Dragon Age: Inquisition GPU & CPU Performance Review

[Guest]

Could you guys in the future test on the more common resolutions please?

I see more people with 1920x1080 and 2560x1440 (or maybe even just downsampling) than with 1920x1200 and 2560x1600.

Also, if possible, try including 4K tests in the future with SLI 980x and CrossFire 290X (since arguably AMD still has the upper hand in 4k performance) and I want too see some valid results to confirm what I suspect to be the performance leader currently at that resolution.

1920x1200 is a better resolution than 1080p. It is a shame computer monitors use the 16:9 aspect ratio on all the cheap models. In any case is it really a big issue for you, we are talking 1 or 2fps at most going from 1920x1200 to 1920x1080.

 

[Pastuch]

Steve where are the Mantle benchmarks? Why are you handicapping the AMD cards with DirectX?

 

[Steve]

By this benchmark I think it goes without saying that a R9 290 and an FX 8320 are the best bang for the buck for maxing out this game with breaking the bank.

Question: Wasn't DA:I supposed to support Mantle as well? That would put AMD GPUs and CPUs in an even better position.

It might help the AMD CPU’s a little but it will do nothing for the AMD GPU’s.

Steve where are the Mantle benchmarks? Why are you handicapping the AMD cards with DirectX?

There seems to be a lot of confusion about what Mantle does. Let me make it clear, the AMD GPU’s are in no way at a disadvantage because we didn’t test with Mantle. Mantle is an API that removes CPU overhead, it doesn’t make GPUs faster.

We used a high-end Intel CPU so at 1920x1200 and 2560x1600 the GPU was the bottleneck, not the CPU.

Mantle is only useful for those using a low-end processor such as a Phenom II with a high-end GPU such as the R9 290 or R9 290X.

We have already conducted in-depth testing with Mantle in games such as Battlefield 4 and Thief. Those using a Core i5, Core i7 or in many cases even a Core i3 have nothing to gain from using Mantle, unless they are running at low resolutions or using low quality settings.

So an i3 definitely can't run this game? Gamegpu had some i3s in their results and I was planning to get a 4360. This review is saying it won't work so I'm a little confused.

Core i3 should work since it offers 4 threads. We weren't able to test the Core i3's in time due to the issues I was having with Origin.

 

[dividebyzero]

There seems to be a lot of confusion about what Mantle does. Let me make it clear, the AMD GPU’s are in no way at a disadvantage because we didn’t test with Mantle. Mantle is an API that removes CPU overhead, it doesn’t make GPUs faster.

Indeed. Mantle doesn't seem overly consistent in its application either. Referencing the GameGPU review you mentioned earlier, the 290X shows a modest gain at 25x16, and a small loss at 19x10. The HD 7970/280X also shows a slight decrease in framerate in two of three resolutions (the other being equal) using Mantle vs DX11. These would tend to indicate that the game dev and AMD still have some optimization ahead of them.

All in all, I think AMD would be happy for everyone to bench with DX11 based on these results.

 

[Steve]

Indeed. Mantle doesn't seem overly consistent in its application either. Referencing the GameGPU review you mentioned earlier, the 290X shows a modest gain at 25x16, and a small loss at 19x10. The HD 7970/280X also shows a slight decrease in framerate in two of three resolutions (the other being equal) using Mantle vs DX11. These would tend to indicate that the game dev and AMD still have some optimization ahead of them.

All in all, I think AMD would be happy for everyone to bench with DX11 based on these results.

The problem for us is testing Mantle means doing almost everything twice and obviously given the work that already went into this article that isn’t possible.

If there was a real need to test Mantle then we would but we already know what Mantle does.

While it is great sites likes GameGPU have included some Mantle results they are mostly pointless.

They tested Mantle using the world’s fastest desktop CPU, the Core i7-5960X. As I said Mantle has nothing to offer Core i5 and Core i7 users providing they are using appropriate quality settings. We even found in many cases when testing BF4 and Thief that Mantle offers Core i3 users very little.

So in order to include useful Mantle results we would have to test all the AMD cards in both DX11 and Mantle using a huge range of CPUs.

More over if you haven’t been a wally and paired a high-end GPU with a low-end CPU then you don’t need to worry about Mantle in the first place.

Why no FX-9370 or FX-9590?

I get the FX-8350 was overclocked to 4.5Ghz but the FX-9590 is faster again out of the box and would have been interesting to see the results

I must thank you Sgt Bilko, AMD is now sending us an FX-9590 so you can expect that to be in our next performance article.

 

[Alpha Gamer]

Wow, awesome article, thanks for the marathon! I'm not sure, but I'd say your numbers are substantially better than mine. Maybe there's something wrong with my gtx 980, I could be imagining things, though. I'll only be sure on Friday, when I'll next get to use my rig.

info about how the graphic quality goes down by reducing Ultra to High and how that impacts the FPS.

I'd like this info as well.

 

[Sgt Bilko]

Indeed. Mantle doesn't seem overly consistent in its application either. Referencing the GameGPU review you mentioned earlier, the 290X shows a modest gain at 25x16, and a small loss at 19x10. The HD 7970/280X also shows a slight decrease in framerate in two of three resolutions (the other being equal) using Mantle vs DX11. These would tend to indicate that the game dev and AMD still have some optimization ahead of them.

All in all, I think AMD would be happy for everyone to bench with DX11 based on these results.

I'm running a FX-9590 and a R9 295x2 at 1440p, in DX I see around 40-50fps avg and in Mantle I'm running 50-60 avg so Mantle is working for me and with a Positive effect.

I must thank you Sgt Bilko, AMD is now sending us an FX-9590 so you can expect that to be in our next performance article.

You're Welcome I guess?

Not sure what I had to do with it exactly but it's good you are getting one, they are a nice chip to be sure

 

[poohbear]

Correction, the ultra preset doesn't put EVERYTHING to their maximum settings, "Fade touch" is the maximum setting for the texture option, not Ultra. It does make a difference in the crispness of the textures, with ultra textures they didn't look modern, with fade touch they look much better.

 

[GhostRyder]

Wow you really removed all the limiters on this article and went all in. There is such a wide range you almost cannot hit a setup here that does not at least give you an idea of what to expect from your machine. I would also say it seems this game is actually decently optimized for all range of products instead of being driver limited like some recent games have been. I am actually surprised at how well the range of processors at 1080p and GPU's at 1080-1600p performed given the games demanding nature.

Though I am shocked that no dual core would even allow the game to run period. That seems to be quite a shocking result, I suppose because you did not include the i3 @Steve that the hyper threading made no difference in it running on it? Odd enough that a game like this would come out and pose such a problem though because that ruins the fun of using the Pentium AE processor for sure.

Great review!

 

[Keithian]

Could you guys in the future test on the more common resolutions please?

I see more people with 1920x1080 and 2560x1440 (or maybe even just downsampling) than with 1920x1200 and 2560x1600.

Also, if possible, try including 4K tests in the future with SLI 980x and CrossFire 290X (since arguably AMD still has the upper hand in 4k performance) and I want too see some valid results to confirm what I suspect to be the performance leader currently at that resolution.

1920x1200 is a better resolution than 1080p. It is a shame computer monitors use the 16:9 aspect ratio on all the cheap models. In any case is it really a big issue for you, we are talking 1 or 2fps at most going from 1920x1200 to 1920x1080.

Sorry I agree with the guest above. Testing should be at common resolutions (1440P, 1080P) so assumptions don't need to be made. Regardless, the results were very interesting and gave a sense of what it takes at multiple levels so I loved the article. The only thing I don't get as I've ready several posts at other forums of folks running 4K with 980s in SLI with a much higher FPS so not sure why the author is only seeing 30s. I guess others may have done some tweaking who knows. Regardless, I've already concluded that the only way I would buy this game and a corresponding new PC (which I need soon) would be when Nvidia releases the next card like a 980ti or the 990. The PS4 plays this game brilliantly on a PC monitor with just about zero issues. A PS4 is equivalent to a PC on High Settings (developers confirmed) and to spend a few thousand with dual 980s, a high end process, a large SSD drive, new Gsync monitor in 4K just for a little better texture or more grass is a total waste of money. Maybe with the next Nvidia card and optimized drivers and Bioware patches, the purchase will be worth it in the first quarter of next year...again...only for 4K if you are already going to spend the money because I'm playing perfectly on a PS4 for $60 lol.

 

[DAOWAce]

DRM everyone. The best thing for consumers.

 

[Divine_Assassin]

This is another game that shows AMD's cards outperforming similarly- or better-specced NVIDIA cards. The other one being Far Cry 4. It's surprising to see a Radeon HD 7950 Boost performing the same as a GeForce GTX 770 at 1920x1200. And take a look at the R9 280X performing the same as a GTX Titan.

Also, the quad-core CPU requirement seems unfair to me. The Core i3 is a very capable CPU when it comes to gaming. Battlefield 4 has no issues running on a CPU with 2 cores so why is it an issue with Dragon Age: Inquisition?

This has more to do with the fact that Kepler is weak at Compute which is used extensively in Far Cry 4 for all the candy effects. If you look at the benchmarks both GCN and Maxwell do very well in this game because they are absolute beasts at compute.

 

[Jad Chaar]

Wow AMD has their work cut out for them... their CPUs are performing horribly in newer titles. It is either that or the game is poorly optimized I really do not know. It is definitely time for an update though.

 

[Divine_Assassin]

This is another game that shows AMD's cards outperforming similarly- or better-specced NVIDIA cards. The other one being Far Cry 4. It's surprising to see a Radeon HD 7950 Boost performing the same as a GeForce GTX 770 at 1920x1200. And take a look at the R9 280X performing the same as a GTX Titan.

Also, the quad-core CPU requirement seems unfair to me. The Core i3 is a very capable CPU when it comes to gaming. Battlefield 4 has no issues running on a CPU with 2 cores so why is it an issue with Dragon Age: Inquisition?

It would seem more and more games are shifting game effects from being done with raw shader power to Compute which is why Kepler based GPUs are lagging behind due to weak GPGPU performance.

 

[Steve]

Wow AMD has their work cut out for them... their CPUs are performing horribly in newer titles. It is either that or the game is poorly optimized I really do not know. It is definitely time for an update though.

AMD's CPU department has had their work cut of for quite some time now mate, that's old news There is a reason why their FX-8350 is priced along side the cheapest Core i5.

 

[amstech]

Just wanted to say, first time I commented I skimmed over the article then wrote some junk about 4K. But earlier this evening on my tablet I re-read this and what you went through with the systems and hard drives is great. I laughed harder as I read more.
As an experienced professional IT Tech its amazing sometimes some of the issues we run across throughout the years.
Nice work staying diligent and focused throughout that escapade!

 

[emirhasa]

Wow AMD has their work cut out for them... their CPUs are performing horribly in newer titles. It is either that or the game is poorly optimized I really do not know. It is definitely time for an update though.

Well with everything maxed out it the AMD CPU's are even better than those i5's and i7's sometimes.

Also keep into consideration how much cheaper AMD is than Intel. You get what you pay for. For it's price, AMD is excellent, and offers better performance per $ ratio.

 

[Eddo22]

"Forget about 4K gaming unless you have a few (more than two) high-end GPUs as the GTX 980 averaged just 37fps at 2560x1600 while the R9 290X offered 35fps."

I calling you out on this again. What you have up there is not proof and no way to back up a statement like that.

The other thing is that you don't need AA when using 4k and you can numb down the settings slightly and still have a great time gaming without sacrificing much in the way of looks.

 

[amstech]

"Forget about 4K gaming unless you have a few (more than two) high-end GPUs as the GTX 980 averaged just 37fps at 2560x1600 while the R9 290X offered 35fps."

I calling you out on this again. What you have up there is not proof and no way to back up a statement like that..

Not the sharpest knife in the drawer are ya?

 

[Eddo22]

"Forget about 4K gaming unless you have a few (more than two) high-end GPUs as the GTX 980 averaged just 37fps at 2560x1600 while the R9 290X offered 35fps."

I calling you out on this again. What you have up there is not proof and no way to back up a statement like that..

Not the sharpest knife in the drawer are ya?

You probably shouldn't have said anything because you clearly don't know what I'm referring to.

 

[plonk420]

Why no FX-9370 or FX-9590?

I get the FX-8350 was overclocked to 4.5Ghz but the FX-9590 is faster again out of the box and would have been interesting to see the results

Why no FX-9370 or FX-9590?

I get the FX-8350 was overclocked to 4.5Ghz but the FX-9590 is faster again out of the box and would have been interesting to see the results

Wow AMD has their work cut out for them... their CPUs are performing horribly in newer titles. It is either that or the game is poorly optimized I really do not know. It is definitely time for an update though.

AMD's CPU department has had their work cut of for quite some time now mate, that's old news There is a reason why their FX-8350 is priced along side the cheapest Core i5.

is an i7 Nehalem or Haswell Pentium or i3 out of the question? I'm currently on a dying former, looking to go to the latter (my budget is only currently allowing for a Pentium at the moment) or just get another Nehalem mobo ... and wondering how much an i3 will benefit me in games...

 

[Cryio]

I have been playing Dragon Age Inquisition at 1440p ultra with an FX 8350 and I just upgraded to an Intel 4790k (not in your benchmark), but I can say for sure at higher resolutions my FX 8350 barely pushed over 30 fps with a GTX 980. When I upgraded to the Intel I was pushing another 10 FPS at least in the benchmark which is about 20fps with regular gameplay as the benchmark is a stress test of sorts.

I have been an AMD guy for the past ten years or so, but if you are going to play over 1080p, you really should get an Intel if you don't already have one.

Or if you had a R9 290, 290X or 295x2, running that FX 8350 with Mantle (especially with the latest patch and AMD driver update) would have net you the same performance as with the Intel one at a significant lower cost.

Plus, once you ran into a GPU limitated situation, it wouldn't have mattered if you had a 8350 or a 4790.

 

[Jad Chaar]

is an i7 Nehalem or Haswell Pentium or i3 out of the question? I'm currently on a dying former, looking to go to the latter (my budget is only currently allowing for a Pentium at the moment) or just get another Nehalem mobo ... and wondering how much an i3 will benefit me in games...

Make a separate post on the forums and we can help you out!

 

[DanUK]

I've been really enjoying this game although I've been suffering quite a lot of bluescreens which was odd. I can run the game 1920x1200 ultra with no slowdown or lag. Realised last night that this game is massively cpu intensive, and my cpu was overheating :S I have a corsair h80i cooler on an Core i7 4790k, with 16 gig ram and a gtx 970 (which doesnt go above 60 degrees). It's because I have the fans on the h80i set to quiet mode (anything else and the fans sound like a plane taking off). A bit annoying that for this game only whenever I want to play it I have to crank the fan speed up in fear of a meltdown! Never had this problem with any other games, not sure if thats "just the way it is" or if this game is "poorly optimised" or whatever. I may just buy two noctua nf-f12 fans for the h80i that will hopefully be able to give me performance and quietness.

I guess this is just a heads up to anyone else.

 

[Steve]

Does your system crash if you run a program like Prime95 with the H80i set to silent? If not then heat isn’t the issue.

Unless your fan on the H80i is spinning abnormally slow the system shouldn’t be crashing due to heat with a stock Core i7-4790K. Either the water-block isn’t seated correctly or not enough thermal paste has been applied.

Anyway I would diagnose this with a program such as Prime95 in windows and monitor the temperature as you do.