• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

VUDA is a CUDA-Like Programming Interface for GPU Compute on Vulkan (Open-Source)

But it's not OpenCL. It's using the flexibility of Vulkan to execute non-rendering code. It's impossible to predict how it performs because it comes at the problem from a very different angle.
But the argument was not about how it works. It was about VUDA replacing CUDA (making easy ports possible). That is not true. The code is different. End of story. :)
You remember Mantle, how cool that was? It's called Vulkan now & is in your PC, probably. If we didn't encourage free alternatives, we'd never have Linux or Android.
Linux is not "free". You're just paying with time, not with money. It's just a different way of paying. Money is better - that's why we invented it.
Android is almost impossible to evaluate. As a client, you can't download Android and use it on your PC (well... you can, but it's pointless). On the other hand, Android is in the end the main source of income for Google.

Mantle was a joint venture of AMD and gaming studios. It wasn't a hobby project run by some bloke.

Also, isn't it funny that Mantle was mentioned next to claims that AMD is such a FOSS and Linux contributor? :-D
 
Last edited:
But the argument was not about how it works. It was about VUDA replacing CUDA (making easy ports possible). That is not true. The code is different. End of story. :)
But how difficult would it be to convert/port over?
As a client, you can't download Android and use it on your PC (well... you can, but it's pointless).
http://www.android-x86.org/
http://www.phoenixos.com/download_x86
Solidly stable, supporting tons of hardware and very useful. You were saying?
 
You remember Mantle, how cool that was? It's called Vulkan now & is in your PC, probably. If we didn't encourage free alternatives, we'd never have Linux or Android.
I remember Mantle.

I also remember that it wasn't a 2 day project with 0 releases by a single dude on github, when I got excited about it.
 
So basically using Vulkan's compute shader but wrap it up to make it look like CUDA?
 
But the argument was not about how it works. It was about VUDA replacing CUDA (making easy ports possible). That is not true. The code is different. End of story. :)
vuda class calls instead of cuda class calls, of course it is different in that regard, but performance matters. If, for example, VUDA performs only 10% slower than CUDA, VUDA code could be widely implemented in software with no specific CUDA path because it's not worth debugging. VUDA could get fairly wide spread use.

http://www.android-x86.org/
http://www.phoenixos.com/download_x86
Solidly stable, supporting tons of hardware and very useful. You were saying?
I play some Android companion games through BlueStacks Android emulator on my computer (don't have an Android phone). Never ran into issues other than not being able to "touch" in two places like smart phones support.

So basically using Vulkan's compute shader but wrap it up to make it look like CUDA?
Yup. CUDA-like code executed through Vulkan.
 
vuda class calls instead of cuda class calls, of course it is different in that regard, but performance matters. If, for example, VUDA performs only 10% slower than CUDA, VUDA code could be widely implemented in software with no specific CUDA path because it's not worth debugging. VUDA could get fairly wide spread use.
But why? What makes VUDA so special that you couldn't do with OpenCL already?

And BTW: 10% performance gap is very optimistic. :-D
 
But why? What makes VUDA so special that you couldn't do with OpenCL already?

And BTW: 10% performance gap is very optimistic. :-D
CUDA developers/existing code ported to/started on VUDA so it's hardware agnostic.
 
Targeting Vulkan devices is probably a smart move. A lot of GPUs support it at this point and it's not like you're installing additional SDKs or anything to work with it. Personally I think this is a very interesting and ambitious idea.
 
Back
Top