Crunching Wet and Dry

[Abarbarian]

So would you like to post your CPU bench marks for your riggs along with cpu type and model and operating system .
I thought it might be of interest to see how different setups perform .

Asgard
Xp Pro 32 Bit
Athalon 64 3400 Clawhammer

2174 Floating point MIPS Whetstone per cpu
3972 Integer point MIPS Dhrystone per cpu


XXXX
Xp Pro 32 Bit
Athalon 64 X2 4800+ Brisbane Dual Core

2528 Floating point MIPS Whetstone per cpu
4671 Integer point MIPS Dhrystone per cpu


XXXX
Gutsy Gibbon 7.10 64 Bit
Athalon 64 X2 4800+ Brisbane Dual Core

2211 Floating point MIPS Whetstone per cpu
6764 Integer point MIPS Dhrystone per cpu


Haven't a clue what whetstone and dhrystone are all about I thought that they were descriptions of the walls here in the lakes .

 

[Spezi]

AMD Athlon 64 FX-60 Dual Core
2921 floating point MIPS (Whetstone) per CPU
5339 integer MIPS (Dhrystone) per CPU

AMD Athlon 64X2 4400+
2254 floating point MIPS (Whetstone) per CPU
4131 integer MIPS (Dhrystone) per CPU

AMD Athlon 64X2 3800+
1981 floating point MIPS (Whetstone) per CPU
3694 integer MIPS (Dhrystone) per CPU

AMD Turion 64X2 Mobile (Two identical units like this one crunching)
1579 floating point MIPS (Whetstone) per CPU
2919 integer MIPS (Dhrystone) per CPU

My other crunchers are not immediately accessible as they reside elsewhere.

These are the main work horses.

All run XP PRO 32 Bit at the moment.

 

[Abarbarian]

Hi Spezi love your avatar gif . Those marks you posted are very interesting. I'm suprised that my old Clawhammer is getting better results than your X2 3800+ . If all the folks post there could be some real eye openers in store .

 

[Taffycat]

Here are mine... I've often wondered about Whetstone and Dhrystone too

AMD Athlon 64 X2 5200+ Dual Core (Vista)

2610 floating point MIPS (Whetstone) per CPU
4457 integer MIPS (Dhrystone) per CPU


AMD Athlon 64 4000+ Single Core (XP-Pro SP2)

2372 flating point MIPS (Whetstone) per CPU
4124 integer MIPS (Dhrystone) per CPU

 

[Abarbarian]

Hi Taffycat love your avatar too .
I'm not suprised that your 5200 is better on the Whetstone but I am suprised that my 4800 is better on the Dhrystone in XP .
Interesting results arriving thick and fast .

 

[crazylegs]

Point me to the benchmarking software Abe..I don't normally do any of this benchmarking palava..

As long as the pc boots thats my benchmark..

 

[Spezi]

Hi Spezi love your avatar gif . Those marks you posted are very interesting. I'm suprised that my old Clawhammer is getting better results than your X2 3800+ . If all the folks post there could be some real eye openers in store .

Wonder if it might have something to do with that unit not currently having a matched pair of memory sticks?

It has a generic 1GB stick along with another stick of 512MB so I'm definitely not benefitting from the double data rate.

I pilfered the two matched Corsair 1GB RAM sticks that were in that machine and put them in the 4400+ unit figuring I'd get more bang for my buck there though I haven't gotten around to overclocking that baby.

 

[Adywebb]

Point me to the benchmarking software Abe..I don't normally do any of this benchmarking palava..

As long as the pc boots thats my benchmark..

Its the benchmark of the BOINC software we crunchers use

Here are mine:

5 x Intel Q6600's @3.4GHz (XP Pro SP2)
3355 Floating point MIPS Whetstone per cpu
7432 Integer point MIPS Dhrystone per cpu

5 x Intel QX6700's @3GHz (XP Pro SP2)
2954 Floating point MIPS Whetstone per cpu
6441 Integer point MIPS Dhrystone per cpu

Intel Xeon X5340 (x2) @3GHz (Windows Server 2003 64-bit)
2869 Floating point MIPS Whetstone per cpu
8905 Integer point MIPS Dhrystone per cpu

 

[vaio]

Ram timings and latency probably impact numbers to some extent.

My 6300 Allendale: 468 x 7

12/03/2008 10:22:21|| 3207 floating point MIPS (Whetstone) per CPU
12/03/2008 10:22:21|| 7278 integer MIPS (Dhrystone) per CPU


X2 3800+ Manchester: 250 x 10

2547
4683.

 

[nivrip]

Its the benchmark of the BOINC software we crunchers use

Where do we find it?

And, more to the point, what does it all mean?

 

[Abarbarian]

Ram timings and latency probably impact numbers to some extent.

My 6300 Allendale: 468 x 7

12/03/2008 10:22:21|| 3207 floating point MIPS (Whetstone) per CPU
12/03/2008 10:22:21|| 7278 integer MIPS (Dhrystone) per CPU


X2 3800+ Manchester: 250 x 10

2547
4683.

Them results for the 3800 are impresive , what OS are you using ??

 

[Taffycat]

Where do we find it?

And, more to the point, what does it all mean?

Hi Niv, it's to be found here: Open BOINC, click the "Advanced" tab and then click "Run CPU Benchmarks"

 

[Abarbarian]

Where do we find it?

And, more to the point, what does it all mean?

Open Boinc in advanced mode , click on the advanced tab on the top toolbar and you will see a CPU Benchmark , click on it and the results will show in the main window .

The benchmarking will tell us which set up will give the best results for crunching . For my rigg Linux is showing better results than XP and is showing better results than some riggs with much better cpu's .

What I am trying to show is that a rigg dedicated to crunching will give much better results on linux than on windows . If I am right and the results might show this then linux is a greener option for crunching as you get more points generated for the power used .

Now if a technonumptie like me can get linux up and running with Boinc then anyone can .

So if my idea is right then if everyone here swapped over their dedicated crunchers to linux we could really kick some ass in the bunny race .

Oooops I mean that we could generate better results for the good of mankind . An kick some ass .

 

[Abarbarian]

Its the benchmark of the BOINC software we crunchers use

Here are mine:

5 x Intel Q6600's @3.4GHz (XP Pro SP2)
3355 Floating point MIPS Whetstone per cpu
7432 Integer point MIPS Dhrystone per cpu

5 x Intel QX6700's @3GHz (XP Pro SP2)
2954 Floating point MIPS Whetstone per cpu
6441 Integer point MIPS Dhrystone per cpu

Intel Xeon X5340 (x2) @3GHz (Windows Server 2003 64-bit)
2869 Floating point MIPS Whetstone per cpu
8905 Integer point MIPS Dhrystone per cpu

How come my 4800 is out performing your QX6700 ?? That is some suprising result .

XXXX
Gutsy Gibbon 7.10 64 Bit
Athalon 64 X2 4800+ Brisbane Dual Core

2211 Floating point MIPS Whetstone per cpu
6764 Integer point MIPS Dhrystone per cpu


Do we have a guru who could tell us the relevence of the Whetstone and Dhrystone tests . i.e is it better to have higher Wet than Dry etc etc .

 

[muckshifter]

Where do we find it?

And, more to the point, what does it all mean?

It means very little ...

I can't remember what CPU I have.
12/03/2008 11:45:39 am|| Number of CPUs: 2
12/03/2008 11:45:39 am|| 2032 floating point MIPS (Whetstone) per CPU
12/03/2008 11:45:39 am|| 3426 integer MIPS (Dhrystone) per CPU

... ah, found it; an x2 3800 2gig ram Vista

 

[Abarbarian]

x2 3800 2gig ram Vista Number of CPUs
2032 floating point MIPS (Whetstone) per CPU
3426 integer MIPS (Dhrystone) per CPU


AMD Athlon 64X2 3800+ XP Pro 32 bit
1981 floating point MIPS (Whetstone) per CPU
3694 integer MIPS (Dhrystone) per CPU


X2 3800+ Manchester: 250 x 10 Unknown os
2547 Floating point MIPS (Whetstone) per CPU
4683. integer MIPS (Dhrystone) per CPU

Same cpu totaly different results . In a dedicated cruncher running 24/7 that would make quite a difference . I think a 1000 point difference is considerable thats around 27% certainly not to be sniffed at .

I'm guessing that the bottom set of results for the unknown os are produced on a pc with linux .

 

[nivrip]

technonumptie like me can get linux up and running

Surely, a contradiction in terms, LOL.

 

[Adywebb]

How come my 4800 is out performing your QX6700 ?? That is some suprising result .

XXXX
Gutsy Gibbon 7.10 64 Bit
Athalon 64 X2 4800+ Brisbane Dual Core

2211 Floating point MIPS Whetstone per cpu
6764 Integer point MIPS Dhrystone per cpu

Its not surprising as it has nothing to do with the CPU, its because you are using a 64 bit OS - they always seem to produce higher Integer marks, in the same way my Windows Server 2003 64 bit rig does:

Intel Xeon X5340 (x2) @3GHz (Windows Server 2003 64-bit)
2869 Floating point MIPS Whetstone per cpu
8905 Integer point MIPS Dhrystone per cpu

When I was using Ubuntu 64bit on one of my Q6600 I was getting over 14000 for the Integer mark compared to over 7000 with XP

Makes no difference really, as the quorum reduces the claim anyway

What I am trying to show is that a rigg dedicated to crunching will give much better results on linux than on windows . If I am right and the results might show this then linux is a greener option for crunching as you get more points generated for the power used .

As indicated above, the benchmark is almost irrelevant to returns in the real world as the claim will be reduced within the quorum.

However Linux can crunch units slightly faster in some circumstances, but its nowhere near the '27%' figure you are going on in relation to the benchmark.............to be honest you need to get away from thinking that a higher benchmark on the same CPU with different OS's means that one CPU is crunching units faster than the other.

The BOINC benchmark is really purely artificial now and has little relation to how many points you get with most crunching projects these days.

 

[Abarbarian]

Its not surprising as it has nothing to do with the CPU, its because you are using a 64 bit OS - they always seem to produce higher Integer marks, in the same way my Windows Server 2003 64 bit rig does:

Intel Xeon X5340 (x2) @3GHz (Windows Server 2003 64-bit)
2869 Floating point MIPS Whetstone per cpu
8905 Integer point MIPS Dhrystone per cpu

When I was using Ubuntu 64bit on one of my Q6600 I was getting over 14000 for the Integer mark compared to over 7000 with XP

Makes no difference really, as the quorum reduces the claim anyway


As indicated above, the benchmark is almost irrelevant to returns in the real world as the claim will be reduced within the quorum.

However Linux can crunch units slightly faster in some circumstances, but its nowhere near the '27%' figure you are going on in relation to the benchmark.............to be honest you need to get away from thinking that a higher benchmark on the same CPU with different OS's means that one CPU is crunching units faster than the other.

The BOINC benchmark is really purely artificial now and has little relation to how many points you get with most crunching projects these days.

Ah , now if I'd had that explanation when I posted me question several days ago . I wouldn't have wasted so much time on this linux project .

https://www.pcreview.co.uk/forums/thread-3439855.php

 

[Adywebb]

You misunderstand me - 64bit Linux WILL give you more points.....but nowhere near the benchmark increase of 27%.

In my comparisons using Ubuntu 64bit it seemed to crunch units slightly faster, and if you are lucky enough to be matched in the quorum with another using the same you WILL get more points.....but sadly it won't happen very often.

You CAN expect to get an increase, but it won't be that much is all I'm saying - but it should be noticeable I hope..........and anyway its good to experience other OS's