D²OL Benchmark - DBench 1.1 !!!

[bachus_anonym]

Hey guys,

Did you have any idea that there's actually a benchmark for D²OL project ???

I didn't until now - DBench 1.1

I haven't had a chance to check it out - gotta go now - but let's see what those crunchers are capable of!

Shall we ???

Michal

[Jeff]

Cool!

[[XC]thewildblue]

Pimp it for new members !!!!

[shmaa]

Pimp it for new members !!!!

yeah tell them to more accurately benchmark, they have to run the actual D2OL program for at least a year.

[haPpydUde]

Sweet, good idea. This will probly get a little more ppl intrested in this GJ

[Rodzilla]

bachus... OMG man!!!! That is the coolest find ever! You are truly the man!

Unlike me you don't post that much here... but when you do it's absolute gold! (I try and make up with quantity over quality )

Thanks a lot! Now we all have some more stats to drool over!

[chunkylover77]

I ran the benchmark but how do I see the results?

[Jeff]

Should be in the DBench.txt file that's in the DBench directory.

[chunkylover77]

I did not install it. Just ran it from winrar. It does not have an installer. Any ideas?

[bachus_anonym]

Allright... Since you guys seem to be shy, then I'll be the first to post sample result
This is one of my full-time crunchers at it's 24/7 setting:

DFI NF4/A64 Winchester 3200+ @ 2615MHz DDR580 7-3-3-2.5-1T ---> 149 seconds



@chynkylover

I just extracted contents of that archive to this folder: SengentD2OL\D2OL\res\data\bin\

This benchmark needs either Gridwin.exe or Dockwin.exe to work... Those two apps are in that above mentioned folder...

[haPpydUde]

Chunky do what i did just copy the stuff in that folder to your bin directory. It will find the grindwin and dockwin in the same directory and run it

[angrysquirrel]

Now we need to find whats better to crunch with, AMD, Cyrix, Intel, PowerPC, or SPARC!

Good find!

[chunkylover77]

Thanks bachus and Happy I got 152

[bachus_anonym]

bachus... OMG man!!!! That is the coolest find ever! You are truly the man!

Unlike me you don't post that much here... but when you do it's absolute gold! (I try and make up with quantity over quality )

Thanks a lot! Now we all have some more stats to drool over!

Thanks Rod

Actually, I just asked on the Official D2OL forums and one helpful fellow gave me the link

I was looking for a little proggy like that for some time and finally found it...

I really wonder what those Xeon rigs need to complete this one... I mean, two CPUs mean 4 instances... So let's say, just run one instance and divide the time by 4... Am I right ???

[Jerunk]

Barton 2500+@ 2.4ghz .. 11 x 218

[Jeff]

Here's the result from my dual 3.2GHz Xeon rig with HT enabled and running only one instance of DBench...

Target: SARS IV
Candidate: nci_340557.aq
Number of Atoms:    15                  Rotatable Bonds:    0
 1: 355      -5.88   00:12      11: 332      -5.91   00:12
 2: 345      -5.87   00:12      12: 353      -5.88   00:12
 3: 351      -5.88   00:12      13: 340      -5.88   00:13
 4: 359      -5.85   00:13      14: 367      -5.87   00:12
 5: 351      -5.88   00:12      15: 350      -5.85   00:12
 6: 345      -5.87   00:12      16: 349      -5.91   00:12
 7: 356      -5.88   00:13      17: 337      -5.85   00:13
 8: 355      -6.17   00:12      18: 325      -5.88   00:12
 9: 359      -5.91   00:12      19: 352      -5.88   00:12
10: 346      -5.88   00:13      20: 342      -5.85   00:12

14 Mar 2005  06:32:07 PM	0	nci_340557.aq	SARS IV	15	0	-6.17	-5.85	00:04:05	2	29791	00:00:40	00:04:48

Total Time: 288 seconds

200MHz x 16 with BH-5 at only 200MHz(2-2-2-6)

[Jerunk]

AMD looks pretty good so far.

[haPpydUde]

How do you run it on multiple cpu's? This is really only good for you single cpu boys so far. Still nice to have something

[Rodzilla]

Hmmmm... How would we run this on HT P4 or Xeons? ... to get an acurate measure.

[haPpydUde]

lol same thing i was thinking rod

[haPpydUde]

I really wonder what those Xeon rigs need to complete this one... I mean, two CPUs mean 4 instances... So let's say, just run one instance and divide the time by 4... Am I right ???

I missed that lol Dividing by 4 is unrealistic 2.5~3 mabye but no way a 3.2ghz xeon with ht preforms 4x better then single 3.2ghz without ht. I rember reading somewhere that a p4 @ 3.2 with HT preforms like a p4 @ 4ghz in d2ol somewhere... But better then dividing to get unreaslistic results someone with some skill could purhaps modify this program.

[Jerunk]

Or perhaps we could all run 2 instances at the same time? And add the scores?

[Rodzilla]

No... I just tried it with 2x benches running at the same time...

On my P4E @ 3.74GHz I got...

One showed 332 sec and the other 337. It maxed out the CPU. 1 copy used 50%.

So if you average them together you get 334.5 sec. Divide that by 2 (Because each node is averaging 1 Cand every 334.5) and you get 167.25

This is the only way to get an accurate score. Run 2x or 4x at the same exact time. You have to start them as close together as possible. I started them within 1 sec of eachother.

[Rodzilla]

Also... you only need to copy Dockwin.exe and Gridwin.exe into the same dir as the bench... this way you can keep the 2 separate.

[Rodzilla]

So on my dual XEON LV rig @ 3.2GHz and mem @ 228MHz 2-2-2-5... I get...

378+377+379+376=1510

So I average 4 cands evey 377.5 Sec
or 1 every 94.375 sec