Monday, August 31st 2015
Lack of Async Compute on Maxwell Makes AMD GCN Better Prepared for DirectX 12
It turns out that NVIDIA's "Maxwell" architecture has an Achilles' heel after all, which tilts the scales in favor of competing AMD Graphics CoreNext architecture, in being better prepared for DirectX 12. "Maxwell" lacks support for async compute, one of the three highlight features of Direct3D 12, even as the GeForce driver "exposes" the feature's presence to apps. This came to light when game developer Oxide Games alleged that it was pressured by NVIDIA's marketing department to remove certain features in its "Ashes of the Singularity" DirectX 12 benchmark.
Async Compute is a standardized API-level feature added to Direct3D by Microsoft, which allows an app to better exploit the number-crunching resources of a GPU, by breaking down its graphics rendering tasks. Since NVIDIA driver tells apps that "Maxwell" GPUs supports it, Oxide Games simply created its benchmark with async compute support, but when it attempted to use it on Maxwell, it was an "unmitigated disaster." During to course of its developer correspondence with NVIDIA to try and fix this issue, it learned that "Maxwell" doesn't really support async compute at the bare-metal level, and that NVIDIA driver bluffs its support to apps. NVIDIA instead started pressuring Oxide to remove parts of its code that use async compute altogether, it alleges.
"Personally, I think one could just as easily make the claim that we were biased toward NVIDIA as the only "vendor" specific-code is for NVIDIA where we had to shutdown async compute. By vendor specific, I mean a case where we look at the Vendor ID and make changes to our rendering path. Curiously, their driver reported this feature was functional but attempting to use it was an unmitigated disaster in terms of performance and conformance so we shut it down on their hardware. As far as I know, Maxwell doesn't really have Async Compute so I don't know why their driver was trying to expose that. The only other thing that is different between them is that NVIDIA does fall into Tier 2 class binding hardware instead of Tier 3 like AMD which requires a little bit more CPU overhead in D3D12, but I don't think it ended up being very significant. This isn't a vendor specific path, as it's responding to capabilities the driver reports," writes Oxide, in a statement disputing NVIDIA's "misinformation" about the "Ashes of Singularity" benchmark in its press communications (presumably to VGA reviewers).
Given its growing market-share, NVIDIA could use similar tactics to keep game developers away from industry-standard API features that it doesn't support, and which rival AMD does. NVIDIA drivers tell Windows that its GPUs support DirectX 12 feature-level 12_1. We wonder how much of that support is faked at the driver-level, like async compute. The company is already drawing flack for using borderline anti-competitive practices with GameWorks, which effectively creates a walled garden of visual effects that only users of NVIDIA hardware can experience for the same $59 everyone spends on a particular game.
Sources:
DSOGaming, WCCFTech
Async Compute is a standardized API-level feature added to Direct3D by Microsoft, which allows an app to better exploit the number-crunching resources of a GPU, by breaking down its graphics rendering tasks. Since NVIDIA driver tells apps that "Maxwell" GPUs supports it, Oxide Games simply created its benchmark with async compute support, but when it attempted to use it on Maxwell, it was an "unmitigated disaster." During to course of its developer correspondence with NVIDIA to try and fix this issue, it learned that "Maxwell" doesn't really support async compute at the bare-metal level, and that NVIDIA driver bluffs its support to apps. NVIDIA instead started pressuring Oxide to remove parts of its code that use async compute altogether, it alleges.
Given its growing market-share, NVIDIA could use similar tactics to keep game developers away from industry-standard API features that it doesn't support, and which rival AMD does. NVIDIA drivers tell Windows that its GPUs support DirectX 12 feature-level 12_1. We wonder how much of that support is faked at the driver-level, like async compute. The company is already drawing flack for using borderline anti-competitive practices with GameWorks, which effectively creates a walled garden of visual effects that only users of NVIDIA hardware can experience for the same $59 everyone spends on a particular game.
196 Comments on Lack of Async Compute on Maxwell Makes AMD GCN Better Prepared for DirectX 12
latency numbers done by the professionals.
Also, can you link to something that states AMD has priority for HBM2?
As posted earlier in this thread, the original post was from Oxide i.e. read the full post from
www.overclock.net/t/1569897/various-ashes-of-the-singularity-dx12-benchmarks/1200#post_24356995
Sure the game needs much more than a 20% boost that's for sure but still.
steamcommunity.com/app/346110/discussions/0/594820656447032287/
Typical hype bleeding edge crap anyway...
No one seems to remember the geforce FX and how they where not fully DX9 cards...
By the time games started using DX9 everyone had a geforce 6XXX or 7XXX.
DX11 came out in 2010?
How many games now use DX 11? A few more but it took a few years....
On top of that ...
Hopefully steam pulls there head out of there arse and makes steam os decent.... And developers start developing for linux.
As I don't like some of the features in windows 10....
This is not how it works, we are talking about performance in games, and not features or a tech-demo like engines abusing a single api feature. Think about the performance impact and usage of dx11.1 or 11.2 in games, because I can't recall if it ever made a huge impact in any game, and I read everything from dev documents through beyond3d to reddit or this forum. There are features which Intel does the best, and one could write a program which would abuse such feature (let's say ROV) to beat both NV and AMD, yet nobody in their right mind would think that Intel has a chance in games against the big boys. On the subject, Kepler and Fermi might be slower indeed, but don't forget that they had a wider bus which could come handy for them to keep a bit up with the younger chips, I think they won't be that bad with dx12 titles, but we will see.
AMD came with DX10.1 back then and what's happened ?!
Our "pigs in the mud" nvidia damaged its progress as well.
When they throw so much money so some titles run better on their hardware, why didn't they even a single time think of the option to make some progress, not to spoil all customers' experience ?!
www.anandtech.com/show/2549/7
As for HBM-2, I have a sneaking suspicion there will be no mass production of HBM and will only be mass production of HBM-2. AMD is the only client for HBM, no? Unless SK Hynix can get more buyers, ramping up production doesn't make much sense. The bulk of memory orders are still DDR3, DDR3L, DDR4, and GDDR5. I wonder what SK Hynix said to AMD to get them to sign up. HBM-2 may be amazing but HBM seems like more trouble than its worth.
I am going to buy for my friends the R9 280 for 168 euros now.
And it won't be my new card - I will recommend and buy several R9 280 because there is nothing better, for my friends. ;)
It can't be in performance /watt...
It can't be in frame time in CFx v SLI...
It can't be in highest FPS/performance...
Bang for your buck? CHECK.
Utilizing technology to get (TOO FAR) ahead of the curve? CHECK.
I'm spent.
bon appétitI think rvalencia is attempting to bridge that divide.