CrossFireX low fps

[jnmbhj]

I just got a 7970 ghz into my system and I tried running unigine at 1080p and it was very smooth I had an average of 60+ fps (I can't remember the exact number) and with heavy action it will dip down to 45 fps

But I tried playing battlefield 4 and both cards spun up but I had some micro stutters and the Fps was just horrible I usually get 55 -65 fps But with crossfire enabled I got 110 fps in places with no action but once there was even a little action like explosions or gunshots the Fps was plummet down to 25 fps making the game almost unplayable. And that happens very often. Does anyone know a fix? I am running the newest drivers (omega which is supposted to make it even smoother for crossfire applications) 

[ExcitezR]

open MSI afterburner when you are playing and see if the second GPU is maxed out at 100 is so you will need to disable ULPS 

[Altecice]

Its possibly your CPU bottlenecking. I had this issue with my 8350 and my 290x's I swapped out the CPU for an I7-4790K (with no other changes) and I gained FPS and no dips (my origional issue was with Tomb Raider and it juddering when loading new areas)... A good test for a CPU bottleneck is to run a benchmark at stock CPU speeds (so 4ghz for you) mark the average FPS. Then bump it back to OC speeds and run the test again. Do you gain FPS? if so thats a clear way to show that your GPU is being bottlenecked by the CPU as making the CPU faster = it can give more instructions per second to the GPU so speeding it up.

 

I can further verify this by my 4790K in that when I oc from 4ghz to 4.6ghz 0 FPS is gained so its shows my CPU is more than capable of "feeding" the GPU's enough data to keep them at 100% usage when they need it and my bottleneck lies in other hardware.

[jnmbhj]

open MSI afterburner when you are playing and see if the second GPU is maxed out at 100 is so you will need to disable ULPS 

I was using the MSI afterburner (frogot to mention it) 

The GPU 1 had 60% usage 

GPU 2 had 50% usage 

I will disable ULPS now and see if makes a difference 

[Altecice]

I was using the MSI afterburner (frogot to mention it) 

The GPU 1 had 60% usage 

GPU 2 had 50% usage 

I will disable ULPS now and see if makes a difference 

 

Not 100% usage of both GPU's is also another sign of a bottleneck

[jnmbhj]

Not 100% usage of both GPU's is also another sign of a bottleneck

I know, and thats what I was worried about. 

[Altecice]

I know, and thats what I was worried about. 

 

read my longer post above and try it. But I fear you have the same issue I did.

[ExcitezR]

I know, and thats what I was worried about. 

hey man i have the same set up as you and there is no bottle neck are you running your games in full screen because you will need to do that 

[jnmbhj]

hey man i have the same set up as you and there is no bottle neck are you running your games in full screen because you will need to do that 

Yes they are run at full screen at 1080P Thanks for the reassurance 

[Priller]

I just got a 7970 ghz into my system and I tried running unigine at 1080p and it was very smooth I had an average of 60+ fps (I can't remember the exact number) and with heavy action it will dip down to 45 fps

But I tried playing battlefield 4 and both cards spun up but I had some micro stutters and the Fps was just horrible I usually get 55 -65 fps But with crossfire enabled I got 110 fps in places with no action but once there was even a little action like explosions or gunshots the Fps was plummet down to 25 fps making the game almost unplayable. And that happens very often. Does anyone know a fix? I am running the newest drivers (omega which is supposted to make it even smoother for crossfire applications) 

 

Sounds like one of your cards has gone bad.

 

I had the same issues and it turns out my secondary card went belly up.

 

Try switching them around 1 card at a time and see if one gives you issues. Also look at the stock voltage reported in MSI afterburner.

[ExcitezR]

Yes they are run at full screen at 1080P Thanks for the reassurance 

yea man my cards dont run at 100% unless they are being stressed so try and put every thing to ultra 

[jnmbhj]

Sounds like one of your cards has gone bad.

 

I had the same issues and it turns out my secondary card went belly up.

 

Try switching them around 1 card at a time and see if one gives you issues. Also look at the stock voltage reported in MSI afterburner.

Both my cards are gigabyte and they dont show their voltages. I am know that both cards are fine I have tested them both this morning seperatly using the unigine heavan benchmark 

[jnmbhj]

yea man my cards dont run at 100% unless they are being stressed so try and put every thing to ultra 

I set Ran the ULPS and I get steady fps of 55-60 fps everything ultra and 4x AA but the gpu usage is only at 50% on both my cards 55% on the gpu 1 and 52% on GPU 2. But I tried running unigine at 1080p at full screen and both gpu went to ~98%. 

[ExcitezR]

I set Ran the ULPS and I get steady fps of 55-60 fps everything ultra and 4x AA but the gpu usage is only at 50% on both my cards 55% on the gpu 1 and 52% on GPU 2. But I tried running unigine at 1080p at full screen and both gpu went to ~98%. 

hey man set the res scale to 150% in BF4 and then see what you get 

[Timmy-P]

hey man set the res scale to 150% in BF4 and then see what you get 

Wouldn't that make things worse?

[ExcitezR]

Wouldn't that make things worse?

no because it will fores the cards to run harder 

[Timmy-P]

no because it will fores the cards to run harder 

I don't play Battlefield, forgive me, but to clarify resolution scaling downsamples the image, correct?

Higher resolution = more information handled by the API, thus CPU = more likely of CPU bottleneck.

[ExcitezR]

I don't play Battlefield, forgive me, but to clarify resolution scaling downsamples the image, correct?

Higher resolution = more information handled by the API, thus CPU = more likely of CPU bottleneck.

no i dont think it works like that but i have the same setup as him and when i play on one screen the GPUs both are about 50-60% but i play on 3 screens and they both go to 100% so i thought that would work 

[SkilledRebuilds]

I don't play Battlefield, forgive me, but to clarify resolution scaling downsamples the image, correct?

Higher resolution = more information handled by the API, thus CPU = more likely of CPU bottleneck.

Your right the CPU can be worked harder an amount, but it's not as much as the GPU hit, if you render 200%, your effectively doing 4K over 1080p, thats a hugely GPU taxing feature to use, moreso than the CPU tax.

Great for limited GPU usage (which is not great) but to be further used when your cannot gain any more FPS, but can gain some extra details, at the same performance levels.

[Timmy-P]

Your right the CPU can be worked harder an amount, but it's not as much as the GPU hit, if you render 200%, your effectively doing 4K over 1080p, thats a hugely GPU taxing feature to use, moreso than the CPU tax.

True.  Only problem is that the only thing you're accomplishing is a better looking game at the same framerates.

[SkilledRebuilds]

True.  Only problem is that the only thing you're accomplishing is a better looking game at the same framerates.

Yeah,..I know,. but if thats all your options are... and the game wasn't particularly fast paced/input latency dependent, I'd still do it.

[thorpj]

Its possibly your CPU bottlenecking. I had this issue with my 8350 and my 290x's I swapped out the CPU for an I7-4790K (with no other changes) and I gained FPS and no dips (my origional issue was with Tomb Raider and it juddering when loading new areas)... A good test for a CPU bottleneck is to run a benchmark at stock CPU speeds (so 4ghz for you) mark the average FPS. Then bump it back to OC speeds and run the test again. Do you gain FPS? if so thats a clear way to show that your GPU is being bottlenecked by the CPU as making the CPU faster = it can give more instructions per second to the GPU so speeding it up.

 

I can further verify this by my 4790K in that when I oc from 4ghz to 4.6ghz 0 FPS is gained so its shows my CPU is more than capable of "feeding" the GPU's enough data to keep them at 100% usage when they need it and my bottleneck lies in other hardware.

"No other changes" Well you changed motherboard, and that can make a huge difference.

[thorpj]

Wouldn't that make things worse?

no, it's basically supersampling, so you're rendering the frames at 150% res.

 

It'll stress the GPUs more.

[Timmy-P]

no, it's basically supersampling, so you're rendering the frames at 150% res.

 

It'll stress the GPUs more.

As explained earlier, a higher resolution would demand from the API, and therefore require more from the CPU, though not as much as the GPU.

[thorpj]

As explained earlier, a higher resolution would demand from the API, and therefore require more from the CPU, though not as much as the GPU.

So you're saying that requiring the GPUs to render frames at a higher resolution is not going to stress the GPU much more.

 

Right, how does that make sense.