Ashes of the Singularity User Benchmarks Thread

[VR Enthusiast]

I have to disappoint you. No async shader benefit in Thief. You forgot to post that you picked a CPU limited case with a very weak AMD CPU. No "free performance".

You forgot that as the resolution increases so does the chances for separate bottlenecks to affect FPS appear. None of those benchmarks proves that async shaders do or don't work.

Also we know Thief is broken on GCN 1.2. Is that the future we gonna see?

If by broken you mean bad DX12 performance on Nvidia cards, yes I believe that's the future we will see.

I thought it was made clear I dont support your view.

If you don't believe what Kanter, Carmack, AMD or Nvidia says about VR, who/what do you believe?

 

[Enigmoid]

I don't believe this to be the case. I believe that the method being employed in async shaders (developed on GCN) just doesn't work very well on Maxwell.

Ashes might be worst case but I don't think so. I think that when devs really start using async shaders ie pushing all 8 ACEs on a PS4 for example, then you'll start to see worst case for Nvidia.

Look at that graph from Carfax above, you can see even the 370 gains with DX12 and it only has 2 ACEs. There are no benchmarks showing lower fps on AMD in DX12. That probably means that Ashes is nowhere near pushing the maximum limit (8 ACEs) on later GCN cards.

I think some of the variation in the totals we see is probably down to it not being a set benchmark. Because of that we can't say for sure that it's an awesome DX12 benchmark but the overall picture should be obvious.

Again. You are making assumptions about ACE units and assuming that you can play the numbers game in 1 ACE = 1 Nvidia equivalent when there is absolutely nothing supporting this claim. The two approaches Nvidia and AMD are taking seem to be completely different and you are comparing them one to one.

There is no reason to believe that the more queues you have the better it will be (8 ACE units can actually be used). Having more queues allows for better utilization but you only get better utilization if hardware is not being used - you can't keep scaling because eventually you run out of unutilized hardware. As Silverforce said, it is unlikely that you will be able to get more than a few compute queues + 1 graphics queue scaling properly (you will start taking resources away from other queues if you add more queues).

Well it certainly isn't "spare cycles".

DX12 performance for NV in Ashes.
1. No async shaders (no dynamic lights), draw call test = Big perf boost.
2. Heavy test with dynamic lights (which the devs CONFIRMED is via async compute) = Small perf drop. Across the Kepler & Maxwell series.

A few possible conclusions you can draw from that. But to say there's no evidence that Async Compute is causing NV hardware some issues, is wrong. In fact, there's NO evidence it does not. There's NO evidence Maxwell excels at async shaders.

You would have evidence IF and WHEN you find gamedevs that works with DX12 which publicly say Maxwell is great at it, or a demo. Both will be viable evidence.
https://forum.beyond3d.com/threads/dx12-performance-thread.57188/page-4#post-1867649

You cannot draw any conclusions from this benchmark about DX12 nor anything about async shaders. You can draw conclusions about how Ashes will run but that is pretty much it. It would be the same for any DX11 game as well - there simply is not enough of a sample to make any sort of claims.

The problem is you have no idea if something like this is happening on Nvidia cards.

Mantle performs worse on the 285 because the developers have not optimized mantle. Looking at AOS and seeing that Nvidia is doing poorly on an alpha demo gives no conclusions.

Here was the star swarm alpha.

(770 on 3960X)

Look at DX11 later.

Both on Extreme RTS mode. AT's are done with a 4960X @ 4.2. Nvidia's performance has improved by > 2.5x. You can't make any conclusions about DX12 based on alpha performance.

 

[Leadbox]

Here was the star swarm alpha.

(770 on 3960X)

Look at DX11 later.

Both on Extreme RTS mode. AT's are done with a 4960X @ 4.2. Nvidia's performance has improved by > 2.5x. You can't make any conclusions about DX12 based on alpha performance.

Is that the anandtech chart that also had videos showing the geforces not rendering certain effects? What ever became of that?

 

[VR Enthusiast]

Again. You are making assumptions about ACE units and assuming that you can play the numbers game in 1 ACE = 1 Nvidia equivalent when there is absolutely nothing supporting this claim. The two approaches Nvidia and AMD are taking seem to be completely different and you are comparing them one to one.

No I'm not, I'm saying that whatever Nvidia is doing with their async compute, it's not very good. Doesn't matter about numbers when it's a bad base.

There is no reason to believe that the more queues you have the better it will be (8 ACE units can actually be used). Having more queues allows for better utilization but you only get better utilization if hardware is not being used - you can't keep scaling because eventually you run out of unutilized hardware. As Silverforce said, it is unlikely that you will be able to get more than a few compute queues + 1 graphics queue scaling properly (you will start taking resources away from other queues if you add more queues).

The technology is new from a development perspective and I'm sure many new ideas will come from its use. We don't know if 8 is too many or not but we can certainly guess that 1 or 2 might not be enough.

But again, it doesn't matter how many you have when you're doing it wrong like Nvidia seems to be.

Remember (this is important) that they cannot get similar performance to AMD in VR either. At some point you have to accept that this is a hardware limitation in Maxwell. They don't have the ACEs, people like David Kanter say that ACEs are needed for proper VR. Nvidia admits they cannot get below 20ms which is critical for good VR. Game developers everywhere are developing VR titles on LiquidVR.

Nvidia doesn't have the ACEs and they have poor performance in DX12 in a game that we know uses asynchronous shaders. Performance that is even worse than their DX11 results.

This isn't rocket science. Maxwell's async shaders are at worst a liability or at best offering very small benefits depending on usage. AMD's ACEs consistently give better performance no matter what the application is.

 

[Silverforce11]

You cannot draw any conclusions from this benchmark about DX12 nor anything about async shaders.

You can, whether its true or not, we won't know until there's more evidence to back one of the conclusions.

1. Oxide did not optimize DX12 async compute for NV (because, alpha/closed beta or they are unethical due to AMD sponsorship!).
2. Kepler/Maxwell has issues with async compute.
3. Kepler/Maxwell has issues with async compute AS they are implemented by Oxide in Ashes.

You can probably think up a few more reasons..

But I wouldn't say for a FACT that Oxide is corrupt. Likewise, I can't say for a FACT that Kepler/Maxwell are both gimped for async compute.

 

[Glo.]

8 ACEs with 8 Queues each compared to 1 with 32 queues will always be better for parallel computing.

1 ACE with 32 Queues gets bottlenecked by the amount of work it schedule for the GPU to do in one cycle.

Whereas 8 ACEs are doing less work. The work is split into smaller bits, in order to make proper schedules on GCN hardware. Therefore it can be made in parallel, not multithreaded. At the same time. Thats where the difference is.

Of course this is gimping Nvidia hardware. But using all capabilities of asynchronous compute is really going "upstream" the Nvidia hardware? Or maybe that is the nature of hardware you have?

In multithreaded environment the faster your cores are the better for the code - Maxwell.
In Parallel environment, the more ACEs you have, and the less work they have to do in one cycle the better for your hardware - GCN. Thats what makes difference. GCN is proper(?) Out-of-Order architecture here. With ONE ACE schedule Maxwell2 is completely non Out-Of-Order architecture.

We can argue to death about it: is it gimping one vendor or not? Using all of the capabilities of your API is not gimping hardware.

 

[TheELF]

8 ACEs with 8 Queues each compared to 1 with 32 queues will always be better for parallel computing.

Will it though?
Having to issue 8 commands each cycle instead of just one will be faster?

1 ACE with 32 Queues gets bottlenecked by the amount of work it schedule for the GPU to do in one cycle.

Forget the one cycle,gpus work at ~1ghz that is 1 billion cycles per second, 1x32 can do just as much work in one second or even more,doesn't have to but it can.

 

[sontin]

Seriously, every GPU is a parallel processing monster. Feeding the compute units with information is the bread and butter for nVidia and AMD.

And btw: Hyper-Q allows input from 32 different streams. So no, it has not "one" whatever.

 

[AtenRa]

Heh, seems the same situation with Tessellation and GTX480/580 vs HD5870/6970.

 

[ShintaiDK]

Heh, seems the same situation with Tessellation and GTX480/580 vs HD5870/6970.

Seems more like Star Swarm before and after.

 

[Samwell]

You can, whether its true or not, we won't know until there's more evidence to back one of the conclusions.

1. Oxide did not optimize DX12 async compute for NV (because, alpha/closed beta or they are unethical due to AMD sponsorship!).
2. Kepler/Maxwell has issues with async compute.
3. Kepler/Maxwell has issues with async compute AS they are implemented by Oxide in Ashes.

You can probably think up a few more reasons..

I'll take Nr. 4 which a friend who is gamedev told me:
Nvidia and AMD have different architecture approaches. Nvidia is trying to use every shader up tp 100% during normal gameload. Async compute won't help them, because you can't make more than to fully utilize your shaders. AMDs approach is different. Outside of gpgpu you can never use all shaders efficiently because other things are limiting, but they build more of them. Now with Async Compute you are able to get a much better utilization. So is Nv having issues? No, but still Amd will get an advantage by using Async compute.

 

[Glo.]

Will it though?
Having to issue 8 commands each cycle instead of just one will be faster?


Forget the one cycle,gpus work at ~1ghz that is 1 billion cycles per second, 1x32 can do just as much work in one second or even more,doesn't have to but it can.

In one post you have given question and answered it. 8 commands with 8 queues are better than 1 command with 32 queues. Because of the horsepower of the GPUs.

Thats where you get bottlenecked in parallel world, not multithreaded. DX11 was multithreaded environment. DX12 is parallel.

 

[Azix]

I'll take Nr. 4 which a friend who is gamedev told me:
Nvidia and AMD have different architecture approaches. Nvidia is trying to use every shader up tp 100% during normal gameload. Async compute won't help them, because you can't make more than to fully utilize your shaders. AMDs approach is different. Outside of gpgpu you can never use all shaders efficiently because other things are limiting, but they build more of them. Now with Async Compute you are able to get a much better utilization. So is Nv having issues? No, but still Amd will get an advantage by using Async compute.

I am seriously curious about what this means for the 980ti. On paper it's not much faster than a 290x/390x. If the situation was that it was using as many shaders as it could and the 290x just was not, then what will happen when the 290x is? What will happen to the entire pecking order of current GPUs?

It also calls into question the value of a fury x over the hawaii cards. I think the difference there will be resolution being played at if the bottleneck for both is in the same place.

I guess it depends on how much is being done with compute.

 

[dogen1]

In one post you have given question and answered it. 8 commands with 8 queues are better than 1 command with 32 queues. Because of the horsepower of the GPUs.

Thats where you get bottlenecked in parallel world, not multithreaded. DX11 was multithreaded environment. DX12 is parallel.

According to zlatan the penalty from switching contexts is a bigger deal than the front end design(the ACE units, vs whatever maxwell uses), and that in practice more than a couple of async pipes might not even be useful.

http://forums.anandtech.com/showpost.php?p=37634942&postcount=10

 

[Glo.]

Azix, I think the best way to understand what will happen is to look at compute power of GPUs.

I am starting to wonder right now if the Asynchronous Shading is Using the Nvidia GPUs in 100%.

I don't think so, thats why we see dump in gaming performance. 8 ACEs on GCN are utilizing the GPU in 100% in Parallel environment. But what about Maxwell? You cannot expect it to be parallel because of one ACE in that architecture. It is again multithreaded only. And that is why we see dumps in performance. It may possibly not be due to the code of the game or API, but in fact Asynchronous Shading of Nvidia hardware. Its parallel vs multithreaded the key to understanding what is going on.

Edit. Dogen1. And that exactly is what we are debating here. Parallel vs multithreaded environment.

 

[sontin]

Edit. Dogen1. And that exactly is what we are debating here. Parallel vs multithreaded environment.

A GPU is a massive parallel processor. It only performs when there are thousend of threads flying around to get executed by the units at the same time.

 

[Azix]

Some more information from OCN

= = = =

Hi xxxxxxxxx,

Thanks for your interest in the Ashes of the Singularity benchmark.

In order to get an accurate picture of how well a given CPU will perform, it's important to look at the CPU Frame rate with Infinite GPU on and off ( a check box exists on the benchmark settings panel ). Note, while on, you may see some graphical corruption due to use of async shaders, however the results will be valid.

With Infinite GPU, you should see %90+ workload on your CPU. In this mode, we do not "wait" in the case where the GPU is still busy. You should see excellent scaling between 4-16 thread machines.This can only be tracked on DX12.

Without Infinite GPU, the CPU will "Wait" on a signal from the GPU that the ready to process another frame. During this wait, the CPU tends to power down when there isn't any additional work to do and effectively serializes a portion of the frame. This serialization is what causes the CPU frame rate discrepancy between Infinite GPU on and off.

In addition, due to this "wait", one interesting stat to track is your power draw. On DX11 the power draw tends to be much higher than on DX12, as the additional serial threads that the driver needs to process the GPU commands effectively forces the CPU to be active even if it is only using a fraction of it's cores. This tends to be an overlooked benefit to DX12 since the API is designed so that engines can evenly distribute work.


Regarding specific CPU workloads and the differences between AMD and Intel it will be important to note a few things.

1. We have heavily invested in SSE ( mostly 2 for compatibility reasons ) and a significant portion of the engine is executing that code during the benchmark. It could very well be 40% of the frame. Possibly more.

2. While we do have large contiguous blocks of SSE code ( mainly in our simulations ) it is also rather heavily woven into the entire game via our math libraries. Our AI and gameplay code tend to be very math heavy.

3. The Nitrous engine is designed to be data oriented ( basically we know what memory we need and when ). Because of this, we can effectively utilize the SSE streaming memory instructions in conjunction with prefetch ( both temporal and non temporal ). In addition, because our memory accesses are more predictable the hardware prefetcher tends to be better utilized.

4. Memory bandwidth is definitely something to consider. The larger the scope of the application, paired with going highly parallel puts a lot of pressure on the Memory System. On my i7 3770s i'm hitting close to peak bandwith on 40% of the frame.


I hope this information helps point you in the right direction for your investigation into the performance differences between AMD and Intel. We haven't done exhaustive comparative tests, but generally speaking we have found AMD chips to compare more favorably to Intel than what is displayed via synthetic benchmarks. I'm looking forward to your results.

# # #

http://www.overclock.net/t/1569897/...singularity-dx12-benchmarks/850#post_24335709

 

[Despoiler]

Sorry for the late reply. Extended my vacation by a day. DX11 is a stutter and hitching fest. DX12 is way smoother at all points in the benchmark. My bet is the 10525 Win10 build they pushed to insiders. Pure garbage.

Edit:Final numbers updated on all benchmarks. Made sure there was no throttling on any bench. Win10 issue found and resolved.

This is the DX12 GPU bound test.

===========================================================
Oxide Games
Ashes Benchmark Test - ©2015
C:\Users\green\Documents\My Games\Ashes of the Singularity\Benchmarks\Output_15_08_25_1917.txt
Version 0.49.11978
08/25/2015 19:21
===========================================================

== Hardware Configuration =================================
GPU:		AMD Radeon R9 200 Series
CPU:		GenuineIntel
		       Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz
Physical Cores:			4
Logical Cores:			4
Physical Memory: 		17131261952 bytes
Allocatable Memory:		140737488224256 bytes
===========================================================


== Configuration ==========================================
API:				DirectX 12
Resolution:			1920x1080
Fullscreen:			True
Bloom Quality:			High
PointLight Quality:		High
Glare Quality:			High
Shading Samples: 		16
Terrain Shading Samples: 		16
Shadow Quality:			High
Temporal AA Duration:		12
Temporal AA Time Slice:		2
Multisample Anti-Aliasing:		4
Texture Rank :		1
===========================================================


== Results ================================================
BenchMark 0
TestType: Full System Test
==Sub Mark Normal Batch ================================== 
Total Time: 60.066971
Avg Framerate : 33.167847 ms (30.149681 FPS)
Weighted Framerate : 34.272232 ms (29.178141 FPS)
CPU frame rate (estimated framerate if not GPU bound): 12.032233 ms (83.110092 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 2606.245850
Average Batches per frame: 10196.457031
==Sub Mark Medium Batch ================================== 
Total Time: 61.988529
Avg Framerate : 42.545319 ms (23.504349 FPS)
Weighted Framerate : 43.700924 ms (22.882813 FPS)
CPU frame rate (estimated framerate if not GPU bound): 14.759303 ms (67.753876 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 3768.980225
Average Batches per frame: 21734.085938
==Sub Mark Heavy Batch ================================== 
Total Time: 57.945545
Avg Framerate : 50.607464 ms (19.759930 FPS)
Weighted Framerate : 51.374081 ms (19.465069 FPS)
CPU frame rate (estimated framerate if not GPU bound): 22.826572 ms (43.808590 FPS)
Percent GPU Bound: 99.920235%
Driver throughput (Batches per ms): 4091.177734
Average Batches per frame: 48135.554688
===========================================================
===========================================================
==Shot mid artemis battle shot ================================== 
Total Time: 4.970352
Avg Framerate : 40.740589 ms (24.545546 FPS)
Weighted Framerate : 40.910221 ms (24.443769 FPS)
CPU frame rate (estimated framerate if not GPU bound): 14.227588 ms (70.285980 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 2224.667480
Average Batches per frame: 12812.114258
==Shot  far artemis battle shot ================================== 
Total Time: 5.004176
Avg Framerate : 41.701466 ms (23.979973 FPS)
Weighted Framerate : 41.743404 ms (23.955879 FPS)
CPU frame rate (estimated framerate if not GPU bound): 14.788525 ms (67.620003 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 3483.738525
Average Batches per frame: 20711.583984
==Shot satellite shot ================================== 
Total Time: 4.971457
Avg Framerate : 54.037582 ms (18.505640 FPS)
Weighted Framerate : 54.351871 ms (18.398630 FPS)
CPU frame rate (estimated framerate if not GPU bound): 22.922436 ms (43.625381 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 3919.023926
Average Batches per frame: 47643.640625
==Shot fighters satellite chase ================================== 
Total Time: 12.029092
Avg Framerate : 26.150198 ms (38.240627 FPS)
Weighted Framerate : 26.532988 ms (37.688934 FPS)
CPU frame rate (estimated framerate if not GPU bound): 9.924479 ms (100.760948 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 2087.398438
Average Batches per frame: 5230.299805
==Shot surrounded T3 shot ================================== 
Total Time: 5.051164
Avg Framerate : 35.571575 ms (28.112333 FPS)
Weighted Framerate : 35.911373 ms (27.846331 FPS)
CPU frame rate (estimated framerate if not GPU bound): 11.818226 ms (84.615074 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 3088.761719
Average Batches per frame: 13027.302734
==Shot overhead top battle shot ================================== 
Total Time: 4.963708
Avg Framerate : 32.655975 ms (30.622267 FPS)
Weighted Framerate : 32.675716 ms (30.603765 FPS)
CPU frame rate (estimated framerate if not GPU bound): 12.289145 ms (81.372627 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 3608.273682
Average Batches per frame: 16167.716797
==Shot sideline battle shot ================================== 
Total Time: 4.972024
Avg Framerate : 45.614899 ms (21.922663 FPS)
Weighted Framerate : 45.977909 ms (21.749575 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.176436 ms (65.891617 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 3423.498291
Average Batches per frame: 18980.357422
==Shot bomber chase cam ================================== 
Total Time: 7.026407
Avg Framerate : 53.636692 ms (18.643953 FPS)
Weighted Framerate : 54.144970 ms (18.468935 FPS)
CPU frame rate (estimated framerate if not GPU bound): 18.650915 ms (53.616673 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 3498.458008
Average Batches per frame: 28985.771484
==Shot long shot ================================== 
Total Time: 9.992028
Avg Framerate : 35.432724 ms (28.222500 FPS)
Weighted Framerate : 35.593052 ms (28.095371 FPS)
CPU frame rate (estimated framerate if not GPU bound): 12.828273 ms (77.952820 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4598.856934
Average Batches per frame: 21694.601563
==Shot fighters chase cam 1 ================================== 
Total Time: 15.000812
Avg Framerate : 53.574326 ms (18.665657 FPS)
Weighted Framerate : 54.288212 ms (18.420204 FPS)
CPU frame rate (estimated framerate if not GPU bound): 14.382218 ms (69.530304 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4827.747070
Average Batches per frame: 22646.125000
==Shot satellite shot 2 ================================== 
Total Time: 7.995778
Avg Framerate : 51.254986 ms (19.510298 FPS)
Weighted Framerate : 51.857342 ms (19.283672 FPS)
CPU frame rate (estimated framerate if not GPU bound): 23.181801 ms (43.137287 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4428.077637
Average Batches per frame: 50732.468750
==Shot top battle shot ================================== 
Total Time: 4.980666
Avg Framerate : 48.830063 ms (20.479187 FPS)
Weighted Framerate : 49.049416 ms (20.387604 FPS)
CPU frame rate (estimated framerate if not GPU bound): 16.312384 ms (61.303120 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4995.383301
Average Batches per frame: 32310.853516
==Shot sideline battle shot 2 ================================== 
Total Time: 10.002266
Avg Framerate : 47.629841 ms (20.995241 FPS)
Weighted Framerate : 47.687401 ms (20.969900 FPS)
CPU frame rate (estimated framerate if not GPU bound): 22.727474 ms (43.999611 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 2581.974609
Average Batches per frame: 31573.982422
==Shot fighters chase cam 2 ================================== 
Total Time: 9.995485
Avg Framerate : 41.997837 ms (23.810749 FPS)
Weighted Framerate : 43.716564 ms (22.874624 FPS)
CPU frame rate (estimated framerate if not GPU bound): 13.752467 ms (72.714226 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 3665.660400
Average Batches per frame: 16659.783203
==Shot  long shot 3 ================================== 
Total Time: 5.023561
Avg Framerate : 54.603931 ms (18.313700 FPS)
Weighted Framerate : 54.752277 ms (18.264080 FPS)
CPU frame rate (estimated framerate if not GPU bound): 26.908653 ms (37.162766 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4381.352539
Average Batches per frame: 65818.281250
==Shot  mountain air battle shot ================================== 
Total Time: 10.000275
Avg Framerate : 32.258949 ms (30.999149 FPS)
Weighted Framerate : 32.386639 ms (30.876932 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.218138 ms (65.711060 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 2449.261963
Average Batches per frame: 17286.818359
==Shot sideline battle shot 3 ================================== 
Total Time: 5.012171
Avg Framerate : 40.749359 ms (24.540262 FPS)
Weighted Framerate : 40.917667 ms (24.439320 FPS)
CPU frame rate (estimated framerate if not GPU bound): 13.057414 ms (76.584839 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4339.979492
Average Batches per frame: 17900.478516
==Shot panoramic battle shot ================================== 
Total Time: 9.989502
Avg Framerate : 51.759075 ms (19.320284 FPS)
Weighted Framerate : 51.896755 ms (19.269028 FPS)
CPU frame rate (estimated framerate if not GPU bound): 22.095650 ms (45.257774 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4638.026367
Average Batches per frame: 51648.644531
==Shot mountain air battle shot 2 ================================== 
Total Time: 4.997723
Avg Framerate : 34.467056 ms (29.013212 FPS)
Weighted Framerate : 34.692169 ms (28.824949 FPS)
CPU frame rate (estimated framerate if not GPU bound): 16.387821 ms (61.020920 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4938.664063
Average Batches per frame: 36533.503906
==Shot low vista ================================== 
Total Time: 9.997131
Avg Framerate : 57.454777 ms (17.404993 FPS)
Weighted Framerate : 57.798950 ms (17.301352 FPS)
CPU frame rate (estimated framerate if not GPU bound): 24.398920 ms (40.985420 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 4182.386230
Average Batches per frame: 52273.257813
==Shot fighters satellite chase 2 ================================== 
Total Time: 7.006032
Avg Framerate : 27.474636 ms (36.397205 FPS)
Weighted Framerate : 27.907381 ms (35.832813 FPS)
CPU frame rate (estimated framerate if not GPU bound): 10.766924 ms (92.877045 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 2718.024170
Average Batches per frame: 7401.745605
==Shot high vista ================================== 
Total Time: 4.968132
Avg Framerate : 59.857010 ms (16.706480 FPS)
Weighted Framerate : 60.294773 ms (16.585186 FPS)
CPU frame rate (estimated framerate if not GPU bound): 27.430426 ms (36.455868 FPS)
Percent GPU Bound: 99.069618%
Driver throughput (Batches per ms): 4732.766113
Average Batches per frame: 69527.578125
==Shot  small mountain battle ================================== 
Total Time: 5.026293
Avg Framerate : 32.427696 ms (30.837839 FPS)
Weighted Framerate : 32.722191 ms (30.560301 FPS)
CPU frame rate (estimated framerate if not GPU bound): 13.377983 ms (74.749680 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 2018.567383
Average Batches per frame: 10574.116211
==Shot bomber chase cam 2 ================================== 
Total Time: 11.024861
Avg Framerate : 38.414150 ms (26.032074 FPS)
Weighted Framerate : 38.901142 ms (25.706186 FPS)
CPU frame rate (estimated framerate if not GPU bound): 13.418015 ms (74.526672 FPS)
Percent GPU Bound: 100.000000%
Driver throughput (Batches per ms): 1980.907104
Average Batches per frame: 9400.495117
===========================================================

 

[Despoiler]

Here is the DX12 CPU bound test.

===========================================================
Oxide Games
Ashes Benchmark Test - ©2015
C:\Users\green\Documents\My Games\Ashes of the Singularity\Benchmarks\Output_15_08_25_1913.txt
Version 0.49.11978
08/25/2015 19:16
===========================================================

== Hardware Configuration =================================
GPU:		AMD Radeon R9 200 Series
CPU:		GenuineIntel
		       Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz
Physical Cores:			4
Logical Cores:			4
Physical Memory: 		17131261952 bytes
Allocatable Memory:		140737488224256 bytes
===========================================================


== Configuration ==========================================
API:				DirectX 12
Resolution:			1920x1080
Fullscreen:			True
Bloom Quality:			High
PointLight Quality:		High
Glare Quality:			High
Shading Samples: 		16
Terrain Shading Samples: 		16
Shadow Quality:			High
Temporal AA Duration:		12
Temporal AA Time Slice:		2
Multisample Anti-Aliasing:		4
Texture Rank :		1
===========================================================


== Results ================================================
BenchMark 0
TestType: CPU Test (infinitly fast GPU)
==Sub Mark Normal Batch ================================== 
Total Time: 60.032566
Avg Framerate : 14.624255 ms (68.379555 FPS)
Weighted Framerate : 15.107683 ms (66.191490 FPS)
CPU frame rate (estimated framerate if not GPU bound): 14.607916 ms (68.456039 FPS)
Percent GPU Bound: 60.381462%
Driver throughput (Batches per ms): 3260.413086
Average Batches per frame: 5579.907715
==Sub Mark Medium Batch ================================== 
Total Time: 62.004711
Avg Framerate : 15.551721 ms (64.301567 FPS)
Weighted Framerate : 16.091860 ms (62.143223 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.534580 ms (64.372513 FPS)
Percent GPU Bound: 67.983711%
Driver throughput (Batches per ms): 4036.208984
Average Batches per frame: 7785.166016
==Sub Mark Heavy Batch ================================== 
Total Time: 57.968128
Avg Framerate : 20.304073 ms (49.251202 FPS)
Weighted Framerate : 21.524710 ms (46.458233 FPS)
CPU frame rate (estimated framerate if not GPU bound): 20.287279 ms (49.291973 FPS)
Percent GPU Bound: 69.504768%
Driver throughput (Batches per ms): 4371.764160
Average Batches per frame: 19406.695313
===========================================================
===========================================================
==Shot mid artemis battle shot ================================== 
Total Time: 4.980194
Avg Framerate : 16.013485 ms (62.447372 FPS)
Weighted Framerate : 16.909775 ms (59.137394 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.997529 ms (62.509655 FPS)
Percent GPU Bound: 65.321640%
Driver throughput (Batches per ms): 2224.080078
Average Batches per frame: 4731.568848
==Shot  far artemis battle shot ================================== 
Total Time: 4.986956
Avg Framerate : 16.244156 ms (61.560600 FPS)
Weighted Framerate : 17.003408 ms (58.811741 FPS)
CPU frame rate (estimated framerate if not GPU bound): 16.227695 ms (61.623043 FPS)
Percent GPU Bound: 70.497627%
Driver throughput (Batches per ms): 3801.470947
Average Batches per frame: 8642.937500
==Shot satellite shot ================================== 
Total Time: 5.015798
Avg Framerate : 19.592960 ms (51.038742 FPS)
Weighted Framerate : 21.019047 ms (47.575897 FPS)
CPU frame rate (estimated framerate if not GPU bound): 19.575914 ms (51.083183 FPS)
Percent GPU Bound: 73.198006%
Driver throughput (Batches per ms): 4057.278564
Average Batches per frame: 17135.636719
==Shot fighters satellite chase ================================== 
Total Time: 12.001231
Avg Framerate : 13.439229 ms (74.409035 FPS)
Weighted Framerate : 13.708800 ms (72.945847 FPS)
CPU frame rate (estimated framerate if not GPU bound): 13.423486 ms (74.496300 FPS)
Percent GPU Bound: 54.245876%
Driver throughput (Batches per ms): 2606.942871
Average Batches per frame: 3334.403320
==Shot surrounded T3 shot ================================== 
Total Time: 4.999869
Avg Framerate : 14.044578 ms (71.201859 FPS)
Weighted Framerate : 14.467849 ms (69.118774 FPS)
CPU frame rate (estimated framerate if not GPU bound): 14.029033 ms (71.280754 FPS)
Percent GPU Bound: 65.102242%
Driver throughput (Batches per ms): 3240.572754
Average Batches per frame: 4757.354004
==Shot overhead top battle shot ================================== 
Total Time: 4.992150
Avg Framerate : 13.790470 ms (72.513840 FPS)
Weighted Framerate : 14.201637 ms (70.414413 FPS)
CPU frame rate (estimated framerate if not GPU bound): 13.773757 ms (72.601830 FPS)
Percent GPU Bound: 63.646435%
Driver throughput (Batches per ms): 3568.279541
Average Batches per frame: 6276.370605
==Shot sideline battle shot ================================== 
Total Time: 5.002644
Avg Framerate : 14.542568 ms (68.763641 FPS)
Weighted Framerate : 14.930200 ms (66.978340 FPS)
CPU frame rate (estimated framerate if not GPU bound): 14.524733 ms (68.848083 FPS)
Percent GPU Bound: 69.228111%
Driver throughput (Batches per ms): 3768.087402
Average Batches per frame: 6149.244141
==Shot bomber chase cam ================================== 
Total Time: 7.007099
Avg Framerate : 15.606011 ms (64.077873 FPS)
Weighted Framerate : 16.052052 ms (62.297329 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.589210 ms (64.146935 FPS)
Percent GPU Bound: 70.198891%
Driver throughput (Batches per ms): 3524.996338
Average Batches per frame: 6968.427734
==Shot long shot ================================== 
Total Time: 9.990016
Avg Framerate : 15.298647 ms (65.365257 FPS)
Weighted Framerate : 15.885934 ms (62.948769 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.281797 ms (65.437332 FPS)
Percent GPU Bound: 63.649494%
Driver throughput (Batches per ms): 4443.299805
Average Batches per frame: 8942.050781
==Shot fighters chase cam 1 ================================== 
Total Time: 15.010671
Avg Framerate : 15.254746 ms (65.553368 FPS)
Weighted Framerate : 15.563836 ms (64.251511 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.237394 ms (65.628014 FPS)
Percent GPU Bound: 75.122375%
Driver throughput (Batches per ms): 6348.291504
Average Batches per frame: 7278.958008
==Shot satellite shot 2 ================================== 
Total Time: 7.981266
Avg Framerate : 20.360373 ms (49.115013 FPS)
Weighted Framerate : 21.544725 ms (46.415073 FPS)
CPU frame rate (estimated framerate if not GPU bound): 20.343641 ms (49.155411 FPS)
Percent GPU Bound: 70.245720%
Driver throughput (Batches per ms): 4540.005859
Average Batches per frame: 19740.083984
==Shot top battle shot ================================== 
Total Time: 5.013419
Avg Framerate : 17.052444 ms (58.642620 FPS)
Weighted Framerate : 17.682720 ms (56.552387 FPS)
CPU frame rate (estimated framerate if not GPU bound): 17.034851 ms (58.703186 FPS)
Percent GPU Bound: 68.636856%
Driver throughput (Batches per ms): 4882.496094
Average Batches per frame: 10504.251953
==Shot sideline battle shot 2 ================================== 
Total Time: 9.993227
Avg Framerate : 19.404324 ms (51.534904 FPS)
Weighted Framerate : 20.629059 ms (48.475307 FPS)
CPU frame rate (estimated framerate if not GPU bound): 19.386745 ms (51.581631 FPS)
Percent GPU Bound: 70.833344%
Driver throughput (Batches per ms): 2781.861328
Average Batches per frame: 12954.695313
==Shot fighters chase cam 2 ================================== 
Total Time: 10.013479
Avg Framerate : 14.533351 ms (68.807259 FPS)
Weighted Framerate : 15.057006 ms (66.414268 FPS)
CPU frame rate (estimated framerate if not GPU bound): 14.516018 ms (68.889412 FPS)
Percent GPU Bound: 71.805618%
Driver throughput (Batches per ms): 4198.450195
Average Batches per frame: 5028.534180
==Shot  long shot 3 ================================== 
Total Time: 4.984147
Avg Framerate : 21.299772 ms (46.948860 FPS)
Weighted Framerate : 22.700747 ms (44.051418 FPS)
CPU frame rate (estimated framerate if not GPU bound): 21.283747 ms (46.984207 FPS)
Percent GPU Bound: 71.866570%
Driver throughput (Batches per ms): 4797.245605
Average Batches per frame: 25188.876953
==Shot  mountain air battle shot ================================== 
Total Time: 10.007873
Avg Framerate : 15.735648 ms (63.549969 FPS)
Weighted Framerate : 16.581257 ms (60.309063 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.718360 ms (63.619869 FPS)
Percent GPU Bound: 58.479774%
Driver throughput (Batches per ms): 2516.821777
Average Batches per frame: 7746.384766
==Shot sideline battle shot 3 ================================== 
Total Time: 4.986118
Avg Framerate : 15.581620 ms (64.178177 FPS)
Weighted Framerate : 15.944702 ms (62.716755 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.564406 ms (64.249161 FPS)
Percent GPU Bound: 66.719612%
Driver throughput (Batches per ms): 4160.448242
Average Batches per frame: 6641.562500
==Shot panoramic battle shot ================================== 
Total Time: 10.012203
Avg Framerate : 20.858757 ms (47.941494 FPS)
Weighted Framerate : 21.884075 ms (45.695332 FPS)
CPU frame rate (estimated framerate if not GPU bound): 20.842842 ms (47.978100 FPS)
Percent GPU Bound: 70.357933%
Driver throughput (Batches per ms): 5704.416016
Average Batches per frame: 24086.257813
==Shot mountain air battle shot 2 ================================== 
Total Time: 4.994242
Avg Framerate : 17.900507 ms (55.864338 FPS)
Weighted Framerate : 18.617165 ms (53.713875 FPS)
CPU frame rate (estimated framerate if not GPU bound): 17.883354 ms (55.917919 FPS)
Percent GPU Bound: 52.077347%
Driver throughput (Batches per ms): 5026.096680
Average Batches per frame: 18664.373047
==Shot low vista ================================== 
Total Time: 9.998945
Avg Framerate : 20.874624 ms (47.905052 FPS)
Weighted Framerate : 22.055693 ms (45.339767 FPS)
CPU frame rate (estimated framerate if not GPU bound): 20.857580 ms (47.944202 FPS)
Percent GPU Bound: 71.878639%
Driver throughput (Batches per ms): 4110.664063
Average Batches per frame: 17585.076172
==Shot fighters satellite chase 2 ================================== 
Total Time: 7.013573
Avg Framerate : 15.655298 ms (63.876141 FPS)
Weighted Framerate : 15.873577 ms (62.997772 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.639821 ms (63.939350 FPS)
Percent GPU Bound: 47.015717%
Driver throughput (Batches per ms): 3653.437988
Average Batches per frame: 5933.826172
==Shot high vista ================================== 
Total Time: 4.988244
Avg Framerate : 22.673836 ms (44.103699 FPS)
Weighted Framerate : 24.155138 ms (41.399059 FPS)
CPU frame rate (estimated framerate if not GPU bound): 22.657654 ms (44.135197 FPS)
Percent GPU Bound: 70.562141%
Driver throughput (Batches per ms): 4806.547852
Average Batches per frame: 25106.412109
==Shot  small mountain battle ================================== 
Total Time: 5.004429
Avg Framerate : 15.398243 ms (64.942474 FPS)
Weighted Framerate : 16.024614 ms (62.403999 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.380653 ms (65.016747 FPS)
Percent GPU Bound: 46.469582%
Driver throughput (Batches per ms): 1359.447632
Average Batches per frame: 4251.633789
==Shot bomber chase cam 2 ================================== 
Total Time: 11.027662
Avg Framerate : 15.294955 ms (65.381035 FPS)
Weighted Framerate : 15.805305 ms (63.269894 FPS)
CPU frame rate (estimated framerate if not GPU bound): 15.278620 ms (65.450935 FPS)
Percent GPU Bound: 65.649757%
Driver throughput (Batches per ms): 4679.486816
Average Batches per frame: 9689.198242
===========================================================

 

[Despoiler]

Here is the DX11 GPU bound test.

===========================================================
Oxide Games
Ashes Benchmark Test - ©2015
C:\Users\green\Documents\My Games\Ashes of the Singularity\Benchmarks\Output_15_08_25_1921.txt
Version 0.49.11978
08/25/2015 19:25
===========================================================

== Hardware Configuration =================================
GPU:		AMD Radeon R9 200 Series
CPU:		GenuineIntel
		       Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz
Physical Cores:			4
Logical Cores:			4
Physical Memory: 		17131261952 bytes
Allocatable Memory:		140737488224256 bytes
===========================================================


== Configuration ==========================================
API:				DirectX
Resolution:			1920x1080
Fullscreen:			True
Bloom Quality:			High
PointLight Quality:		High
Glare Quality:			High
Shading Samples: 		16
Terrain Shading Samples: 		16
Shadow Quality:			High
Temporal AA Duration:		12
Temporal AA Time Slice:		2
Multisample Anti-Aliasing:		4
Texture Rank :		1
===========================================================


== Results ================================================
BenchMark 0
TestType: Full System Test
==Sub Mark Normal Batch ================================== 
Total Time: 60.171352
Avg Framerate : 35.311825 ms (28.319124 FPS)
Weighted Framerate : 36.876751 ms (27.117357 FPS)
Average Batches per frame: 10308.742188
==Sub Mark Medium Batch ================================== 
Total Time: 61.917240
Avg Framerate : 50.094856 ms (19.962130 FPS)
Weighted Framerate : 51.272846 ms (19.503500 FPS)
Average Batches per frame: 19538.451172
==Sub Mark Heavy Batch ================================== 
Total Time: 57.918991
Avg Framerate : 78.057938 ms (12.810997 FPS)
Weighted Framerate : 80.720375 ms (12.388446 FPS)
Average Batches per frame: 42937.027344
===========================================================
===========================================================
==Shot mid artemis battle shot ================================== 
Total Time: 4.965755
Avg Framerate : 45.979218 ms (21.748957 FPS)
Weighted Framerate : 46.680290 ms (21.422318 FPS)
Average Batches per frame: 12246.916992
==Shot  far artemis battle shot ================================== 
Total Time: 4.949141
Avg Framerate : 54.990452 ms (18.184975 FPS)
Weighted Framerate : 55.156525 ms (18.130220 FPS)
Average Batches per frame: 20643.189453
==Shot satellite shot ================================== 
Total Time: 4.992797
Avg Framerate : 76.812271 ms (13.018754 FPS)
Weighted Framerate : 77.245895 ms (12.945671 FPS)
Average Batches per frame: 42365.445313
==Shot fighters satellite chase ================================== 
Total Time: 12.046316
Avg Framerate : 26.302000 ms (38.019920 FPS)
Weighted Framerate : 26.873192 ms (37.211807 FPS)
Average Batches per frame: 4846.550293
==Shot surrounded T3 shot ================================== 
Total Time: 5.016349
Avg Framerate : 42.154194 ms (23.722433 FPS)
Weighted Framerate : 42.744827 ms (23.394644 FPS)
Average Batches per frame: 13196.152344
==Shot overhead top battle shot ================================== 
Total Time: 4.970282
Avg Framerate : 41.767075 ms (23.942303 FPS)
Weighted Framerate : 41.811790 ms (23.916698 FPS)
Average Batches per frame: 16949.951172
==Shot sideline battle shot ================================== 
Total Time: 4.995646
Avg Framerate : 50.461063 ms (19.817259 FPS)
Weighted Framerate : 50.543171 ms (19.785067 FPS)
Average Batches per frame: 18016.261719
==Shot bomber chase cam ================================== 
Total Time: 7.018539
Avg Framerate : 55.702698 ms (17.952452 FPS)
Weighted Framerate : 56.005733 ms (17.855314 FPS)
Average Batches per frame: 20132.857422
==Shot long shot ================================== 
Total Time: 9.968552
Avg Framerate : 49.106167 ms (20.364042 FPS)
Weighted Framerate : 49.225487 ms (20.314680 FPS)
Average Batches per frame: 20391.531250
==Shot fighters chase cam 1 ================================== 
Total Time: 15.001339
Avg Framerate : 59.293831 ms (16.865160 FPS)
Weighted Framerate : 59.667068 ms (16.759663 FPS)
Average Batches per frame: 20689.416016
==Shot satellite shot 2 ================================== 
Total Time: 8.036322
Avg Framerate : 81.174973 ms (12.319068 FPS)
Weighted Framerate : 83.597572 ms (11.962070 FPS)
Average Batches per frame: 42694.617188
==Shot top battle shot ================================== 
Total Time: 4.959628
Avg Framerate : 61.995346 ms (16.130243 FPS)
Weighted Framerate : 62.683098 ms (15.953264 FPS)
Average Batches per frame: 28455.099609
==Shot sideline battle shot 2 ================================== 
Total Time: 10.000641
Avg Framerate : 63.698353 ms (15.698994 FPS)
Weighted Framerate : 63.801750 ms (15.673552 FPS)
Average Batches per frame: 29362.687500
==Shot fighters chase cam 2 ================================== 
Total Time: 10.041683
Avg Framerate : 49.711304 ms (20.116150 FPS)
Weighted Framerate : 50.861469 ms (19.661249 FPS)
Average Batches per frame: 16850.683594
==Shot  long shot 3 ================================== 
Total Time: 4.933547
Avg Framerate : 107.251030 ms (9.323919 FPS)
Weighted Framerate : 108.068893 ms (9.253357 FPS)
Average Batches per frame: 58386.000000
==Shot  mountain air battle shot ================================== 
Total Time: 10.063139
Avg Framerate : 35.433586 ms (28.221811 FPS)
Weighted Framerate : 36.330624 ms (27.524986 FPS)
Average Batches per frame: 16380.383789
==Shot sideline battle shot 3 ================================== 
Total Time: 4.961229
Avg Framerate : 49.121082 ms (20.357859 FPS)
Weighted Framerate : 49.311089 ms (20.279413 FPS)
Average Batches per frame: 16524.355469
==Shot panoramic battle shot ================================== 
Total Time: 9.965940
Avg Framerate : 85.178986 ms (11.739985 FPS)
Weighted Framerate : 85.360428 ms (11.715030 FPS)
Average Batches per frame: 47109.242188
==Shot mountain air battle shot 2 ================================== 
Total Time: 5.056481
Avg Framerate : 51.075562 ms (19.578835 FPS)
Weighted Framerate : 51.726719 ms (19.332369 FPS)
Average Batches per frame: 42112.617188
==Shot low vista ================================== 
Total Time: 9.970628
Avg Framerate : 85.953690 ms (11.634171 FPS)
Weighted Framerate : 86.451385 ms (11.567194 FPS)
Average Batches per frame: 45370.257813
==Shot fighters satellite chase 2 ================================== 
Total Time: 7.041343
Avg Framerate : 28.623346 ms (34.936516 FPS)
Weighted Framerate : 29.303627 ms (34.125469 FPS)
Average Batches per frame: 7366.938965
==Shot high vista ================================== 
Total Time: 4.962653
Avg Framerate : 115.410538 ms (8.664721 FPS)
Weighted Framerate : 117.562897 ms (8.506085 FPS)
Average Batches per frame: 61376.023438
==Shot  small mountain battle ================================== 
Total Time: 5.038957
Avg Framerate : 33.818501 ms (29.569613 FPS)
Weighted Framerate : 35.035690 ms (28.542324 FPS)
Average Batches per frame: 9792.140625
==Shot bomber chase cam 2 ================================== 
Total Time: 11.050678
Avg Framerate : 36.470886 ms (27.419130 FPS)
Weighted Framerate : 36.740715 ms (27.217760 FPS)
Average Batches per frame: 12413.159180
===========================================================

 

[Despoiler]

Here is the DX11 CPU bound test.

===========================================================
Oxide Games
Ashes Benchmark Test - ©2015
C:\Users\green\Documents\My Games\Ashes of the Singularity\Benchmarks\Output_15_08_25_1926.txt
Version 0.49.11978
08/25/2015 19:29
===========================================================

== Hardware Configuration =================================
GPU:		AMD Radeon R9 200 Series
CPU:		GenuineIntel
		       Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz
Physical Cores:			4
Logical Cores:			4
Physical Memory: 		17131261952 bytes
Allocatable Memory:		140737488224256 bytes
===========================================================


== Configuration ==========================================
API:				DirectX
Resolution:			1920x1080
Fullscreen:			True
Bloom Quality:			High
PointLight Quality:		High
Glare Quality:			High
Shading Samples: 		16
Terrain Shading Samples: 		16
Shadow Quality:			High
Temporal AA Duration:		12
Temporal AA Time Slice:		2
Multisample Anti-Aliasing:		4
Texture Rank :		1
===========================================================


== Results ================================================
BenchMark 0
TestType: CPU Test (infinitly fast GPU)
==Sub Mark Normal Batch ================================== 
Total Time: 60.084259
Avg Framerate : 35.198746 ms (28.410105 FPS)
Weighted Framerate : 37.288040 ms (26.818251 FPS)
Average Batches per frame: 9411.399414
==Sub Mark Medium Batch ================================== 
Total Time: 61.985985
Avg Framerate : 50.808186 ms (19.681870 FPS)
Weighted Framerate : 52.077930 ms (19.201992 FPS)
Average Batches per frame: 19803.763672
==Sub Mark Heavy Batch ================================== 
Total Time: 57.931034
Avg Framerate : 77.035950 ms (12.980952 FPS)
Weighted Framerate : 79.816910 ms (12.528674 FPS)
Average Batches per frame: 41977.636719
===========================================================
===========================================================
==Shot mid artemis battle shot ================================== 
Total Time: 4.950612
Avg Framerate : 44.600105 ms (22.421471 FPS)
Weighted Framerate : 44.822109 ms (22.310419 FPS)
Average Batches per frame: 13440.072266
==Shot  far artemis battle shot ================================== 
Total Time: 4.953705
Avg Framerate : 55.659607 ms (17.966351 FPS)
Weighted Framerate : 56.038048 ms (17.845018 FPS)
Average Batches per frame: 20469.023438
==Shot satellite shot ================================== 
Total Time: 5.055498
Avg Framerate : 74.345551 ms (13.450704 FPS)
Weighted Framerate : 74.771507 ms (13.374078 FPS)
Average Batches per frame: 41592.101563
==Shot fighters satellite chase ================================== 
Total Time: 11.988623
Avg Framerate : 26.348623 ms (37.952648 FPS)
Weighted Framerate : 26.885054 ms (37.195389 FPS)
Average Batches per frame: 4932.793457
==Shot surrounded T3 shot ================================== 
Total Time: 4.994822
Avg Framerate : 42.328995 ms (23.624468 FPS)
Weighted Framerate : 42.785015 ms (23.372669 FPS)
Average Batches per frame: 13212.923828
==Shot overhead top battle shot ================================== 
Total Time: 5.000854
Avg Framerate : 40.657345 ms (24.595802 FPS)
Weighted Framerate : 40.789852 ms (24.515900 FPS)
Average Batches per frame: 14713.568359
==Shot sideline battle shot ================================== 
Total Time: 4.983840
Avg Framerate : 51.914997 ms (19.262257 FPS)
Weighted Framerate : 51.959297 ms (19.245832 FPS)
Average Batches per frame: 17217.958984
==Shot bomber chase cam ================================== 
Total Time: 6.977224
Avg Framerate : 59.634396 ms (16.768847 FPS)
Weighted Framerate : 59.947765 ms (16.681189 FPS)
Average Batches per frame: 22114.505859
==Shot long shot ================================== 
Total Time: 10.041292
Avg Framerate : 50.206459 ms (19.917755 FPS)
Weighted Framerate : 50.255436 ms (19.898344 FPS)
Average Batches per frame: 21024.875000
==Shot fighters chase cam 1 ================================== 
Total Time: 14.985825
Avg Framerate : 59.232510 ms (16.882620 FPS)
Weighted Framerate : 59.751106 ms (16.736092 FPS)
Average Batches per frame: 20280.455078
==Shot satellite shot 2 ================================== 
Total Time: 7.983154
Avg Framerate : 81.460762 ms (12.275849 FPS)
Weighted Framerate : 83.806770 ms (11.932211 FPS)
Average Batches per frame: 42277.222656
==Shot top battle shot ================================== 
Total Time: 5.015170
Avg Framerate : 63.483162 ms (15.752208 FPS)
Weighted Framerate : 64.150978 ms (15.588226 FPS)
Average Batches per frame: 30494.253906
==Shot sideline battle shot 2 ================================== 
Total Time: 9.992455
Avg Framerate : 63.646206 ms (15.711856 FPS)
Weighted Framerate : 63.771912 ms (15.680885 FPS)
Average Batches per frame: 28620.988281
==Shot fighters chase cam 2 ================================== 
Total Time: 10.022514
Avg Framerate : 53.029175 ms (18.857544 FPS)
Weighted Framerate : 54.933403 ms (18.203861 FPS)
Average Batches per frame: 17456.845703
==Shot  long shot 3 ================================== 
Total Time: 4.939523
Avg Framerate : 107.380936 ms (9.312640 FPS)
Weighted Framerate : 108.188896 ms (9.243093 FPS)
Average Batches per frame: 57836.347656
==Shot  mountain air battle shot ================================== 
Total Time: 10.100965
Avg Framerate : 35.441982 ms (28.215126 FPS)
Weighted Framerate : 36.257763 ms (27.580301 FPS)
Average Batches per frame: 16251.631836
==Shot sideline battle shot 3 ================================== 
Total Time: 4.927947
Avg Framerate : 48.791550 ms (20.495352 FPS)
Weighted Framerate : 48.897266 ms (20.451040 FPS)
Average Batches per frame: 17047.851563
==Shot panoramic battle shot ================================== 
Total Time: 9.985281
Avg Framerate : 83.909927 ms (11.917541 FPS)
Weighted Framerate : 84.071426 ms (11.894648 FPS)
Average Batches per frame: 46544.398438
==Shot mountain air battle shot 2 ================================== 
Total Time: 5.020296
Avg Framerate : 48.272079 ms (20.715908 FPS)
Weighted Framerate : 48.742294 ms (20.516064 FPS)
Average Batches per frame: 38286.300781
==Shot low vista ================================== 
Total Time: 10.005028
Avg Framerate : 85.513062 ms (11.694119 FPS)
Weighted Framerate : 85.988220 ms (11.629499 FPS)
Average Batches per frame: 44831.242188
==Shot fighters satellite chase 2 ================================== 
Total Time: 7.031800
Avg Framerate : 28.584551 ms (34.983932 FPS)
Weighted Framerate : 29.385027 ms (34.030937 FPS)
Average Batches per frame: 7127.605469
==Shot high vista ================================== 
Total Time: 4.949795
Avg Framerate : 115.111511 ms (8.687229 FPS)
Weighted Framerate : 117.346733 ms (8.521754 FPS)
Average Batches per frame: 62232.093750
==Shot  small mountain battle ================================== 
Total Time: 5.044833
Avg Framerate : 33.857937 ms (29.535172 FPS)
Weighted Framerate : 35.103867 ms (28.486891 FPS)
Average Batches per frame: 9883.261719
==Shot bomber chase cam 2 ================================== 
Total Time: 11.050242
Avg Framerate : 34.969120 ms (28.596659 FPS)
Weighted Framerate : 37.081375 ms (26.967716 FPS)
Average Batches per frame: 7704.920898
===========================================================

 

[Mahigan]

Well,

It looks like I may have caused quite the sh!t storm.

Look,

Lets all leave the partisanship aside. Partisanship serves no purpose.

Rather than yell at one another... why not try and discuss instead?

Peace.

 

[Despoiler]

It is not about bugs. This engine is still under development and the developer warned reviewers not to use MSAA because their DX12 implementation is still experimental.

Correction, NVIDIA told reviewers not to use MSAA because they felt it was a bug in the benchmark/game. It's been 100% proven to be an Nvidia driver issue. Nvidia has acknowledged it as their issue to fix in the last week or so. This is from the AoTS forums.

 

[Silverforce11]

Found some very pertinent info relating to context switch and why it may incur a performance hit (as mentioned by @Zlatan awhile ago, some of the members here attack him personally which is very shameful). This is directly from NV:

https://developer.nvidia.com/sites/...works/vr/GameWorks_VR_2015_Final_handouts.pdf

p31

All our GPUs for the last several years do context switches at draw call boundaries. So when the GPU wants to switch contexts, it has to wait for the current draw call to finish first.

Basically the pipeline is in-order, serial. Here they are referring to Async Shaders to perform timewarp:

So, even with timewarp being on a high-priority context, it’s possible for it to get stuck behind a longrunning draw call on a normal context. For instance, if your game submits a single draw call that happens to take 5 ms, then async timewarp might get stuck behind it, potentially causing it to miss vsync and cause a visible hitch.

So i think I understand when gamedevs describe Kepler/Maxwell as being in-order, and GCN out-of-order (stateless, no context switch necessary).

This can cause issues if developers do not optimize or design their engine around the limitations of NV's uarch when they use Async Shaders on NV GPUs. I believe this is what we are seeing with the performance drop on DX12 in Ashes.

As for how this impacts their VR latency, the timewarp uses async shaders to quickly render the last frame in response to user movement. That frame gets priority context, else it lags. The problem for NV's uarch is it isn't true timewarp/priority because it still has to wait for traffic in front to be cleared (finish rendering the last frame first!), so NV suggest to developers to break up their draw calls into many smaller batches to avoid long delays which can cause stutter/perf drops. If devs/engines do this, it can reduce the latency when async shaders are used, but if they do not, we get problems.

 

[tential]

FYI, PS4 has 8 ACE (same as recent GCN SKUs), XBone has 2. So devs will have to tone down usage for the Xbone, PC will probably have sliders, no worries.

Ya... PC will probably have sliders sure.

Just like that Hairworks slider right? Look how hard that was to get.

Don't be surprised if it's the same across PC, PS4, Xbone for 60%+ of titles (And that's conservative).

Or have you forgotten who Nvidia is? I'm sure they'll remind you soon enough....