R9 295X2 TERRIBLE Crossfire Utilization In Games

[BlackPanthaa]

Hello!

 

So basically a month ago I purchased the R9 295X2 GPU to plop into my system, massive upgrade from the R9270X I thought! But was it? Not really...

 

I'm using an FX-9590, which people say will cause a bottleneck, but really, it shouldn't cause something this bad.

 

I had these issues for a month, but assumed it was to do with my 850W PSU, which I have upgraded to 1000w a few days ago. Nope, same issues. Just no random shutdowns on full CPU + GPU load.

 

Running Furmark or Unigine Heaven, my GPUs are constantly maxed out as they should be. Drops to only 99% usage of the GPUs. 130+ fps on Furmark at 1080p, with 8100ish score. More than double that of an R9 290X as it should be. Unigine never really drops below 60fps with tessellation on extreme, the benchmark on ultra with no AA at 1440p.

 

Now, when it comes to the two latest AAA PC games I have, they are GTA V & The Crew. My results are literally pathetic and I feel I have wasted money on this card due to these. I have spent the past few days trying to figure it out around work times.

 

GTA V @ 1440p, max (advanced settings all off) no aa, no vsync - indoors I get about 50-70fps, outside I get 40-55fps. Pathetic.

The Crew @ 1080p, ultra, 4x aa - 45-50fps

 

Now one thing I tried, is going into borderless window to disable crossfire and see my results. Practically the same. A drop of about 5fps, which is never stable when the crossfire is on anyway. With both games.

 

I should also state that when I enabled X4 AA on GTA V, the usage between the cards jumped up to around 60-75% on each card... so there is definitely room there not being used...

 

Much googling and using software, my GPUs for some reason as making up for eachother rather than using their full power together. For example, when one is at 50%, the other is at 50%. When one goes to 75%, the other goes to 25%. They sometimes manage to get to 105%+ usage added together, but that is only for a second or two, which does boost my FPS to the higher points for a short time.

 

When I switch to one card (borederless window) I get 100% GPU usage on GPU1 and 0% on two. (I tried this to disable crossfire and see)

 

This makes no sense to me as in the benchmarking tools, everything runs perfectly topped out. But in games... I fail to hit a stable 60fps in any of my new games... one of the R9 290Xs alone should be able to do that...

 

Any ideas people? It's been driving me insane... any suggestions would be super helpful.

 

Thanks!

Theo.

 

My system:

CPU: AMD FX-9590

MOBO: Gigabyte 990FXA-UD3 SKT-AM3+

GPU: MSI R9 295X2

RAM: 8GB DDR3 1333MHz

CASE: Corsair Obsidian Series 750D

STORAGE: 120GB OCZ Agility 3, 1TB WD Black HDD. 3TB Green HDD

OS : Windows 10 preview, same issues on Windows 8.1

PSU: Corsair RM1000

 

 

----------------------------

New findings guys. So I tried Car Mechanic Simulator 2015, a somewhat light game. CPU usage was round 30-40%, FPS doubled under crossfire. Could this be a CPU thin then? Maybe I should attempt a slight overclock and see what happens?

[Badger906]

amd cards just don't work well on GTA let alone crossfire.

GTA is also very cpu bound and the weakest link in your build is your current cpu.

[PineappleHolidays]

You need to run games full screen to use cossfire properly,

Yes, your 9590 is a bottleneck probably ( at least a little bit, check usage on GTA V ) .

You should be getting ~80-90 fps 1440p gta ( I had 2 R9 290's crossfire with a 5820k and had a little lower than that ).

[Johannes_Lazor]

Your CPU is the problem, GTA 5 can only use 6 cores+2 cores a little.

If you upgrade your CPU to a i5 4690k you will see Significantly more performance.

[Samfisher]

SLI and Crossfire don't work on anything except full screen.

[Notree]

amd cards just don't work well on GTA let alone crossfire.

 

lol Such a fanboy remark. Im sure that helped him a lot. http://www.techspot.com/review/991-gta-5-pc-benchmarks/page3.html

[Badger906]

lol Such a fanboy remark. Im sure that helped him a lot. http://www.techspot.com/review/991-gta-5-pc-benchmarks/page3.html

Its not a fan boy remark in the slightest.. I own and own more amd products than I do Intel and nvidia..

everyone knows amd cards aren't very well optimised for GTA, its been on every forum and tech magazine I read since the launch of GTA..

and as for helping him.. do you want me to magic optimised drivers out of my ass..

its an optimisation issue.. it can't be fixed on a forum..

[BlackPanthaa]

You need to run games full screen to use cossfire properly,

Yes, your 9590 is a bottleneck probably ( at least a little bit, check usage on GTA V ) .

You should be getting ~80-90 fps 1440p gta ( I had 2 R9 290's crossfire with a 5820k and had a little lower than that ).

I do, I just ran it in borderless window to see how it ran without crossfire. And it ran almost identical.

 

Yeah that's what I was expecting. It runs the exact same at 1080p also when I should get FPS over in the hundreds.

 

Also, CPU usage hovers around 70-80% of each core. Never higher.

[Badger906]

I do, I just ran it in borderless window to see how it ran without crossfire. And it ran almost identical.

Yeah that's what I was expecting. It runs the exact same at 1080p also when I should get FPS over in the hundreds.

Also, CPU usage hovers around 70-80% of each core. Never higher.

then you aren't being cpu bound. are you looking at all cores or just an over all % as some games (especially dying light) hammer 1 or 2 cores and don't really utilise the others.

have you tried the latest beta drivers?

they may give better cfx support as amd tend to constantly work on the profiles for a month or 2 after launch.

[BlackPanthaa]

then you aren't being cpu bound. are you looking at all cores or just an over all % as some games (especially dying light) hammer 1 or 2 cores and don't really utilise the others.

have you tried the latest beta drivers?

they may give better cfx support as amd tend to constantly work on the profiles for a month or 2 after launch.

Yep the utilization is around the same across all cores. Most do not go above 80%, one goes to no more than 84% every now and again but drops back below for the most part.

As for the beta drivers. that's what I have been running. Started off with 15.4 and I'm now running 15.5, same issues.

I originally had this in Windows 8.1, but my install bogged down due to other reasons so did a clean Windows 10 preview install on an older SSD. Same issues.

 

It's rather confusing and frustrating as you can imagine. It's performing perfectly in benchmark applications, but in games it's running as if it's an R9 290X or worse. I feel like I'm missing a setting or something somewhere but I have played around with as much as I can come across.

[Badger906]

Yep the utilization is around the same across all cores. Most do not go above 80%, one goes to no more than 84% every now and again but drops back below for the most part.

As for the beta drivers. that's what I have been running. Started off with 15.4 and I'm now running 15.5, same issues.

I originally had this in Windows 8.1, but my install bogged down due to other reasons so did a clean Windows 10 preview install on an older SSD. Same issues.

It's rather confusing and frustrating as you can imagine. It's performing perfectly in benchmark applications, but in games it's running as if it's an R9 290X or worse. I feel like I'm missing a setting or something somewhere but I have played around with as much as I can come across.

the only downside to sli and cfx is the game profile support.

some games and devs nail it and you get super optimisation on launch and really get a boost from multiple gpus and others you don't.

it took dying light about 2 weeks to support sli or crossfire properly and even then I heard it wasn't great.

I'd be frustrated having the world's fastest gpu and not getting the full benefits.

I was 1 click way from owning a 295x2 a few weeks ago but held off because of things like this.

don't be put off though you'll get more good experiences than bad.

[Notree]

 

 

everyone knows amd cards aren't very well optimised for GTA, its been on every forum and tech magazine I read since the launch of GTA..

 

Pretty sure the benchmarks show otherwise and the countless others with 295x2  ,290 crossfire and other AMD cards that are working fine in gta5. And all the issues you read were before the beta driver came out. But whatever this argument isn't helping the OP any  :lol:

[BlackPanthaa]

Hey guys. Could it be due to me using dual monitors at all? It's all I can really think of at this stage.

 

New findings guys. So I tried Car Mechanic Simulator 2015, a somewhat light game. CPU usage was round 30-40%, FPS doubled under crossfire. Could this be a CPU thin then? Maybe I should attempt a slight overclock and see what happens.

[Notree]

Hey guys. Could it be due to me using dual monitors at all? It's all I can really think of at this stage.

 

New findings guys. So I tried Car Mechanic Simulator 2015, a somewhat light game. CPU usage was round 30-40%, FPS doubled under crossfire. Could this be a CPU thin then? Maybe I should attempt a slight overclock and see what happens.

It could be. I've heard of people having issues with crossfire and dual monitors before.  Try unplugging your second monitor .Its worth a try.  

[Badger906]

Hey guys. Could it be due to me using dual monitors at all? It's all I can really think of at this stage.

New findings guys. So I tried Car Mechanic Simulator 2015, a somewhat light game. CPU usage was round 30-40%, FPS doubled under crossfire. Could this be a CPU thin then? Maybe I should attempt a slight overclock and see what happens.

normally a cpu bottleneck could be seen if all cores are maxed out and or gpu usage is very low or not near 100%

an overclock never does any harm.

[BlackPanthaa]

I bring more findings. So Battlefield 4 is another game I had, so decided to jump on that. Flawless.

120FPS ultra on crossfire @ 1440p 2x aa

50-70fps ultra no crossfire @ 1440p 2x aa

 

This is rather odd to me.

[Columbo]

You need a 4690K or 4790K or better to run the R9 295x2 properly. Your CPU can't handle that kind of power, it's bottlenecked badly. The 5820K is a step above the 4790K. 

[Tacitus]

Get an i5. I use 290x crossfire and it works an absolute treat for me. I use V-sync so I don't always get 100% utilisation on both cards, but it's noticeably faster than when I disable crossfire. I'm yet to play GTA V on PC, so I don't know what the situation there is, but I had thought AMD got crossfire profiles out for that game prior to launch. Plus, when you look at games for which drivers have had a bit of time to mature, more often than not the scaling on AMD cards is better than Nvidia cards, and I've found microstutter to be a complete non-issue. Anyone who says crossfire is a broken technology doesn't know what they're on about.

 

The 295x2 remains a fantastic, cost-effective setup for anyone doing high-resolution gaming.

[TheWildOnesEU]

I bring more findings. So Battlefield 4 is another game I had, so decided to jump on that. Flawless.

120FPS ultra on crossfire @ 1440p 2x aa

50-70fps ultra no crossfire @ 1440p 2x aa

 

This is rather odd to me.

 

This is because the Graphics Card is 2 GPU's combined, and it is expected for you to get more performance on using that feature, As for dual 1440p monitors it depends if your using Eyefinity technology. As for the CPU, 8 Cores are too much for gaming, The CPU is even under-powered in other ways too, If you are getting OK results with Crossfire. Just leave it on, it won't effect your experience.  

[BlackPanthaa]

Hello all, back again. Still having these issues but I have some more info that might help solve this, as I doubt I can be the only one? 

 

To be clear, this is on every game (except Battlefield 4 as stated above). I get identical frames in windowed AND full screen. (so with crossfire enabled and disabled)

 

Fire Strike benchmark with new Win 10 driver (yes, had the issue in Win 8 too if not worse) http://www.3dmark.com/fs/5291653

 

Below is my 1440p performance on GTA V, not acceptable for my £500 card at all. 

 

Not CPU issue, not GPU issue (works fine in benchmarks)... I can't understand. Anyone? Thankya!

 

[MahtXL]

amd cards just don't work well on GTA let alone crossfire.

GTA is also very cpu bound and the weakest link in your build is your current cpu.

its funny cuz his cpu shits on your i5. but keep on fanboyin

[Stanvard]

lol Such a fanboy remark. Im sure that helped him a lot. http://www.techspot.com/review/991-gta-5-pc-benchmarks/page3.html

You're just not reading the graph right. The GTX 680 is clearly superior to the R9 295X2 in that test.

[Nergale]

Well actually that might be CPU related. Almost no game has any 8 core optimisation at all.
The reason why you perform good in heaven and firestrike is simpy because they're synthetic with barely any CPU load at all.
There are some things to validate this theory.
Try Dragon Age Inquisition, The Crew, Watch Dogs, GTA V and Assassin's Creed Unity- if neither oft these gains anything from CFX its very likely that you got a CPU bottleneck.
Try Crysis 3 Witcher 3 and Tomb Raider 2013. If they have an almost linear scaleing from 1 card to 2 cards, then your Problem is not GPU related.

[alphaproject]

This is why I don't trust anything Xfire.  A game did that to me once and ever since then, just not dealing with it.

 

It's always up to the drivers or game devs.... you never know if it's going to be GREAT... or suck....

[EJ993]

Hello all, back again. Still having these issues but I have some more info that might help solve this, as I doubt I can be the only one? 

 

To be clear, this is on every game (except Battlefield 4 as stated above). I get identical frames in windowed AND full screen. (so with crossfire enabled and disabled)

 

Fire Strike benchmark with new Win 10 driver (yes, had the issue in Win 8 too if not worse) http://www.3dmark.com/fs/5291653

 

Below is my 1440p performance on GTA V, not acceptable for my £500 card at all. 

 

Not CPU issue, not GPU issue (works fine in benchmarks)... I can't understand. Anyone? Thankya!

 

Firestrike score is definitely too low. I get 12900 with overclocked 980, also running latest win10 build.