No DX12 Software is Suitable for Benchmarking *spawn*
- Thread starter trinibwoy
- Start date
D
Deleted member 13524
Guest
I don't know where that obscure test came from, but it really didn't.
I think those results are from the slowest 0.1% frames.
Last edited by a moderator:
D
Deleted member 13524
Guest
I didn't know it was from extremetech and the graph itself only says "0.1% framerate" which isn't completely self-explanatory.
DavidGraham
Veteran
I couldn't care less about the redundant useless 3GB 1060, NVIDIA is moronic to use less VRAM with that GPU, which will only serve to hinder it further down the line, same thing that happened to their 3GB high-end Kepler cards, eventually leading to their shortened life span. Heck my old PC has the venerable 3GB 660Ti, sometimes it struggles with max textures @720p!
We seem to diverge here, I say there is a definitive VRAM limitation, just not at 4GB. GPUs with less than 4GB will suffer (like the 3GB 1060), while those with 4GB and more will most likely not. FuryX being hit that hard could be Geometry performance, bad drivers, or bad coding on the developer's part (It's a DX12 game after all, they are ALL plagued with this kind of issues).
Silent_Buddha
Legend
It's easier for non-technical people to understand. That also happens to be the target audience of 99% of review sites.
Regards,
SB
DavidGraham
Veteran
With the imminent release of Gears of War 4, pcgameshardware posted the latest results for Gears 1 at max details:
http://www.pcgameshardware.de/Gears.../Specials/DirectX-12-Benchmarks-Test-1208296/
http://www.pcgameshardware.de/Gears.../Specials/DirectX-12-Benchmarks-Test-1208296/
DavidGraham
Veteran
More Quantum Break DX11 results, NV cards seem to gain massive fps here, 980Ti is now leading FuryX by a respectable margin.
http://www.pcgameshardware.de/Quantum-Break-Spiel-15745/Specials/Technik-Test-Steam-1209192/
In a direct head to head comparison GTX 1080 is up to 25% faster in DX11 than DX12.
Same for GTX 1060, 970 is gaining up to 45% here, but RX 480 is gaining nothing! In the end it seems DX11 of GTX 1060 is faster than both DX11 and DX12 of the RX 480:
http://www.pcgameshardware.de/Quantum-Break-Spiel-15745/Specials/Technik-Test-Steam-1209192/
In a direct head to head comparison GTX 1080 is up to 25% faster in DX11 than DX12.
Same for GTX 1060, 970 is gaining up to 45% here, but RX 480 is gaining nothing! In the end it seems DX11 of GTX 1060 is faster than both DX11 and DX12 of the RX 480:
Last edited:
It's kinda ridiculous that the game can be faster in DX11 vs DX12.
DavidGraham
Veteran
The game was never designed as a true DX12 part, Microsoft forced them to port to DX12 to work with Win 10. Now that they are free from the shackles of a bogus DX12 implementation, performance seems to have returned to "proper" and adequate levels.
Last edited:
DavidGraham
Veteran
More Quantum Break DX11/D12 comparisons from computerbase, showing the same massive (30-40%) gains for NV GPUs. DX11 path for NV is even faster than DX12 path for AMD!
When CPU limited AMD GPUs take a massive hit from the DX11 path:
https://www.computerbase.de/2016-09/quantum-break-steam-benchmark/3/
When CPU limited AMD GPUs take a massive hit from the DX11 path:
https://www.computerbase.de/2016-09/quantum-break-steam-benchmark/3/
Adding to this, somewhere on this forum Sebbbi comments that many dx12 implementations are basically emulation of the dx11 pipeline using dx12 commands, often coming up short.
It will be some time until we see true dx12 engines.
Sent from my iPhone using Tapatalk
Silent_Buddha
Legend
Adding to that, it's quite likely we'll see a true game designed around Vulkan (none yet) before we see a game designed around Dx12 due to it not being limited to Windows 10 and higher.
While I understand the technical limitations for Dx12 implementation on older versions of Windows, it does open up a huge window of opportunity for Vulkan. Something OGL could never take advantage of with Dx10/11's limitations due to OGL being quite crap.
Regards,
SB
D
Deleted member 13524
Guest
There really isn't. At least not for Windows 8, for example.
I like the idea of Vulkan ending up being more successful than DX12.
If only Valve got back to making games...
Silent_Buddha
Legend
It has an opportunity and in the short term should be capable of potentially supplanting Dx12.
However, long term it's going to be contingent on Khronos maintaining enough control to not allow one or more IHVs from turning Vulkan into an extension shit show shackled by legacy crap similar to OGL. Otherwise we'll just see a repeat of Vulkan losing support from developers long term. All assuming it gets big developer support in the near term.
Regards,
SB
D
Deleted member 13524
Guest
Are you sure Khronos has that kind of authority?
When Epic starts including nvidia extensions in their Vulkan path for UE4, what can Khronos do?
Theres an entirely different build with embedded nvapi, and extensions in engine aren't the same as extensions for the api...
DOOM uses gcn intrinsics, why are we not in arms about this too?
It seems that lately whenever anything uses nvidia specific extensions/libraries gameworks whatever it's bad, if amd does it its a huge victory
