Crossfire performance

[memnoch 0]

I upgraded my system to a Crossfire 1950/1950XTX system today, previously running with just the single 1950 XTX, and could not see any performance increase whatsoever. I have double checked in CCC and it is definitely enabled but does nothing for system performance at all as far as I can see. I did some Google searches and found reference to a bug with SLI systems, where you can rename the arma.exe to 3dmark06.exe. I tried it anyway but that makes no difference either. Anyone else with a Crossfire system care to comment?

System specs:

XP SP2

MSI 975X Platinum PE

Intel Core 2 Extreme X6800

2GB Corsair Dominator 3-4-3-9

ATI 1950 CE + 1950 XTX

C Labs X-Fi Fatal1ty

[Paco454 0]

Try the FEAR.EXE profile. Some users running SLI nVidia are getting up to 49% FPS increases, and that includes me. I'm in the process right now of asking nVidia to make a custom profile, however their first response was simply to make one myself. DUH!! They now have been told about the 31 optimaztion flags which can only be seen in nHancer. I'm now waiting to see what info nVidia will respond with. I have asked where I can make the request to have the game optimized using the optimaztion flags and added to the profile list, instead of using the FEAR profile.

PACO454

[memnoch 0]

Interesting. Problem is I don't have an Nvidia card. The 1950s are ATI. I only mentioned the SLI thing as it was the only thing I could find referencing ArmA and dual card setups. Thanks anyway.

[memnoch 0]

I've just finished some testing with FRAPS in a forest on the north of the

island.

With Crossfire on I get around 17-20 fps in the forest, looking straight down

this will increase to high 30s low 40s. Oddly if I then reduce

the view distance to 500 it hardly budges at all and if I whack it up to 10000

I lose around 5fps.

Without Crossfire the values are almost identical. Perhaps a few fps less when

view distance is set to 10000. The highest I saw was when I had ran out of the

forst down to the coast and looked straight up and got 64fps. That was more

like it. I can only assume that either the game needs some severe

optimisations or that the way the game works just doesn't seem to take much,

if any at all, advantage of Crossfire.

All tests were done at 1024 x 768 with the default High detail setting. Only

the view distance slider was altered during tests.

[shadow 6]

Interesting. Problem is I don't have an Nvidia card. The 1950s are ATI. I only mentioned the SLI thing as it was the only thing I could find referencing ArmA and dual card setups. Thanks anyway.

What he is saying is you should try renaming Arma.exe to fear.exe. I believe this works for crossfire too.

Quote[/b] ]Some users running SLI nVidia are getting up to 49% FPS increases

Isnt it closer to 90%?

[memnoch 0]

I'll certainly give it a try. I wrongly assumed that this was a purely Nvidia software thing. The 40% is what I had read elsewhere. What I have found with my tinkering is that from the default High setting which just has Shader and Prost Processing set to high, knock Post Processing to Low and I can put all the others to High and it sill runs faster, even in the middle of a dense forest. Not sure what its supposed to be doing on High but it doesn't look any worse.

[memnoch 0]

Just tried chaning to fear.exe and again it doesn't seem to make much difference. Not the extra 20-30fps you would expect to get on paper certainly. My best best I think is to leave Post Proc. on Low and wait for the new patch and see what that brings.

[Paco454 0]

Interesting. Problem is I don't have an Nvidia card. The 1950s are ATI. I only mentioned the SLI thing as it was the only thing I could find referencing ArmA and dual card setups. Thanks anyway.

What he is saying is you should try renaming Arma.exe to fear.exe. I believe this works for crossfire too.

Quote[/b] ]Some users running SLI nVidia are getting up to 49% FPS increases

Isnt it closer to 90%?

Ya....LOL sorry my mistake. It's more like almost double the FPS!!!!!!!.. By the way my request to have ArmA added and optimized by nVidia has reached level 2 tech at nVidia. Should know something soon.

PACO454

[Paco454 0]

Interesting. Problem is I don't have an Nvidia card. The 1950s are ATI. I only mentioned the SLI thing as it was the only thing I could find referencing ArmA and dual card setups. Thanks anyway.

Yes, I meant to try the FEAR profile.

PACO454

[DaSquade 0]

Maybe it is just me (non SLI user, but interested if it all 'pays' off), but does the SLI/crossfire adventace only kicks in at high resolution. Meaning if you are not planning to run 'what ever supported or non support game tittle' not higher then 1280 you wont have much improvement, compaired to resolution of 1600 and more. Key question is what are the results once past the 1280 resolution?

[Paco454 0]

Maybe it is just me (non SLI user, but interested if it all 'pays' off), but does the SLI/crossfire adventace only kicks in at high resolution. Meaning if you are not planning to run 'what ever supported or non support game tittle' not higher then 1280 you wont have much improvement, compaired to resolution of 1600 and more. Key question is what are the results once past the 1280 resolution?

Doesn't matter what res. The cards in SLI share and therefore increase FPS. RES depends on how much video ram there is. A single card with 256 meg of video ram in a non-SLI system compared to a 512 meg single non-SLI card will have higher frame rates at higher RES because it has more video memory to buffer the higher res.

PACO454

[memnoch 0]

Just been doing some searching and found these two pages:

http://www.driverheaven.net/windows....am.html

and:

http://acat.arti-sieraden.nl/modules....start=0

that aren't that old that specfically state that there is no profile YET for Crosfire and in fact that renaming the .exe appears to make no difference at all. This leaves me with two scenarios:

1. These guys are right and people who say there are getting a boost are wrong or the boost is something else they did rather than Crossfire.

2. They are wrong, for whatever reason that me be, driver revisions etc.

These posts aren't that old so I'm HOPING for my sake that number 1 is the case.