" Personal TEST " CF ATI HD 5870 + HD 5850 on BENCHMARKING and GAMING !

[Gaul]

RIG :

- q 9550 @ 4.2 GH
- P45 UD3-P
- 4 GB DDR 2
- 5870
- 5850
- TX - 850

Software :

CAT 10.6 WHQL, WIN 7 64 Ultimate


Look's

- 5870 as primary
- 5850 as secondary
- run on 8x 8x bandwith cause P45






REMEMBER :

- when i say SINGLE = 5870 only
- CFX = 5870 + 5850 CFX


POWER CONSUMPTION





HEAVEN 2.1



SINGLE




CFX

[Gaul]

VANTAGE

SINGLE





CFX




CRYSIS WARHEAD


SINGLE





CFX

[Gaul]

JUST CAUSE 2


SINGLE





CFX





AVP

SINGLE





CFX

[spajdr]

1400x900 resolution is too low for CFX testing, you will be limited by CPU.

[damha]

If its going up from 59 to 87 I am pretty sure the CPU isn't a bottleneck at these settings.

That's quite an improvement. A worthwhile upgrade if a 5870 owner finds a good deal on a 5850...

[Razrback16]

When the 5870 is in Crossfire with a 5850, does the 5870 only use 1440 shader processors or does it still use its full 1600 sp? Also will the 5870 have to "downclock" itself to run in sync w/ the 5850 or no?

[zpaf]

I think platform holding back the crossfire setup.
Here is my setup with my i7-960 and two 5870 all at defaults.

[BeepBeep2]

When the 5870 is in Crossfire with a 5850, does the 5870 only use 1440 shader processors or does it still use its full 1600 sp? Also will the 5870 have to "downclock" itself to run in sync w/ the 5850 or no?

I'm not sure about it using 1600 or 1440 but I know for sure it doesn't downclock. Workload is just split up depending on how powerful the second GPU is.

I can run one of my 5770 at 1000 and one at 850 and it still works...same with my 4850 @ 770 and my 4870X2 @ 850.

[keithlm]

I'm not sure about it using 1600 or 1440 but I know for sure it doesn't downclock. Workload is just split up depending on how powerful the second GPU is.

If the Crossfire setup is doing alternate frame rendering, then the faster card is still spending a lot of time waiting. For all practical purposes it is downclocked to the speed of the slower card. (Although it will display the faster speed in GPU-Z etc.)

Anybody with Crossfire can test this:

1. If you have different speeds cards you can just run the "base" test and continue to step #2. If you have the same cards you could decrease the second cards speed and run then run a "base" test. (i.e., you are simulating a slower second card.)

2. Then decrease the first card's speed to match the second card and run the test again.

If you see the same speed on both tests... then even though the first card might display a faster frequency it is basically being slowed down by the second card during actual use.

(NOTE: I might put my 4850x2 card back into my machine and test this. But I haven't used my 4850x2/4850 setup since I went with a 5870. I would probably put the 4850x2 back into the machine if the folding client or similar things worked better with AMD...)

[Gaul]

anyway inc FPS on game is about 40-50 %, not bad after all 4 me ..

[sniper_sung]

anyway inc FPS on game is about 40-50 %, not bad after all 4 me ..

I think your crossfire is incredibly excellent. I only get around 50% fps increase from a second 5870. At the very beginning I should have bought a 5970 instead, which doesn't have heat/noise/pcie lane issue

[Gaul]

I think your crossfire is incredibly excellent. I only get around 50% fps increase from a second 5870. At the very beginning I should have bought a 5970 instead, which doesn't have heat/noise/pcie lane issue

thanx, and the most important thing is only 8x / 8x bandwith 'coz P45 ,

i wonder if i can 16x / 16x