Printable View
142 on a 3200+ at 260X10 mem only running at 173 though as the board only seems to give 2.70v even if i set it to 2.85v (neo2) ive got a ddr booster but it doesnt work as my psu hasnt got a -5v line. i need to get my electronics books out at set up a 5 to -5v flipper then i should be at 260 sync with my bh5
2.4C@3.115 mem at 209 2-2-2-5 = 186 (HT on) useing Rod's formula
124 seconds - ain't bad I guess...
is that with your FX@3ghz?
whatcha mean?Quote:
Originally Posted by Rodzilla
all test were 1:1 if that's what you're asking... dividers on K7 = :down:
Hey enzoR, yep thats running at my current spec. Do you think the score is good for my setup or should I be expecting a faster time?...Quote:
Originally Posted by enzoR
I'm thinking of joining your D²OL team as my PC is sat tuned off all day whilst I'm at work... :mad:
Cinders.
I think that is about right for your setup. Its not a bad idea joining the team. Anything helps us to the top.Quote:
Originally Posted by cinders
I shall be joining this week when I get time to set it all up.Quote:
Originally Posted by moddolicous
:toast:
a) We already have too many... :(Quote:
Originally Posted by matt9669
b) You run the bench just like you run D2OL... multiple copies. There is no better way to do it and it is 100% real world.
c) The bench uses a real candidate to calculate your score. The candidate never changes from bench to bench so you can compare scores... This bench uses the D2OL engine and a real drug candidate. This is so we can compare scores and call it a bench! :D If the candidate changed we couldn't call it a bench. Think of it this way... If 3DMark Randomly put 2, 3, 4, 5, or 6 trolls in the game test of 2DMark 03 you couldn't call it a bench... as it's not performed as a control. The constant (or control) is the work to be done... the variable is you PC and it's setup/hardware. :)
tell that to the barrel on car chase high. :rolleyes:Quote:
Originally Posted by Rodzilla
What? Are you giving me crap or are you trying to be useful? :lol:Quote:
Originally Posted by shmaa
i'm giving that barrel crap... i woke up in a cold sweat many nights after having a nightmare about that barrel skipping across my 3dmark run. :bounce:Quote:
Originally Posted by Rodzilla
Many years of therapy and lots of pills and you will be all right! ...Until then... you scare me! :eek:Quote:
Originally Posted by shmaa
They gave up on the pills years ago...
back on topic.. is this what you wanted to know? Or was that question not directed to me?
http://www.xtremesystems.org/forums/...954#post769954
Yea.. I was referring to this post...Quote:
Originally Posted by shmaa
http://www.xtremesystems.org/forums/...387#post766387
So you ran it with your CPU clock the same each time... But your ram was running at 100MHz and 220MHz... And you only had a 2 sec difference... Hmmmmmm... So not only do Mem timings not make a diff... But Mem clock doesn’t matter either?!?!!?!!??!?! That pretty much means that RAM doesn’t mean anything in D2OL... It may not even matter if you have enough to keep it from paging. (Which is probably true as the speed of your ram has no effect...)
correct.Quote:
Originally Posted by Rodzilla
also note my main rig has 1 GB of ram and it runs the benchmark the same speed (when running at 2.0ghz) as one of my 2.0ghz crunchers with 256mb of ram running WinXP :stick:
Well that makes sense because D2OL only uses about 65MB per node... But what if you had all the physical mem used... and D2OL had to be paged?!?! hell... if it doesn't matter you can stick a 128 stick of the cheapest ram in a dedicated cruncher and it won't matter at all!Quote:
Originally Posted by shmaa
even with only a 128mb stick, if you run win2k and kill all unessential processes and services and have d2ol running, you still wont run into large page file usage.Quote:
Originally Posted by Rodzilla
167 sec 2500+ @ 2500mhz 12.5x200 2x512 CH-5 @ 2-3-2-11 new win install
168 sec 2500+ @ 2500mhz 12x5x200 2x512 CH-5 @ 2-3-2-11 old win install
164 sec 2500+ @ 2594mhz 11.5x225.4 2x512 OCZ VX @ 2-2-2-11 old win install + steam running
149 sec 3000+ @ 2500mhz 9x277.7/133 2x256 BH-6 @ 178.5mhz 2.5-2-2-5 old win install
My winnie is a POS.. 2300mhz @ 1.65v max stable :(
Have you tried this solution for poor clocking winnie's?Quote:
Originally Posted by STEvil
http://www.xtremesystems.org/forums/...ad.php?t=55942
In order to run this, do I have to stop regular D2OL from running?
yep needs to have the normal d2ol turned off otherwise itll slow down and give the result of trying to run 2 at once.
Thanks Cossey. Just got it to work. This is from my main rig. http://www.xtremesystems.org/forums/...id=27145&stc=1Quote:
Originally Posted by Cossey
Here's my dad's P3 700 @ 800. Not much, but every little bit helps out, right. BTW, can someone run a test with 1t and 2t to see if that makes a difference?
Quote:
Originally Posted by shmaa
Yup.
In fact, if I set the CPU voltage by VID only it will remain at 1.47v reguardless of setting.. or 1.37v at some :stick:
If I use the % increase then it all works correctly.. go figure. I used a DMM to verify the vcore was changing each time, too... I dont trust bioses at all.
So.. fast forward to now.. took the CBBID 0451 back and am now holding a CBBHD 04444XPKW 0081.. interesting combination :D
tested it yet? how's it clock?Quote:
Originally Posted by STEvil
Win2Kpro SP4
XP1700 & NF7-S (from PS2pcGamer)
Swiftech H2O system & 2x256MB BH-5 (from Jeff)
XP1700 @ 12x200 2-2-2-11 165 sec
XP1700 @ 12x200 2.5-4-4-11 166 sec
XP1700 @ 13x185 2-2-2-11 169 sec
All results were duplicated to insure consistancy.
Do timings make a difference? Yes, but very little.
Does FSB make a difference? Yes, but not much.
Seems to be like 3D rendering (e.g. Lightwave) where raw Mhz is king.
shmaa - 2600mhz @ 1.58v so far (used DMM to test voltage from missing capacitor spot near cpu, shows at ~1.51v in windows).
Does 2t make a difference? Like 2-2-2-5 2t and 2-2-2-5 1t?
probably not much, i'll give it a go when I remember to ;)
Wonder if mutiprocessor/uniprocessor driver (ricky tweak) affects it?
mine wont get past the "Press Enter" part.
Dual 2.8GHz Xeon HT enabled 1 instance - 269s
Dual 2.8GHz Xeon HT enabled 4 instances - 108.94s
Dual 2.8GHz Xeon HT enabled 8 instances - 85.45s
So is it better to run more than 4 instances?
you have to go to \D2OL\res\data\bin and copy dockwin & gridwin. I had that problem also.Quote:
Originally Posted by Magnj
I have it there. I have to make a copy of them though?
yea, copy them and paste them into the folder.
u have the benchmark in the folder with dockwin and gridwin already? No copies are needed if thats the case...Quote:
Originally Posted by Magnj
Wow... you sure you calculated the times right? :slobber:Quote:
Originally Posted by Helheim
Pretty sure, I used your formula.. Took an average and then divided that by the number of instances running. I'll try again tonight as well.
Then that's an interesting result!Quote:
Originally Posted by Helheim
running 2T on winchester instead of 1T was a 2 second penalty.
Does anyone else have trouble running 8 instances of DBench? On two different machines now two instances just stop running(ie. window closes) during the test. I've tried 4 times now and the same thing happens. One machine has 512Mb and the other has 1024Mb. :shrug:
ive tabulated most of the results just need to add ram timings and order them. should i do it by processor type or by time?
btw the attachment is the table so far done in excel then zipped coz it wont attach .xls
Ok so it's a bit flawed because 4 of the processes finish way before the others and if they were still running like it would be running regular d2ol they wouldn't stop and allow the others to finish..
http://www.xtremesystems.org/forums/...id=27341&stc=1
(521+447+510+490+883+848+801+889)/64=84.20s each
This is interesting:
Dual Xeon 2.8Ghz
Win2003 1 instance - 261s
Win2003 4 instances - 435,435,436,437 = 108.94s
Win2000 SP4 1 instance - 249s
Win2000 SP4 4 instance - 404,404,403,403 = 100.88s
Both were fresh installs with nothing else installed.
Jeff... i get the same thing... and if I can get them to run all the way through 4 of them finish well before the other 4... so it throws the numbers way off...
Well i finally got round to testing it
I got 147 on my 754 2800+ AMD 64 @2475 Mhz @ 275 x 9 7, 3, 3, 3
seems fairly good for a 745 based AMD 64
Anyone else get very mixed results with there dual xeon machine?
Iv been tryiong a few things like changing the priority of the dbench and am finding very mixed results, varying from 84 seconds to 95+ seconds.
the 84 seconds had 2 in the 280 second places and 2 in the 400+ seconds. Thats at realtime priority. So obviously the first 2 processes are being done at 80-90% of one cpu's power ( ht or not).
Also i found running at 3.5Ghz gave me slower timeings than 3.4Ghz, i would assume its something to do with it being near the limit of its overclock maybe throwing errors into it.
3.4Ghz is 100% stable for me while 3.5 is very close to being stable and i belive when it isnt run on an old generic powersupply will clock to 3.5 fully stable.
Just got myself a 83.5 Second stable with 267+263+416+390
obviously this is infact worse than Jeff's 84 seconds due to the fact on the later ending ones it does not take into account the ones that would still be going.
I did a fresh install of Win2000 and it was 261s (@3Ghz) running one instance, installed all the updates and everything and now it's 285s even after I disable all the unnecessary services, with them running it's 301s. :(
Well im baffled, really
2 runs one after another, NOTHING changed.
first one averages to 95, second to 80.
Im not convinced by running multiple copys of it.
Id love to know how repeatable it is for you other dual xeon guys
Mines been dead on 99% of the time...Quote:
Originally Posted by Smurfy
very good ideabachus_anonym' i will try it out :P
Target: SARS IV
Candidate: nci_340557.aq
Number of Atoms: 15 Rotatable Bonds: 0
1: 346 -5.85 00:09 11: 343 -5.88 00:09
2: 346 -5.85 00:08 12: 360 -5.85 00:08
3: 342 -5.58 00:09 13: 350 -5.85 00:09
4: 331 -5.88 00:09 14: 348 -5.91 00:09
5: 333 -5.88 00:08 15: 340 -5.88 00:09
6: 354 -5.85 00:09 16: 345 -5.91 00:09
7: 343 -5.88 00:09 17: 360 -5.85 00:09
8: 352 -5.58 00:09 18: 351 -6.17 00:09
9: 351 -5.88 00:09 19: 344 -5.58 00:08
10: 347 -5.88 00:09 20: 349 -5.70 00:09
29 Mar 2005 09:09:54 PM 0 nci_340557.aq SARS IV 15 0 -6.17 -5.58 00:02:56 2 29791 00:00:46 00:03:44
Total Time: 224 seconds
1 instance on a dual 2.0ghz Athlon XP system (2x2ghz mobiles).
Why only one instance ? Shouldn't it be 2 ?
Yea, with a dual athlon mobo, you should run 2. BTW, what mobo are u using that allows 2 XP's to run simultaneously? I thought they had to be MP's.
He has mobiles I think.
The mobiles are completely unlocked chips... so if my memory serves me correctly, crossing and blowing bridges actaully does something. So I think all you have to do is blow some bridges out on a XP-M and it will turn it into an MP.Quote:
Originally Posted by moddolicous
After some more benching I consistenly scoring below 80 seconds with dual xeons at 3.4Ghz, but like before 2 instances are finishing earlier than the remaining 2, which obviously could mean the scores are inaccurate. They are all stared within 3 seconds of each other and the cpus have HT enabled, im intrested if anyone has and ideas how or why this is happening?
So how do I run two instances? Just run the program twice and set affinity for one of the runs?
I just ran two instances at the same time I didn't worry about the affinity. Does anyone else worry about that?
Well hopefully I am suppose to have some money coming my way in the future and thought why not invest it into a good server something I can run a dedicated CS:S server on from time to time. The real question is which is the best for D2OL? Kind of a little confused on how many instances I can run on a machine. If someone could please clarify how many instances the machines below can run I would be greatful.
1) Dual Xenon ($300 per chip is round about limit, suggestions welcomed)
2) Dual Opteron 246's (no good overclocking boards out :( )
3) AMD64 X2 (previews already point to OC'in well)
Kind of sounds confusing but any help is appreciated, not really well acquainted with intel :confused:
#1) 4 instances
#2 & #3- 2 instances