "San Diego", WinXP - 5 hours, 37 min avg

This last one is my new "cruncher", not yet overclocked, only a couple of results to get a baseline.

____________

PowerMac G5 2.0 GHz ... looks like 4.32 hours for most of them. Used to be 7 to 10 hours. I was not paying attention and had not noticed the speed increase ...
____________

PowerMac G5 2.0 GHz ... looks like 4.32 hours for most of them. Used to be 7 to 10 hours. I was not paying attention and had not noticed the speed increase ...

> And I should not have said that hyperthreading doubled the throughput. I have
> never run it with hyperthreading turned off & expect it would look quite
> bad.

It does not look that bad. HT does not double the performance. It does give a boost to throughput between 10% to as high as 70% more work done per unit time.

We did fairly extensive tests during the late stages of the Beta test program with HT on and off and it does surprisingly well at improving the total production. And I say surprising in that our usage of the processors is very atypical ... we heavily use the math portions and do not have a very "standard" mix of instruction streams.

I mean, my G5 does more work because of the two CPUs than does any of my HT systems, but it is a cheap way to "see" two CPUs.

What is your opinion: I have a choice in the next couple of days to get either a P4 3.4 Gh with HT, or an AMD 64 3800 computer. Which would you choose?

What is your opinion: I have a choice in the next couple of days to get either a P4 3.4 Gh with HT, or an AMD 64 3800 computer. Which would you choose?

Those are the same two options I was looking at... but then I happened to read about the newer "San Diego" core AMD 64 3700+ (socket 939). Chip cost was less than the 3800 or the P4, while cache is larger (more important to SETI than Einstein, but still nice to have) and every indication is that it will overclock to be faster than the 3800 while still running cooler. HT doesn't mean much to me for the other purposes this computer will have, so CPU choice was based on BOINC.

The P4 seems to take 20-21K seconds per WU (40-42K but doing 2 at a time). Without overclocking, this is exactly what the 3700+ is giving me, 20-21Ksec/single WU. The 3800 might be very slightly faster "stock" for Einstein, but I believe it's slightly slower for SETI. Overclocking on the P4 is limited by heat; the 3800 by chip design. The 3700 seems to have much looser limits. I don't know yet how far I'll be able to push the 3700, but others have gotten into the 4000 or even 4200 range, with fans, nothing fancy like liquid cooling. Not being a PC guru, I bought an ASUS motherboard that has "overclocking for dummies" built in, and a case with extra fans - "stock", my CPU is running at 40-41 degrees, so there's a lot of room left to play - I'll hopefully know more in a few days.

Hm... just realized I didn't answer your question, but instead added a third option to the mix!

The only thing is, this computer is for me at work, so I may need to be a bit careful with how much I can play around with it in terms of overclocking.

You (and everyone else here probably, I'm a Mac guy) I'm sure know more about overclocking than I do. From what I've read about 3700 "San Diego"s though, 10% (2.42 GHz) should barely be pushing it. The limit on air seems to be around 30% in fact... although I value accuracy and CPU lifetime too much to push it that far.

All this effort to shave a few seconds off our crunching times... is this a hobby or an obsession? :-) I'll have at least one or two result times at 105% to report back soon.

What is your opinion: I have a choice in the next couple of days to get either a P4 3.4 Gh with HT, or an AMD 64 3800 computer. Which would you choose?

Take a look again at my results:
Pentium 4 3.4 ghz - 11 hrs 16 minutes per thread. (2 threads) 5hrs 38 min average
AMD 64 X2 4200 - 5 hrs 42 min per core or 2 hrs 51 min average.
Athlon 2400 - 6 hrs 25 min average.
Get the AMD if Einstein is your primary project.

For some reason the work computer was delivered as an AMD Athelon 64 3500+. I will know Monday, but I suspect the guy ordering it saw a way to shave off some cost from the 3800+ system :( [I left him about 5 voicemails telling him I was sure it was HP's fault, they had shipped the wrong computer, and that HP should let me use the 3500+ until the correct 3800+ system arrived!]

Anyway, here is the timing (which is nonetheless about 3X better than my old P4 1.4Ghz, which I now know was basically a piece of crap. From now on, it is AMD for me!)

AMD Athelon 64 (Winchester core) 3500+ processor: 20,300 secs. (5.6 hours)
____________
Regards,
Bob P.

AMD Athlon 64 3700+ (San Diego Core) barely overclocked to 105% - 5h 26m (19,553s) avg.

____________

Avg WU time ~ 5 hr, 15 min
Athlon XP-M 2600 Mobile installed in
Asus A7N8X-deluxe 2.0 mobo
512mb Corsair 3200LL ram @ 7-3-3-2.5 timing
2.448 GHz (204FSB x 12) 1.625 core V, easy2do - this 45 watt (@ default speed) CPU has a huge thermal and voltage envelope
non-exotic aircooling (TT Volcano12)

Built in Dec '03, ran Seti Classic 24/7 until August '04, this $US 95 CPU has been running Einstein day & night, often in 85+ degree F environment since March - rock-stable at these settings, as well as concurrent web duty and gaming. Could probably go higher but I think my 350 Watt PSU is the weak link, too broke to upgrade.
____________
microcraft
"The arc of history is long, but it bends toward justice" - MLK

Looks like most of you guys had your WU time cut down consiberably. Last week my laptop speed jumped from 10 hours to nearly 20!!!

HEY!! wait a minute!!.... What did you guys do to me?

Actually, I think something is broken (after nearly 4 solid months of BOINCing). Laptop won't recognize my power adapter anymore, so it won't charge the battery and it cut my processor down to half speed.

____________


"No, I'm not a scientist... but I did stay at a Holiday Inn Express."

AMD Athlon XP 2600+ 2.83 GHz average time per wu is 22,000 seconds or about 6.11 hrs. So far so good. :)
____________
One day man may become.

Quad PIII Xeon, 4 matched 500/100/2048's, 1Gig 50 nanosecond EDO ECC buffered dimms. Yup, it's a slug at 32-33 hours per WU. Only saving grace is that it does 4 WU's at a time and it is very stable.

See ya
____________

Hi

Athlon XP 2800 @ 2120 Mhz Win XP 512 MB Ram

WU times between 22000 and 25000 seconds.

greetz Mike

____________

PentM 1.86GHz - 29,000 to 30,000 sec

Dual P3 933MHz - 63,000 to 64,000 sec

Andy

Mac G4 Dual CPU 1.4GHz - Longest 27,170.41 - Shortest 13,074.32 But most are around 25,600
Mac G4 PowerBook 1.4GHz - Longest 41,673.63 - Shortest 36,700.31 But most are around 36,600

Regards
Phil

____________

We Must look for intelligent life on other planets as,
it is becoming increasingly apparent we will not find any on our own.

CPU-type: AuthenticAMD AMD Athlon(tm) 64 Processor 3200+
OS: Linux 2.6.8-24.18-default
Memory: 512Mb Corsair DDR400
Last two WU's took 28,023.09 and 28,120.62 seconds.

Bob

CPU-Type : AuthenticAMD AMD Athlon(tm) 64 Processor 3500+
Memory : 1 GB

OS : Linux 2.6.13-kami64-32
Application : Einstein 0.15 for Linux

Average over the last 19 WU's : 24850 secs
Min 24718
Max 2530

(System mostly available for E@H )
____________

Hi,
I'm running an Athlon (really K7 Thunderbird) at 1.364 ghz on an A7V mb.
I have 512meg memory and use Win98SE. My WU times run between 34,800
and 36,000 seconds.
I use 4.19 because I believe that the overhead associated with 4.45 uses
1-3% of the CPU cycles. I hope that the new versions absolutely minimize
overhead.


____________
Joe B

OK, here are some speeds from my systems.

einstein_4.81_i686-pc-linux-gnu

AMD Sempron(tm) 2500+ cpu MHz : 1749.890 cache size : 256 KB

32100-32300 seconds (~9 hours)

AMD Athlon(tm) 64 Processor 3000+ (socket 754, PC2700 RAM) cpu MHz : 2010 cache size : 512 KB

26800-27100 seconds (~7,5 hours)

____________

Here are my two machines, they are both running xp:

Athlon 64 3700+ running slight overclock at 2.45 gig(stock is 2.4) 1mg cache socket 754, 1 gig corsair pc3200:
times a little over 19,000 for it, takes a bit over 5hrs per wu.

The other machine is a 2500+ barton core chip, running 512 meg of corsair pc3200:
the time for it's first wu is 25,128.39!!!

I don't think that's too bad for a lowly 2500+ athlon xp!!!
____________

AMD Athlon(tm) 64 X2 Dual Core Processor 4400+


2 * 24500 seconds (~6,8 hours)
____________

AMD Athlon 64 3700+ (San Diego Core) barely overclocked to 105% - 5h 26m (19,553s) avg.

PowerMac G5 Dual 2.7GHz, MacOS X 10.4.2:

SETI@Home: appr. 1h 15 min.
E@H: appr. 3h 45 min.

Measured floating point speed: 4934.05 million ops/sec
Measured integer speed: 18156.87 million ops/sec


____________

Any suggestions on how to speed it up? I haven't seen any optimized clients for E@H.
____________

Any suggestions on how to speed it up? I haven't seen any optimized clients for E@H.

Thanks for the input Michael,
I really don't want to overclock this 2500+, it has a stock cooling solution, and i don't know how long I will have it. I have, and do overclock, my 3700+ has a mild overclock on it, it has a Zalman cooler on it, and I am running a coolergiant ps. I am running an asus KV8-SE deluxe with a via chipset on it, it doesn't have that much in the of options to overclock in the bios, and this chip won't take much, so in the interest of stability and accuracy I only have it running at 2.45g.
I really enjoy overclocking, just to see how fast I can make them go without them losing their smoke!!! I just don't think this is the app to trade stability for speed, the science is the first consideration here.
I hope to put together at least a 4400 x2 next and be able to run it 24/7 crunching numbers, I really would like to be able to put together a dual proc opteron!!! I can't run my machine 24/7 right now, so I just try to maximize the time I can run them.
Thanks for the suggestions again, and HAPPY CRUNCHING EVERBODY!!!!!

http://www.boincstats.com/signature/user_235410.gif
____________

AMD would set the multiplier higher eg 12 x 166.66 = 2.0 gHz and they would call it a XP2700+ (I think) or something like that. When they want to produce a 3200+ it's usually done by running at 200FSB with the multiplier back at 11. So if you are lucky enough to have a 2500+ which has just been marked at that because AMD needed the most popular speed, why not run it at 3200+ speed and see if it works.


Hey Gary,
Got up this morn and said, what the heck, brought the fsb up to 200, it fired right up no probs so far!!! and showed it as being an xp3200+, then i brought the multiplier up to 12, that didn't seem to do anything, still showed 3200+ @ 2.2, guess we'll see how it runs......I wonder if the multiplier is locked down to 11? You have any knowledge of that? If it runs ok here I will try bumping the fsb up some more. It is a dfi board with an nf2 chipset, think it will take a lot of over clock.
I run E@H and S@H on this machine, think I will bring down an optimized client for seti and use that for testing, should give me some feedback a lot quicker on if it is returning valid results. I'll let you know what happens.

Happy Crunching,
Bill
____________

Ran just fine with the 200 fsb, ran 20,901.42 sec for the last wu, thanks for the tip!!! BTW me 3700+ does wu's in hight 18,000's to mid 19,000's, not that much faster!!!

Bill
____________

Ran just fine with the 200 fsb, ran 20,901.42 sec for the last wu, thanks for the tip!!! BTW me 3700+ does wu's in hight 18,000's to mid 19,000's, not that much faster!!!

Bill

LHC@Home, an exact match is required ... is a true statement. THis is why there is no OS-X version ... also, they have had some off and on problems with some chips against other chips ... VERY little tolerance for errors ...
____________

Hey guys,
Thanks for the info on that barton core. I ran an athlon 64 3000+ last year and got a lot of overclock on it, my 3700+ doesn't seem to want to overclock much, have it running at 2.45 instead of 2.4, had no idea how easy and how much you could get out of the barton. Neither one of those chips showed up as anything diff in windows, their speeds were faster, but the chip designation never changed. With the barton being recognized as a 3200+ now, guess that's what the chip "really" is? I fiddled with the fsb a little bit more yesterday, but to no avail. I really don't feel like messing with the voltages, I did bump voltages when running up my 3000+, both cpu and mem, I have a feeling any more increases from here will be incremental, not as drastic as what I have already gotten out of it.
Thanks again you Gary for the info on that chip. The machine is now doing about 4 wu's a day instead of @3!! May be insignificant numbers when looking at the entire project, but every wu crunched helps. I just wish I were able to run my machines 24/7 :-(
I have read it else where, and I will repeat it, too bad we can't take advantage of the processing power of our video cards. My 3700+ machine used to play games, it has a 6800gt OC with 256mg of ddr3 on it, and I have been able to overclock it quite a bit. Now that card just sets there idle, no games anymore, machine just crunches. I do realize that we can't use that power, but what a shame to have a $400 video card just setting there idle.

Bill
____________

Not exactly a WU speed, but closer to this topic than to Milestones. I've just last night finished crunching 24 WUs in 5 days (12/10-12/14), on a single Barton-core Athlon.
____________
microcraft
"The arc of history is long, but it bends toward justice" - MLK

This "Best 3150" laptop from 1999 had reached the end of its useful life in production use. I replaced the defunct Windows 98 with Debian and now the machine serves as a surfboard@home and crunches for Einstein@home. Recent average credit usually varies between 30-40. CPU time per WU is 160...165 ks and claimed credit about 90.

Very little variation from ~ 8hrs 20 min on EAH wu on my P4 2.8 ghz 512kb L2 with the einstein 4.79.
____________

It did average 42,000sec, then I did a defrag & reinstall everthing but Office Pro, onto my computer. Now my Average time is 9,100-8,800 sec.

Running Setie @ the same time @ 50%
Running Einstein @ 66.67% (using VR RAM)
P4 3ghz 2gig RAM + 1064gig VR-RAM
XP Pro
Laptop


____________
I want pancakes! God, do you people understand every language except English? Yo quiero pancakes! Donnez moi pancakes! Click click bloody click pancakes!
-Stewie Griffin, Family Guy

I would like to share comparisons of two Athlon XP 2800's.

host 422996 runs with Window XP Home and on Einstein averaged 22239.78 sec, 69.93 claimed credit and 71.74 granted credit.

host 443142 runs Linspire linux 2.6.10 and on Einstein averaged 28039.41 sec, 88.93 claimed and 76.98 granted.

Windows took 6.19 hours and Linux took 7.79 hours and these were my best Einstein crunchers.

But now with Albert I have this

422996 does 14557.01 sec, 50.65 claimed, 46.13 granted

443142 does 5335.54 sec, 16.94 claimed, 15.63 granted

The alberts with Linux are taking are taking 37 percent of the time of the Windows ones and getting 34 percent of the credit of the Windows machine.

Both machines have BOINC 5.2.13 but I also use optimized BOINC's by Crunch3r and both machines do Seti@home also. The linux machine does LHC when it's available but the Windows machine could never get a valid LHC result. I don't know whether to blame the Kingston ram I put in it or the refurb Gigabyte mobo.

Regards,

Pam





____________

But now with Albert I have this

422996 does 14557.01 sec, 50.65 claimed, 46.13 granted

443142 does 5335.54 sec, 16.94 claimed, 15.63 granted

The alberts with Linux are taking are taking 37 percent of the time of the Windows ones and getting 34 percent of the credit of the Windows machine.

Both machines have BOINC 5.2.13 but I also use optimized BOINC's by Crunch3r and both machines do Seti@home also. The linux machine does LHC when it's available but the Windows machine could never get a valid LHC result. I don't know whether to blame the Kingston ram I put in it or the refurb Gigabyte mobo.

Regards,

Pam
>Hi, Pam.....it's possible you are comparing apples and oranges with the 'Alberts'. There is an variety of these new WUs and they don't have consistant processing times like the old 'Einstein' WUs. They are all called 'Alberts' but the low frequency versions complete in approximately 30% of the time of the 'Einsteins' and the higher frequency ones closer to 'Einstein' times. It's possible one machine (443142) has the faster WU and the other the ones that take more processing. Re. the Kingston RAM: I've had no problems with that brand. Hope this helps....Cheers, Rog.



[/quote]

____________

>Hi, Pam.....it's possible you are comparing apples and oranges with the 'Alberts'. There is an variety of these new WUs and they don't have consistant processing times like the old 'Einstein' WUs. They are all called 'Alberts' but the low frequency versions complete in approximately 30% of the time of the 'Einsteins' and the higher frequency ones closer to 'Einstein' times. It's possible one machine (443142) has the faster WU and the other the ones that take more processing. Re. the Kingston RAM: I've had no problems with that brand. Hope this helps....Cheers, Rog.

[quote]
I remembered that there were different processing times but I forgot a host gets work units from the same set of data until that set is all sent out. Or something like that. I guess eventually, the two machines could switch types of Alberts.

I remembered that there were different processing times but I forgot a host gets work units from the same set of data until that set is all sent out. Or something like that. I guess eventually, the two machines could switch types of Alberts.

On my dual P3 800 the processing times for the new albert app have been in the range of 4-4.25 hours. With the einstein app processing times were just over 19 hours.

____________
Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)

hrm..had my first data glitch on EAH and it's on the albert app

1/4/2006 3:30:32 PM|Einstein@Home|Result r1_0770.0__432_S4R2a_1 exited with zero status but no 'finished' file
1/4/2006 3:30:32 PM|Einstein@Home|If this happens repeatedly you may need to reset the project.
1/4/2006 3:30:32 PM||request_reschedule_cpus: process exited
1/4/2006 3:30:32 PM|Einstein@Home|Restarting result r1_0770.0__432_S4R2a_1 using albert version 437

we'll see if it's a random, or a continuing problem.
Crunching times seem stable at 5 hrs 12 - 15 min on a P4 Prescott, 512L2, 2.8ghz 400 ddr.
____________

another glitch
Granted, I have not done a massive amount of EAH WU, but the first two I've ever had came after Albert and on the heals of each other.

1/4/2006 8:29:18 PM|Einstein@Home|Result r1_0770.0__431_S4R2a_0 exited with zero status but no 'finished' file
1/4/2006 8:29:18 PM|Einstein@Home|If this happens repeatedly you may need to reset the project.
1/4/2006 8:29:18 PM||request_reschedule_cpus: process exited
1/4/2006 8:29:18 PM|Einstein@Home|Restarting result r1_0770.0__431_S4R2a_0 using albert version 437

____________

NIMRUTH: You do not have enought ram to support the project or your CPU/OS. I consider this a help thread. I have at least 1 Sempron 3100 running at 2.4ghz. Check the forums. There is a huge amount of info out there on optimizing your system. 256 meg of ram at your rated speed is crippling your rig. If your mobo does not allow you to adjust HTT and change ram dividers, at least get a stick of 512.

Regards-tweakster


____________