-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Ashes of the Singularity Coming, with DX12 Benchmark in thread.
- Thread starter humbug
- Start date
Simiulalrly, there is often an increase in FPS going from low settigns to high setitngs.
Then there is the fact that in some of the nvidia benches the DX12 result is lower than DX 11 which si apretty obvious sign the game enigne or driver has a flaw.
Its pretty black and white that the numbers coming out of that benchmark are compelte junk, for both vendors. There are some glimpses that AND will do well but it is hard to trust any of it when the numbers have fundamental flaws. Increasing the resolution should never increase average FPS.
Then there is the fact that in some of the nvidia benches the DX12 result is lower than DX 11 which si apretty obvious sign the game enigne or driver has a flaw.
Its pretty black and white that the numbers coming out of that benchmark are compelte junk, for both vendors. There are some glimpses that AND will do well but it is hard to trust any of it when the numbers have fundamental flaws. Increasing the resolution should never increase average FPS.
CPU or not will make absoltuely no difference, icnreaseing resoltuin cannot increase average FPS.
Have people also noticed how the Furyx's 1080 results have caught right up to where they should be when using directx 12?
The increase you are talking about is also in decimals of an fps.
The increase you are talking about is also in decimals of an fps.
Last edited:
If AMD had this problem i would tend to agree its an engine problem, but they don't.
I can see why Nvidia are being so defensive, it looks like a Driver issue on Nvidia's part.
AMD results do shwo many of the same problems, icnreasaing resoltuon is increasing performance, increasing detail form low to high is increasing performance.
Nvidia are being defensive because this is the first public Dx12 benchmark and it is obviously flawed.
Cant certainly see how AMD have been working hard on DX12!! Those results show some amazing performance gains! Outstanding! Wish i could say same for nVidia?? Either small bump or some cases performs better on Dx11 lol. Weird. Kinda puts a question in my mind seen as i have my heart set on a 980Ti! Only thing is it could be at least 2-3 years till we get a good handful of games that run DX12 which AMD hardware can take advantage of. But then again i could be wrong and they could start churning out in less than a year. 
Still great work AMD!
Also D.P i don't see how it's flawed? They released a driver specifically for this benchmark/game and it doesn't perform as well as we hoped or no where near as good as AMD is doing. Is this why its flawed because AMD have better performance numbers?? NOOOOO it can't be lol.
Still great work AMD!
Also D.P i don't see how it's flawed? They released a driver specifically for this benchmark/game and it doesn't perform as well as we hoped or no where near as good as AMD is doing. Is this why its flawed because AMD have better performance numbers?? NOOOOO it can't be lol.
(removed part about 770 wccftech result since they used the wrond benchmark option and tested the FPS that the CPU can output instead of what the GPU is actually rendering.)
The titan X lost 3fps on and 5960x which can be from a number of small things like the directx 12 pathway only rendering real frames unlike DX11 which often will display old frames. So like with AFR the fps numbers can get artificially increased.
I am not saying that the benchmark is perfect though, it is still a pre alpha game engine but it does show massive improvements all around.
Last edited:
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
Nvidia put in a lot of effort to increase DX11 driver performance in the past 18 months, for instance they worked hard and got big increases in that first Mantle demo in DX11 performance.
Ultimately AMD have been focusing on Mantle/DX12 drivers since around the time Mantle was announced likely. Any company has finite resources. With not a huge step forward coming in DX11 titles( in the past 18 months) and as DX12 is going to be extremely widely adopted it has looked for a while like AMD is focusing resources to be 'ready' for DX12. It also looked like they didn't think it was a good investment to put a lot of time and money into improving DX11 driver overhead considering it was effectively the end of DX11 anyway.
Ultimately someone will say Nvidia has more money to have more people to do both but still one guy working on DX11 is a guy who isn't working on DX12. It's also worth pointing out that more AMD cards support more DX12 features, so there is potentially more driver work to be done for DX12 for AMD. With most new DX versions lets say 3 months after a new gen card launches it will still be 95% of gamers on the previous DX version, 5% on the new one but also only one small set of cards to make new drivers for. This time it will be lets say 70% of current gamers having DX12 compatible hardware across multiple generations of cards which also means WAY more work on new drivers. By current gamer I'm talking about people who buy a reasonable amount of games and have max 3-4 year old gpu.
DX12 adoption rate will be entirely different to DX9/10/11 and frankly because of how much access it will give devs and how little overhead it will have finally there will be in the future diminishing returns. Once you get low level and cut out most of the overhead while also enabling proper usage of multithreading it will scale well on cpu/gpu performance for years to come. Getting DX12 drivers right was/is far more important than improving DX11 drivers over the past year.
AMD's all looks fairly consistent, not perfect but one would expect some fluctuations.
http://www.pcper.com/reviews/Graphi...ted-Ashes-Singularity-Benchmark/Results-Avera
Just ebcause soem numebrs are flawed doesn't man all number areflawed, quite liekly some of the numbers are truely representitive but that doesn't mean hat some of them are compeltely bogus. The problem is you have no idea what is what.
PcPer numbers:
390X at 1080 Low DX12 is 28.3FPS. Increase setting to high and the FPS increase to 31.7. That is a rpetty big flaw ificnreasing game detail icnreases resolution then there is an error in the game engine. Go back to low setting and bump the resoltuon to 1600p, FPS go from 28.3 to 31.2 FPS, there is just no way that is possible with a properly working benchmark.
The numbers don't make any sense for AMD or Nvidia, so it impossible to really draw any conclusions.
See my post above, AMD results get faster with increased resoltuon and increased detail.
'often' being the minority, again, and also again being both a tiny difference for the most part (arguably within error margins/consistency) and only present on low-end cpu's.
The instances on nvidia where the DX12 result is lower is almost always the high detail results, to me this implies either the DX11 route is being optimised very well for something present in the high detail settings and that optimisation is either not in place or not possible in the driver for DX12, or that there is a bug somewhere be it driver or benchmark. This is probably your most relevant criticism but not sure it's enough to denounce the entire results just yet.
If it's CPU bound then there are far more factors that could affect the framerate, notably anything windows may or may not decide to do. As mentioned the difference is relatively small, at a time the GPU is basically sat around waiting on the CPU, it could be a fault in the benchmark or it could be something innocuous affecting the results.
I know which situations you mean. But those cases are completely cpu bound. so increasing resolution and detail will increase performance, even if it is a marginal gain.
the game was running on a dual core i3 and an 8370 in those cases. all of the results hit a wall for all cards in those tests. But once you go up to the 5960x then the results look normal and what you expect since the situation is no longer purely cpu bound.
All i can think is that the game simulation is not running across as many cores as the system can throw at it.
Last edited:
The results I have seen so far are total garbage. Any benchmark that does not scale with resolution is totally flawed. Any DX12 bench that does not scale with resolution even more so as the whole idea of the new API is to remove the CPU bottleneck.
I will only take any new DX12 bench seriously when it scales in the same way as Heaven 4 both with resolution and number of GPUs.
I expect there will be a new version of the Heaven bench that uses DX12 and I think that will be a far better guide.
Something else we can take away from here.
Furyx vs 980ti results DX11
Furyx vs 980ti results DX12
People also need to realise that this benchmark can still become CPU bound becasue it is a game benchmark. Directx 12 does reduce the cpu overhead from the graphics rendering system. But the game itself can still become cpu bound and reduce gpu performance due to other computations.
This isn't like a GPU benchmark where only graphics calculations are being performed on the CPU.
Furyx vs 980ti results DX11
Furyx vs 980ti results DX12
People also need to realise that this benchmark can still become CPU bound becasue it is a game benchmark. Directx 12 does reduce the cpu overhead from the graphics rendering system. But the game itself can still become cpu bound and reduce gpu performance due to other computations.
This isn't like a GPU benchmark where only graphics calculations are being performed on the CPU.
Last edited:
Those slides highlight perfectly why the benchmark is flawed, DX12 results end up slower on the 980Ti, pretty damn obvious soemthing is broken somewhere. Either the game engine or Nvidia drivers but since many of the AMd results are completely flawed I owuld go with engine problems primarily. NVidia have stated as much.
Associate
- Joined
- 10 Nov 2012
- Posts
- 2,441
- Location
- Down the road from Diddly Squat Farm
I just became a founder
Soldato
- Joined
- 30 Nov 2011
- Posts
- 11,522
You mean like higher FPS at lower res or lower IQ? it is doing that
No, not always its not, its all over the show, sometimes lower res is lower fps