Science Mark 2 Scores..please post here guys.

[d3birth]

Still pretty fast XD

[Aguanteboca]

Gonna get XP64 today and see what I score. If this is an indication of what 64bit does to performance...I CANT WAIT!

[redpriest]

Hi, and thanks for using our benchmark =)

We've tried hard to incorporate all of the latest technologies, including dual core and hyperthreading. If you are not running ScienceMark 2.0 with at least hyperthreading, give it a try. Also, our 64-bit optimizations help Intel EM64T processors as much as they do the Athlon 64. Give it a shot =)

Alex Goodrich
sciencemark.org

[agenda2005]

Hi, and thanks for using our benchmark =)

We've tried hard to incorporate all of the latest technologies, including dual core and hyperthreading. If you are not running ScienceMark 2.0 with at least hyperthreading, give it a try. Also, our 64-bit optimizations help Intel EM64T processors as much as they do the Athlon 64. Give it a shot =)

Alex Goodrich
sciencemark.org

redpriest welcome to Xtreme systems!

Nice to read from the horses mouth. I'm just curios to know the compiler you use for the new sciencemark revision.
Have you optimized Primordia for SSE3 as some upcoming AMD CPU's will also feature SSE3. I don't think Mol. Dyn. will benefit from SSE3, correct me if I'm wrong, just like to know anyway.

[redpriest]

Thanks agenda2005 =)

Primordia is optimized for Hyperthreading and SSE2. It does not take advantage of SSE3 as of yet. I haven't fully evaluated whether there would be a noticable speed-up there as of yet, but there are some small areas where it could benefit.

The new compiler we use for the 64-bit version of sciencemark comes from the Platform SDK RC2 for Win64 systems, which is based on the "Whidbey 2005" MSVC code.

MolDyn won't benefit from SSE3 unfortunately, as there are no applicable situations where we could take advantage of it. Intel pretty much covered everything that we needed to do with SSE2 instructions.

BLAS will in the future take advantage of SSE3 as the horizontal components of those instructions will help us nicely. We are working on a complete overhaul of the BLAS area.

Some things we've identified that can be optimized:

AES can be optimized for hyperthreading/dual core. I anticipate extreme speedups once established.
BLAS optimized for Intel and AMD vendor libraries.
Primordia - more hyperthreading/dual core opportunities.

Hope that answers your questions =)

[Flak Monkey]

All under 64-bit windows on a 754 3200+ clawhammer @stock 2.0ghz and with lots of programs running in background (Firefox, Steam, D2OL, Trillian, ect..)

Molecular Dynamics
32bit = 113.09382sec
64bit = 42.25440sec
Improvement = 62.64%

Primordia
32bit = 434.4312366290sec
64bit = 411.5966079750sec
Improvement = 5.27%

[Sheik]

Hi redpriest

Warm welcome to XS. This is an honour.

The 64 Bit version is a great show for 64 Bit power.

Keep up the good work.

[Matias]

Hi all, here are my results:

@Stock
Mol Dyn 70.18sec
Primordia 383.91sec
Cipher 11.61sec

@265x10
Mol Dyn 63.62sec
Primordia 342.90sec
Cipher 10.40sec

[Aguanteboca]

RedPriest, thaks for joining us. I love your benchmark and recommend it to everyone for CPU testing. It helps me a lot to tweak my system. Based on the results here, your benchmark alone has me dying to get XP 64!

Matias, que haces che! Good to see you posting man.

[Aguanteboca]

I guess the scores are normal except the Cipher which is too fast compared too your other scores. Did you get final verison? It has an automatic bench all option. Is better if we all run that again.

[Matias]

CPU @ 2650Mhz(10x265, RAM on 166 divider @ 220Mhz 2.5,3,3,6 1T). I have no idea what I'm running here, or wether these scores are any good:

MolDyn - 64.94s
Primordia - 319.48s
Cipher - 9.93

I think they are pretty good numbers. I'm running @265x10 1:1 and you get better Primordia & Cipher scores.

[Matias]

This was done with WinXP 64bits v.1289 @stock

Mod Dyn 30.18sec
Primordia 354.62sec.

[Jeff]

Thanks agenda2005 =)

Primordia is optimized for Hyperthreading and SSE2. It does not take advantage of SSE3 as of yet. I haven't fully evaluated whether there would be a noticable speed-up there as of yet, but there are some small areas where it could benefit.

The new compiler we use for the 64-bit version of sciencemark comes from the Platform SDK RC2 for Win64 systems, which is based on the "Whidbey 2005" MSVC code.

MolDyn won't benefit from SSE3 unfortunately, as there are no applicable situations where we could take advantage of it. Intel pretty much covered everything that we needed to do with SSE2 instructions.

BLAS will in the future take advantage of SSE3 as the horizontal components of those instructions will help us nicely. We are working on a complete overhaul of the BLAS area.

Some things we've identified that can be optimized:

AES can be optimized for hyperthreading/dual core. I anticipate extreme speedups once established.
BLAS optimized for Intel and AMD vendor libraries.
Primordia - more hyperthreading/dual core opportunities.

Hope that answers your questions =)

Noticed I get only 25% to 50% CPU utilization when running on a dual Xeon w/HT on setup. Mostly sits at 25% with bursts of 45%-50% in there once in a while.

Can't wait for "more hyperthreading/dual core opportunities".

Nice benchy!

[Aguanteboca]

here are some venice 3800+ result's..@ 3.6ghz.


lowest primordia yet.

Awesome Promedia but why is your Moldyn at 3.6 slower than mine @ 2744? How are you hitting 3.6 with that Venice...more details please as that some awesome clock..

[Sheik]

Ok my new scores

64 Bit version.

Moldyn = 21.63 seconds
Primordia = 231 seconds

Who's your daddy

[agenda2005]

32-bit Primordia for A64 3800 @ 3600 MHz does 244sec
64-bit Primordia for opty 252 @ 2600 MHz does 231sec (1000MHz lower clock)

64-bit POWA!!!

[Sheik]

32-bit Primordia for A64 3800 @ 3600 MHz does 244sec
64-bit Primordia for opty 252 @ 2600 MHz does 231sec (1000MHz lower clock)

64-bit POWA!!!

Dual Opty 252's + 64 Bit POWA

Molecular Dynamics is lightly multithreaded - one thread takes care of U/I aspects, and the other thread takes care of the computation.

Primordia is multithreaded.

Source

[c7775]

Hi all!
I think Science Mark 2 is one of the best and most accurate benchmarking tools available. Lets see some scores for Moldyn, Chiper, Primordia and Memory please.


Regards

[saaya]

huh?

[Gorod]

Venice 3000+ @ 3006MHz @ 1.65V , 334x9.0 , HTT x3
Vapochill LS
MSI K8N Neo2 Platinum , nForce3 Ultra
2x512Mb , 215MHz @ 2.0-2-2-8-1T

Molecular : 55.86
Primordia : 278.47
Cipher : 8.98

[Charles Wirth]

Machine specs below.

[Highland3r]

Dothan as per sig. Not bad

Mol Dyn: 82.78
Primordia: 389.44
Cyper: 13.2

[the]

system in sig, all stock clock 64bit
moldyn 1833.40 41.31659 secs
Primordia 41.06 392.63825 secs

[agenda2005]

System : Air Cooled A64 3700+ at 2860MHz (260 X 11) 2.5-3-3-6
OS : Windows Server 2003

Best Molecular Dynamics so far:
Molecular Dynamics : 20.09sec
Primordia :283.90sec

MOLECUALR DYNAMICS

<img src="http://216.77.188.54/coDataImages/p/Groups/253/253565/folders/204926/1563106MolDyn.jpg" width="1152" height="864">


PRIMORDIA

<img src="http://216.77.188.54/coDataImages/p/Groups/253/253565/folders/204926/1563105Primordia.jpg" width="1152" height="864">

[agenda2005]

Broke my personal record of molecular dynamic. 20sec barrier now broken!
Molecular dynamics time = 19.77355sec

<img src="http://216.77.188.54/coDataImages/p/Groups/253/253565/folders/204926/1578591Mol-dyn-2.jpg" width="1152" height="864">

Super Pi SS

<img src="http://216.77.188.54/coDataImages/p/Groups/253/253565/folders/204926/1578593spi-1m.jpg" width="1152" height="864">