Argoon1981
New member
It seams a large number of developers are singing praises for async computing.
wccftech.com/async-compute-praised-by-several-devs
wccftech.com/async-compute-praised-by-several-devs
dev's love async computer
- Thread starter Argoon1981
- Start date
noko
New member
The hardware design allows better utilization of shaders allowing for a larger number of shaders in a GPU and still be usefuul. Nvidia went with speed and less shaders using fast context switching methods (more serial design), AMD went with more shaders and a more Parallel design.
To use AMD hardware better, using Async compute/shaders will be very beneficial. For multigpu cards having more control over graphics and compute operations (compute is being used more and more for special effects, hair rendering, post processing, lighting etc.) can only help having this flexibility making AMD hardware potentially better for multipe gpu operations. With AMD giving out Fury Pro Duo's to developers will also help in seeing more developement with multiple gpu's. Which is the direction is going with GPU's possibly having mutiple GPU's on a HBM module.
With the large base of AMD hardware consoles where the large portion of game development profits come from, the use of using Async Compute I can only assume twill increase and will be carried over to the PC games. AMD has several generations now that supports this and the upcoming great perf/$ 480 cards coming out will have a large base of cards that will benefit from Async Compute/shaders of AMD.
Now I do see Nvidia speeding up their design to support more shaders after Polaris depending upon initial sells which I do believe AMD will dominate if they have the manufacturing capability to keep up with demand on the Mainstream.
To use AMD hardware better, using Async compute/shaders will be very beneficial. For multigpu cards having more control over graphics and compute operations (compute is being used more and more for special effects, hair rendering, post processing, lighting etc.) can only help having this flexibility making AMD hardware potentially better for multipe gpu operations. With AMD giving out Fury Pro Duo's to developers will also help in seeing more developement with multiple gpu's. Which is the direction is going with GPU's possibly having mutiple GPU's on a HBM module.
With the large base of AMD hardware consoles where the large portion of game development profits come from, the use of using Async Compute I can only assume twill increase and will be carried over to the PC games. AMD has several generations now that supports this and the upcoming great perf/$ 480 cards coming out will have a large base of cards that will benefit from Async Compute/shaders of AMD.
Now I do see Nvidia speeding up their design to support more shaders after Polaris depending upon initial sells which I do believe AMD will dominate if they have the manufacturing capability to keep up with demand on the Mainstream.
noko
New member
Async Compute is in DX12 but that does not mean Async Shaders. Nvidia can schedule compute items (up to 32 separate compute) on a priority basis. So if you have a hair compute for position based upon wind, gravity etc. with a number of post compute image processes Nvidia can just process all of them using all the streaming processors as needed switching back to graphics. AMD can work on graphics and compute operations at the same time but you are now divvying up the number of streaming processors for both, less for graphics and less for compute operations.
The bottom line is who ever performs the best per price is the winner. Either method is good in DX12. Nvidia faster clocks has less of a hit for switching.
This will be interesting to see any advantages to either one.
Nagorak
Well-known member
AMD really may have scored a coup by being in all the consoles. It almost guarantees that their graphics tech will be used, since so many games are designed to be multiplatform. It pretty much trumps Nvidia's GameWorks efforts.
noko
New member
If the PC side AMD user base increases greatly then this will help carry over those type of optimization to the PC without bending over backwards for Nvidia GPU's. One benefit is that AMD has several generations and APU's supporting this which can only help.