Dragon Age Inquisition benchmarks.

R

Rizen

[H]F Junkie
Joined
Jul 16, 2000
Messages
9,487
Game is unlocked in South Korea as of 10AM EST. If you are in SK, or Origin thinks you are (hint), you can play now.

I just benchmarked the game using the built in benchmark under "Extras". It is very demanding on max. There is a texture setting above Ultra: "Fade Touched". Here are my results. My system is in my signature. I am running 14.11.1 Beta Catalyst drivers and I have a light overclock on my 290Xs (1065 core).

Below frame rates are with VSync on, turn it off for higher averages.

290X CROSSFIRE SCORES
2560x1440, no MSAA, Ultra, High Post Process, "Fade Touched" Textures:
Mantle API = 57fps avg, 55 min
DirectX API: 59 fps avg, 39 min

2560x1440, 2X MSAA, Ultra, High Post Process, "Fade Touched" Textures:
Mantle API = 52fps avg, 47 min
DirectX API = 53fps avg, 35 min

2560x1440, 4X MSAA Ultra, High Post Process, "Fade Touched" Textures:
Mantle API = 43fps avg, 39min
Mantle API VSYNC OFF: 52fps avg, 40.6 min
DirectX API: 48 avg, 31-34min (min was inconsistent).

SINGLE 290X SCORES
2560x1440, 4X MSAA, Ultra, High Post Process, "Fade Touched" Textures:
Mantle API = 29fps avg, 23.4fps min
DirectX API = 28.6fps avg, 23.6fps min

1920x1080, 4X MSAA, Ultra, High Post Process, "Fade Touched" Textures:
Mantle API = 45fps avg, 39.4fps min
DirectX API = 42.2fps avg, 34.6fps min

With "Fade Touched" textures and 4X MSAA I see 3.8GB VRAM usage in DirectX and it shows as 4.4GB VRAM usage in Mantle. Not sure if that's accurate, or how it will perform in-game, but the benchmarks above are accurate - I ran them multiple times. I also tested Ultra versus Fade Touched textures in Mantle and it does not make any impact on frame rates - tested multiple times, they are identical.

THESE ARE NOT GAMEPLAY BENCHMARKS, JUST IN GAME TOOL. Don't want to spoil the game until I can sit down after work and get into it. The game seems to use all 4 logical cores + virtual cores, I was seeing 30-50% load across all threads with 80%+ on a few of the logical cores.

Post your results here.
 
Last edited:
Yes I could, but there's not much of a point. It's basically a timedemo and it plays through as quickly as possible, so you can't see much. I'm also technically at work (I work at home, so I was able to run some tests while working) but I can't spend the time to upload the videos right now :(
 
Afterburner 4.0.0. It's not perfect and probably not accurate for Mantle. Is there another tool that works better for that?
 
Rizen said:
Afterburner 4.0.0. It's not perfect and probably not accurate for Mantle. Is there another tool that works better for that?
Click to expand...

I have found Aida64 to work with Mantle for vram usage. Maybe you could confirm the Afterburner results with it (if you use the trial hopefully this is not a result that will be hidden)
 
Dual 290Xs and you can't break 60fps at 1440p... these games are really stressing new hardware! Hopefully DA:I isn't bug ridden like AC:U.
 
Okay, using Mantle and at 2560, Ultra, 4X MSAA and "Fade Touched" textures, AIDA64 shows 7.5GB memory usage during the benchmark. I assume that the tool is adding both GPU memory pools together?
 
i hope crossfire is broken with this game, otherwise those numbers are terrible.
 
Odellus said:
i hope crossfire is broken with this game, otherwise those numbers are terrible.
Click to expand...
Can't say for certain, but my friend has a very similar system to me with a single 290X and I asked him what his numbers are. He ran a benchmark using Mantle, Ultra everything, 4X MSAA, Fade textures and got 43.6fps average, with minimum just below that. So it seems CrossFire is working, since I have about the same performance with 2X 290s but I am running 2560. The game is demanding. Welcome to games that aren't entirely held back by consoles :)
 
Rizen said:
Okay, using Mantle and at 2560, Ultra, 4X MSAA and "Fade Touched" textures, AIDA64 shows 7.5GB memory usage during the benchmark. I assume that the tool is adding both GPU memory pools together?
Click to expand...

Yes, devide vram by 2, which looks like it corresponds with Afterburner.

Curious on GPU utilization, are they all running at 100%?
 
Yakk said:
Yes, devide vram by 2, which looks like it corresponds with Afterburner.

Curious on GPU utilization, are they all running at 100%?
Click to expand...
The benchmark makes it somewhat hard to tell what actual utilization is, but I did see both cards hit 100% a few times. Typically one card is at 100% and the other is up and down a bit in the benchmark, I imagine actual gameplay will have better utilization since the scene changes aren't so fast.
 
Rizen said:
Can't say for certain, but my friend has a very similar system to me with a single 290X and I asked him what his numbers are. He ran a benchmark using Mantle, Ultra everything, 4X MSAA, Fade textures and got 43.6fps average, with minimum just below that. So it seems CrossFire is working, since I have about the same performance with 2X 290s but I am running 2560. The game is demanding. Welcome to games that aren't entirely held back by consoles :)
Click to expand...

Waaaaaoooooooow (Hi, I'm the referenced friend)

My numbers were at 1080p and I only had time to run one test this morning. I'll update with more numbers when I get a chance. I have a 290 on the way to crossfire with the 290x, but I imagine there will be plenty of benchmarks from others by the time that gets here (Wednesday).
 
Single GPU scores posted for 2560x1440, 1920x1080 in both DirectX and Mantle.
 
subjetive opinion in smooth gameplay?.. results are pretty interesting.. because DirectX have higher minimums (even if are barely) but Mantle have Higher minimums. it have any kind of framerate graphics? to see those spikes.. im interested in single GPU results also.. i know now my 280X will have problems at 1080P.. i'll buy soon and see how it compare with my GTX 780..
 
Haven't actually played any gameplay yet. I have heard that the benchmark seems to score lower than what people are seeing in game, for what it's worth. I'll update this thread with my subjective opinions once I play the game tonight and I'll let you know what settings I end up using in-game.
 
Rizen said:
Can't say for certain, but my friend has a very similar system to me with a single 290X and I asked him what his numbers are. He ran a benchmark using Mantle, Ultra everything, 4X MSAA, Fade textures and got 43.6fps average, with minimum just below that. So it seems CrossFire is working, since I have about the same performance with 2X 290s but I am running 2560. The game is demanding. Welcome to games that aren't entirely held back by consoles :)
Click to expand...

huh? this game came out even on last gen consoles, and it really doesn't look good enough to justify that kind of performance judging by youtube videos.

edit:

i mean i really just can't see how a game that looks like this can't even hit 60 fps with 290X CF, even at 4K, let alone 1440p:

http://i.imgur.com/FmsHdR4.jpg
http://i.imgur.com/UsrKI9K.jpg
http://i.imgur.com/hPEomao.jpg
 
Last edited:
Odellus said:
huh? this game came out even on last gen consoles, and it really doesn't look good enough to justify that kind of performance judging by youtube videos.

edit:

i mean i really just can't see how a game that looks like this can't even hit 60 fps with 290X CF, even at 4K, let alone 1440p:

http://i.imgur.com/FmsHdR4.jpg
http://i.imgur.com/UsrKI9K.jpg
http://i.imgur.com/hPEomao.jpg
Click to expand...
You really need to stop using Youtube videos to judge visual quality of a game. Even JPG stills are going to have image compression which results in a loss of quality. In either case, the in-game visuals from what I have seen look significantly better than those screenshots. Part of the reason the FPS is low is because of 1) MSAA 2) extremely long draw distance 3) tons of post-processing effects.

I haven't tested it on lower end machines yet (I have a 4690k/R9 285 box here too) but I have heard it scales very well across lots of different configurations, AMD and NVIDIA alike, and that the visual differences between Low and Ultra ramp nicely. Typically, the quality differences are not as noticeable between say, High and Ultra as they are between Low and High, so even if you turn it down somewhat I would imagine it will still look very good.
 
Also, I have heard that the game is CPU dependent as well. My benchmarks would suggest this is true. In the single 290X benchmarks at 2560x1440 you can see that DirectX and Mantle are very close. I'm obviously GPU bound. But when you add a second 290X or you drop down to 1080p, Mantle has a significantly higher minimum frame rate over DirectX, which would suggest that the CPU is becoming more of a bottleneck.
 
That german site that had their review up early said it looks fantastic, so i wouldnt go by the youtube videos, they are compressed afterall.
 
seanclayton said:
Good thing i didnt buy gtx980, knowing that it wont last long due to low 4gb of VRAM.
Click to expand...

lmao, people make comments like this every generation. people were freaking out about the 2gb on my 670 too, was never an issue


I wonder how this runs without the MSAA I bet you get an extra 10 fps on average
 
youtube is a perfectly fine way to see how a game looks. the only thing you can't really judge from youtube videos are textures. environment quality, lighting, models, effects, etc. do not suffer too much from the effects of compression, at least not to the point where what you see is completely unlike the actual game. high quality jpg screenshots are indistinguishable from lossless screenshots (or are you really going to tell me you can see a difference between these two?), and even lower quality ones are still way more than enough to properly represent a game.

anyway, i'm downloading it now and i will see for myself. i have higher standards than most, so i can sort of see why you would respond like that, but to me the game looks significantly worse than battlefield 4 (same engine), and the framerate does not reflect that at all.
 
Sorry, but I don't think Youtube is an accurate comparison. The quality is nowhere near as good as it is on my PC whenever I view a youtube video even in HD. As for lossless compression and JPGs, I don't see a difference there, but you can't really be sure how much compression people are using in their screenshots unless A) they use lossless formats or B) you've taken them yourself.

Let me know what you think. I'll be playing the game shortly.
 
Phlorge said:
lmao, people make comments like this every generation. people were freaking out about the 2gb on my 670 too, was never an issue


I wonder how this runs without the MSAA I bet you get an extra 10 fps on average
Click to expand...

around 25% difference 15fps or so depending on resolution.
 
So, I had VSync on in my original benchmarks - disabling VSync improves your average FPS quite a bit. It seems like even though I have the game running at 120Hz on my ROG Swift, that VSync may have been capping the renderer at 60fps anyways.

I just re-ran the benchmark on my PC with VSync OFF, at 2560 Ultra 4X MSAA, High Post Process, Fade Textures, Mantle API and got 52fps avg 40.9fps min.
 
Rizen said:
Game is unlocked in South Korea as of 10AM EST. If you are in SK, or Origin thinks you are (hint), you can play now.

I just benchmarked the game using the built in benchmark under "Extras". It is very demanding on max. There is a texture setting above Ultra: "Fade Touched". Here are my results. My system is in my signature. I am running 14.11.1 Beta Catalyst drivers and I have a light overclock on my 290Xs (1065 core).

290X CROSSFIRE SCORES

2560x1440, 2X MSAA, Ultra, High Post Process, "Fade Touched" Textures:
Mantle API = 52fps avg, 47 min
DirectX API = 53fps avg, 35 min

2560x1440, 4X MSAA Ultra, High Post Process, "Fade Touched" Textures:
Mantle API = 43fps avg, 39min
DirectX API: 48 avg, 31-34min (min was inconsistent).
Click to expand...

And now for GTX 970's at 1516MHz core in SLI :D

2560x1440, 2X MSAA, Ultra, High Post Process, "Fade Touched" Textures:
DirectX API = 72fps avg, 55.8fps min

2560x1440, 4X MSAA Ultra, High Post Process, "Fade Touched" Textures:
DirectX API: 61.9fps avg, 45.7fps min

344.65 Drivers.
 
pandora's box said:
And now for GTX 970's at 1516MHz core in SLI :D

2560x1440, 2X MSAA, Ultra, High Post Process, "Fade Touched" Textures:
DirectX API = 72fps avg, 55.8fps min

2560x1440, 4X MSAA Ultra, High Post Process, "Fade Touched" Textures:
DirectX API: 61.9fps avg, 45.7fps min

344.65 Drivers.
Click to expand...
Very nice. I assume you shut off VSync as well?

I'm curious to see some CPU comparisons for this game. I wonder if having an LGA 2011 CPU is a noticeable boost - although the 4820 is 4-core as well as the 4770K.
 
Rizen said:
Very nice. I assume you shut off VSync as well?

I'm curious to see some CPU comparisons for this game. I wonder if having an LGA 2011 CPU is a noticeable boost - although the 4820 is 4-core as well as the 4770K.
Click to expand...

Yes, Vsync was off.
 
pandora's box said:
And now for GTX 970's at 1516MHz core in SLI :D

2560x1440, 2X MSAA, Ultra, High Post Process, "Fade Touched" Textures:
DirectX API = 72fps avg, 55.8fps min

2560x1440, 4X MSAA Ultra, High Post Process, "Fade Touched" Textures:
DirectX API: 61.9fps avg, 45.7fps min

344.65 Drivers.
Click to expand...

Did you have to force SLI? I'm only getting like 35fps average with 970s.
 
Except for the AA, are the rest of the settings at maximum?

The benchmark doesn't seem to indicate the game, in technical terms, runs as bad as some games
 
Yes, my #s are with every single setting in the game completely maxed out. I just started playing just now - it looks amazing. By far the best RPG graphics I have ever seen in a game. Also, in game seems very very smooth, smoother than what the benchmarks would indicate. I don't know if it's Mantle, which typically has very tight frame times, or if the FPS is just much higher, but it feels great.
 
You must log in or register to reply here.
Back
Top