difference between all 5 of my quad 5870 rigs

[The Koolio]

i have 5 systems all running the below hardware and software. why is that even though my cards are all stable at 970 clock and 185 mem they all produce different values, and i dont just mean between cards, i mean as a system whole. the speeds of each system are below, but i want them all to run at above 400, but having issues:

MB: Asus 890fx
processor: sempron 140 (with unlocked core using MB)
ram: 4gb corsair 1333
hdd 750 gb
graphics: 4 x xfx 5870

windows 7
phoenix rising
afterburner - for temp graphs
amd gpu clock tool for overclocking


system 1
card 1 - 360 mh
card 2 - 370 mh
card 3 - 360 mh
card 4 - 365 mh

system 2
card 1 - 380
card 2 - 380
card 3 - 380
card 4 - 380 (this card is a refference card, when run on a stand alone system i get 440 mh no problems)

system 3
card 1 - 400
card 2 - 400
card 3 - 400
card 4 - 390

system 4
card 1 - 370
card 2 - 370
card 3 - 380
card 4 - 400

system 5
card 1 - 420
card 2 - 420
card 3 - 420
card 4 - 420

so yes there are differences between each card, but systems as a whole have the notable differences. even my refference card only gets 380 mh and as explained ive got 440 out of this on a stand alone system.

Any suggestions will be appreciated!

[Reckman]

Are the harddrives cloned? maybe clone the harddrive off the system getting 420, might be software

[fascistmuffin]

It might be the hardware's component's tolerances coming into play rather noticeably, since you have a lot of identical systems set up.

If you don't know what I'm talking about: All electronic components have tolerances that they operate under, since it's very hard to make them perfectly. Take a 68 Ohm resistor with a 5% tolerance. This means that it'll have a resistance of 68 * +-.05 when measured, giving it a range of 64.6 resistance to 71.4 Ohm resistance. What's happening in your systems might be that the processors, GPU, RAM, HDD, mobo have tolerances that they operate under, and they're adding up and being easily found since you have so many samples to look at.

It's not uncommon for identical electrical systems to run differently. That's why you see computer advertised with hardware, rather than benchmarks (eg 3.0GHz process rather than "opens firefox in 1 second").

[melco]

Are you absolutely sure that core clock, memory clock, voltages are all the same among the systems?

[padrino]

What power supplies are you using for the systems? I have a system that does the same thing because it cannot deliver the wattage required when I up the voltage and clocks.

[Saturn7]

I have a similar setup, 4 triple 5870.

All are the same hardware, but some are faster than others, and some are more stable than others.

If you look at the REV number of your 5870 they might be different

For example, I have 9 5870 which are REV 3.0 and 3 5870 that are REV 1.6 from the same manufacture and same model.

The 1.6 REV's don't go past 400mh but the 3.0 easily go to 420.

[The Koolio]

i checked the clocks, they are different for each system, as some run stable at lower speeds. I cant change the individual clocks as they start crashing :S My psu is an OCZ 1250w, i dont think power is an issue, its either processor bottlenecking or something else?

[computerparts]

OCZ isn't exactly known for providing top quality psu's. That said, I would say it's the tolerances taking effect as mentioned earlier. No two pieces of equipment will perform the same regardless if they're the same model.