stuttering with crossfire/eyefinity

[Tamlin]

Only partially I'm afraid, so I need to read a bit more before I can comment on that thread. I did however read something earlier on B3D, where they are investigating the microstuttering in CF/SLI and the myths around it:
http://forum.beyond3d.com/showthread.php?t=57288

[MaxDarklighter]

Very good info guys!
By the way if I don't whant to use DicplayPort or DP-VDI active adaptor for 3-rd monitor, can I instead use 3-rd DVI from second CrossFire card? And 4-th DVI for LCD TV?

[Dave Baumann]

I'm really surprised no one's mentioned the apparent degrade of xfire scaling with resolution.

Scaling resolution down or up?

You'll always get lower Crossfire scaling with lower resolutions because the lower the resolution the less time is spent rendering a frame of the GPU and the more CPU limited you become.

[Hydralisk_X]

[quote]I'm really surprised no one's mentioned the apparent degrade of xfire scaling with resolution.

Scaling resolution down or up?

You'll always get lower Crossfire scaling with lower resolutions because the lower the resolution the less time is spent rendering a frame of the GPU and the more CPU limited you become.

Scaling up... crossfire scaling at 5760x1080 is always worse than 1920x1080 for me.

[MaxDarklighter]

Guys? :(

[Tamlin]

[quote]I'm really surprised no one's mentioned the apparent degrade of xfire scaling with resolution.

Scaling resolution down or up?

You'll always get lower Crossfire scaling with lower resolutions because the lower the resolution the less time is spent rendering a frame of the GPU and the more CPU limited you become.

Any idea why people experience the opposite? Lower Crossfire scaling with higher resolution? :)

[skipclarke]

Hydralisk_X,

I wonder if you're hitting the CPU limit. Here are my findings on scaling with the Eyefinity6 card.

http://www.widescreengamingforum.com/wiki/ATI_Radeon_5870_Eyefinity6_-_Featured_Review_-_Page_15

[Hydralisk_X]

Hydralisk_X,

I wonder if you're hitting the CPU limit. Here are my findings on scaling with the Eyefinity6 card.

http://www.widescreengamingforum.com/wiki/ATI_Radeon_5870_Eyefinity6_-_Featured_Review_-_Page_15

Do you have Just Cause 2? The only scaling benches I've seen are from Tom's where it appeared that from 1920x1200 to 2560x1600, the scaling suffered a huge loss. But I haven't seen any triple screen benches yet. Personally it's the one game that bothers me a lot since I only see about 20-25% scaling and was one of the reasons I got a 2nd 5870. http://www.tomshardware.com/reviews/geforce-gtx-480-3-way-sli-crossfire,2622-10.html

For the other games, the scaling isn't horrible compared to single screen, but they're most definitely worse going from 1920x1080 to 5760x1080. Do you think my i7 860 @ 3.4 ghz is bottlenecking at such an extreme resolution? I noticed in your test setup you have a i7 920 @ 2.67.

I did browse through the review and I noticed that in general the single screen resolution scaling was better than triple screen except for far cry 2. It's hard to get hard numbers from the graphs since I have to interpolate, so I just eyeballed it and tried to guestimate.

Overall, the scaling isn't bad in triple screen although I'd like it to be better and get closer to single screen scaling, but Just Cause 2's terrible scaling really bothers me.

[tepescovir]

Very good info guys!
By the way if I don't whant to use DicplayPort or DP-VDI active adaptor for 3-rd monitor, can I instead use 3-rd DVI from second CrossFire card? And 4-th DVI for LCD TV?

NOPE, you can only use the outputs from one card, all outputs on seond card are disabled

[Phoenixdancer]

Again i will say for the 100th time on many forums

Give the 10.x or 8.74 driver pack a try, All my tri-fire and eyefinity problems gone

[tepescovir]

and again we say, good for you, but the 8.74 drivers didnt fix it for others, crossfire works better with them for some games, but still not for most. and vsync dosent work at all in most games with the 8.74 drivers so still using older ones

[elior77]

I can vsync gta4 (i see the frames at 60 no more) but still i get tearing/stuttring

[johnnyvd]

I can vsync gta4 (i see the frames at 60 no more) but still i get tearing/stuttring

This is a GTA4 problem.. it has to do with GPU ram which is flooded.. Not really X-fire.. i think you shuld use commandline -memrestrict best google it. :D

[Mach1.9pants]

Well I sold my 5970 and replaced it with a 5870 due to the stuttery performance in CF with EF. A friend of mine tried 5970/5870 CF/5870 single and ended with the 5870. Unless you get 60FPS with your CF rig, EF is pretty unplayable.

I confirmed this in every game I could get the the 5970 down to less than 60 FPS: Crysis, BFBC2, ArmA2, GTA4, etc

[legoman666]

I am getting microstuttering out the ass with 2 5850's at 5760*1200. The two games I tried: WoW and Trackmania, were completely unplayable. I have a mobo with Lucid Hydra, so I'm going to give that a shot.

EDIT: Ok so with Hydra enabled, I'm sitting here getting 40fps with everything at max. No stuttering.
With Hydra disabled and crossfire disabled, I get 43ish fps with no stuttering.
With Hydra disabled and Crossfire ENabled, I get 40ish fps with lots of stuttering.

Clearly the problem is crossfire.

[Hydralisk_X]

I am getting microstuttering out the ass with 2 5850's at 5760*1200. The two games I tried: WoW and Trackmania, were completely unplayable. I have a mobo with Lucid Hydra, so I'm going to give that a shot.

EDIT: Ok so with Hydra enabled, I'm sitting here getting 40fps with everything at max. No stuttering.
With Hydra disabled and crossfire disabled, I get 43ish fps with no stuttering.
With Hydra disabled and Crossfire ENabled, I get 40ish fps with lots of stuttering.

Clearly the problem is crossfire.

I don't have stuttering at all in trackmania. In fact I don't stutter in any games now. I used to stutter in warhead, but it's gone now. I don't have WOW so I can't test it, but games like JC2, BC2, AvP, source games, crysis / warhead, etc. all don't stutter now.

[johnnyvd]

I assume that is more of a performance hit the higher your res and not a scaling issue this should be obvious.

I 2nd this: to see if scaling is ok you should compare Xfire on/off on the three screen reolution. Not compare the single to the three screens..

[Hydralisk_X]

[quote]I assume that is more of a performance hit the higher your res and not a scaling issue this should be obvious.

I 2nd this: to see if scaling is ok you should compare Xfire on/off on the three screen reolution. Not compare the single to the three screens..

I'm not sure how you guys misconstrued the post so badly.

Scaling is defined as how much performance the second card adds. So when I said scaling is worse in triple wide than single monitor that means the addition of the second card produced less of an increase in performance percent wise in triple monitor vs. single monitor.

For example, single monitor xfire off vs. on is 70% increase while triple monitor xfire off vs. on is 50% increase.

[johnnyvd]

Sorry, when i typed my reply i already got the feeling that i could be misreading..

And that you indeed did test on both situations with Xfire on and off.. :oops: