COD:AW Benchmarks

[Flapdrol1337]

Aren't there quite a few disabled units on the 970 compared to the 980 ? A higher percentage than we've seen in the past comparing the top to second top card ?

Yeah, but they both have the same memory bandwidth, and the core clocks much better, if you overclock both there's only ~10% performance difference or something.

 

[amenx]

Re the Guru3d benches, the reviewer states the following:

However, no matter how we benchmark the results remain a little unreliable...

Call of Duty Advanced Warfare will NOT make it into our regular benchmark suite for videocards. As stated in the intro, there are many things odd and off. As such we recommend you to look at the performance benchmarks you've just read with a grain of salt as I have been at the verge and threshold of asking myself the question whether or not to post these results. One test run will result on 50 FPS, the second 46 FPS and another 56 FPS. This means that the results shown today are indicative, not a precise measurement.

Could apply to the other early reviews as well.

 

[raghu78]

Could apply to the other early reviews as well.

The game needs a patch to fix the performance inconsistencies and to enable robust SLI and CF support.

 

[RussianSensation]

Yeah, but they both have the same memory bandwidth, and the core clocks much better, if you overclock both there's only ~10% performance difference or something.

980 maintains a 16-18% lead when overclocked against a max overclocked 970.

http://www.computerbase.de/2014-09/geforce-gtx-980-970-test-sli-nvidia/12/

And sometimes it can even grow to 21% against an overclocked 970. In cases where the game is shader and texture limited/dependent, even a max overclocked 970 cannot touch a stock after-market 980. The more shader/texture limited the game is, the faster 980 becomes over 970.

That's why GM200 with 2800-3200 CUDA cores will be beast mode.

 

[boozzer]

Huh?

I stand corrected

 

[Flapdrol1337]

980 maintains a 16-18% lead when overclocked against a max overclocked 970.

http://www.computerbase.de/2014-09/geforce-gtx-980-970-test-sli-nvidia/12/

And sometimes it can even grow to 21% against an overclocked 970. In cases where the game is shader and texture limited/dependent, even a max overclocked 970 cannot touch a stock after-market 980. The more shader/texture limited the game is, the faster 980 becomes over 970.

That's why GM200 with 2800-3200 CUDA cores will be beast mode.

Don't show the other chart

 

[frozentundra123456]

Here is another benchmark from computerbase.de

http://www.computerbase.de/2014-11/...diagramm-grafikkarten-benchmarks-in-2560-1600

GTX 980 is once again the indisputable leader at all resolutions with a clear margin of 17 - 20% ahead of the next card at all resolutions. GTX 970 edges out R9 290X at 1600p and at 4k its vice-versa. Surprisingly the GTX 780 Ti falls behind the R9 290X at 1600p and even the R9 290 at 4k. Even the guru3d charts show the R9 290X beating the 780 Ti at 1600p and 4k.

It is even worse if you look at the results at 1080p. With an i7 nvidia is 30% faster and something like 50% faster with an FX. Definitely looks like much lower CPU overhead with the 980. I recall an article a couple years ago in TH that showed the same effect, but nowhere close to this extent. Looks like AMD needs a driver update desperately.

 

[Silverforce11]

Surprisingly the GTX 780 Ti falls behind the R9 290X at 1600p and even the R9 290 at 4k. Even the guru3d charts show the R9 290X beating the 780 Ti at 1600p and 4k.

This has been common in recent games, it seems as though NV is not focused on Kepler optimizations in its drivers since Maxwell launched because the R290/X is steadily creeping up to match or exceed the 780/ti while the 970/980 maintains its lead with the 970 matching 780ti (which did not occur during launch).

 

[RaistlinZ]

Anyone try SLI with the new drivers yet?

 

[SolMiester]

I'd say it's the opposite. A lot of these reviews are using reference models, and the clock speed on the reference GTX 970 is much lower than on the reference GTX 980..

When you use aftermarket parts though, the performance gap shrinks significantly..

There is NO reference 970!...they are all AIB partner models with different speeds.

 

[guskline]

Anyone try SLI with the new drivers yet?

Works great on my 670s in SLI

 

[f1sherman]

It is even worse if you look at the results at 1080p. With an i7 nvidia is 30% faster and something like 50% faster with an FX. Definitely looks like much lower CPU overhead with the 980. I recall an article a couple years ago in TH that showed the same effect, but nowhere close to this extent. Looks like AMD needs a driver update desperately.

This is worse than Nvidia's DX11 vs AMD Mantle:

http://gamegpu.ru/images/remote/htt...Duty_Advanced_Warfare-test-cod_proz_intel.jpg
http://gamegpu.ru/images/remote/htt...f_Duty_Advanced_Warfare-test-cod_proz_amd.jpg

I clearly remember the opposite situation pre-270.something NVIDIA drivers.
With Fermi cards being more cpu hungry than Radeons.
From personal experience, and there had been even few tests, on this back then esoteric subject - CPU scaling/utilization NV vs AMD.

 

[wilds]

This game looks pretty good and runs well, but some of the textures are alarmingly low-res. It's a fun game I guess, Kevin Spacey carries the story line.

 

[psolord]

This is worse than Nvidia's DX11 vs AMD Mantle:

http://gamegpu.ru/images/remote/htt...Duty_Advanced_Warfare-test-cod_proz_intel.jpg
http://gamegpu.ru/images/remote/htt...f_Duty_Advanced_Warfare-test-cod_proz_amd.jpg

I clearly remember the opposite situation pre-270.something NVIDIA drivers.
With Fermi cards being more cpu hungry than Radeons.
From personal experience, and there had been even few tests, on this back then esoteric subject - CPU scaling/utilization NV vs AMD.

That's why I got so excited a few posts above.

I still have my Fermis and do test games on them regularly. I have never witnessed anything that would suggest nvidia has better cpu overhead.

I must be something with their latest series. I still don't see better cpu overhead with the newer drivers (but I will keep looking).

I wish some site would do some proper testing on many games, regarding cpu overhead.

 

[desprado]

That's why I got so excited a few posts above.

I still have my Fermis and do test games on them regularly. I have never witnessed anything that would suggest nvidia has better cpu overhead.

I must be something with their latest series. I still don't see better cpu overhead with the newer drivers (but I will keep looking).

I wish some site would do some proper testing on many games, regarding cpu overhead.

Both of these are cpu depended and see how Nvidia dominate in this.
http://www.pcper.com/reviews/Graphi...DIA-33750-Scaling-Demonstrated-Star-Swarm-AM1

http://www.anandtech.com/show/8640/benchmarked-civilization-beyond-earth/2

 

[caswow]

Both of these are cpu depended and see how Nvidia dominate in this.
http://www.pcper.com/reviews/Graphi...DIA-33750-Scaling-Demonstrated-Star-Swarm-AM1

http://www.anandtech.com/show/8640/benchmarked-civilization-beyond-earth/2

no. mantle civ is way smoother than any dx rendered gameplay.

 

[desprado]

no. mantle civ is way smoother than any dx rendered gameplay.

No and Yes.

yes it is better than AMD DX11 and we all know but mantle experience is same as Nvidia Dx11.

 

[Silverforce11]

No and Yes.

yes it is better than AMD DX11 and we all know but mantle experience is same as Nvidia Dx11.

Wrong. Mantle Civ has much higher Minimum FPS which occur frequently in late game map spanning and zoom in/out.

NV's DX11 is more efficient than AMD with less CPU overhead, this is well known. Credit where it's due for sure!

 

[caswow]

No and Yes.

yes it is better than AMD DX11 and we all know but mantle experience is same as Nvidia Dx11.

not too long ago fps was not the first thing to look on because nvidia had a smear campaign against amd because their fps were not "smooth" enough. nobody cares about smoothness anymore but about the fps.

amd has to work on their dx11 drivers, but nvidia cant match the smooth gameplay of mantle rendered gameplay.

 

[stahlhart]

This thread is about Call Of Duty: Advanced Warfare, not Civilization: Beyond Earth. Stay on topic.
-- stahlhart

 

[desprado]

Well this pic explains all that why Nvidia is much better with lower end cpus.

However, we can report that we see the same issue to different degrees on other AMD GPUs too, though the lower down the stack you go, the less noticeable the impact (as you're hitting GPU, rather than CPU limitations) - as we saw when testing with an R7 265, essentially an overclocked version of the classic Radeon HD 7850. Generally speaking though, if you're running anything at the R9 270 level or better, we'd recommend an Intel quad-core processor or the equivalent for best performance with an AMD card, while a Core i3 is potent enough to deal with Nvidia equivalents in this performance range.

http://www.eurogamer.net/articles/digitalfoundry-2014-call-of-duty-advanced-warfare-face-off

 

[Dankk]

Singleplayer seems to run pretty smooth on my rig from what I've seen, but multiplayer is a massive stuttery mess. Have to turn everything down to low in order to get playable framerates, and even then sometimes I can't get a stable framerate. Still waiting for an AMD driver update, or a game patch, or both.

 

[hawtdawg]

From what I've played so far, in single player at least, it runs comparably to BF4 and isn't nearly as visibly impressive. It also uses a ton of ram.

 

[forestpest]

Anandtech and none of you mentioned frametimes ?

http://www.hardwarepal.com/call-duty-advanced-warfare-benchmark-performance/

"Again with the 2560×1440 results the R9-290 is behind the GTX970 in frame rate but better in terms of frame times"

 

[3DVagabond]

Anandtech and none of you mentioned frametimes ?

http://www.hardwarepal.com/call-duty-advanced-warfare-benchmark-performance/

"Again with the 2560×1440 results the R9-290 is behind the GTX970 in frame rate but better in terms of frame times"

Apparently, that's not as important as it used to be. /sarc

The overwhelming use and reporting of FPS (and LowRes is all the rage even more than it typically is with nVidia) as the only measurable metric (besides efficiency, of course) is astounding. People reviewing and buying nVidia no longer show any concern at all.