2 ms·

Yeah, I know, that's why I mentioned it being a target. I realize it's meant to be used for extremely low-level stuff. Which is fine, I don't get annoyed at x86

Yeah, I know, that's why I mentioned it being a target. I realize it's meant to be used for extremely low-level stuff. Which is fine, I don't get annoyed at x86_64 assembly being difficult to write. As I said, the thing that bothers me is when people act like Vulkan is a replacement for OpenGL (presumably because they've never actually written either and they just saw some YouTube demos), and it really isn't, or at least that's an incomplete picture. It's a replacement for OpenGL in the same way writing JVM bytecode directly is a replacement for Java; they fundamentally do the same or similar things, but one is sort of categorically different in its goal. I do wish that there was a more consumer-friendly API announced in addition to Vulkan though; I'm fine with replacing OpenGL with something newer and better-fitting for newer cards, but that's not what we got. That's why I was trying to say that I was annoyed that Apple didn't open source Metal, because I feel like it really could have taken that role. Also, while I'm aware that Vulkan doesn't inherently imply "graphics", I would also like to point out that CUDA and rocM are also considerably easier to write than the Vulkan equivalents, at least in the little examples I played with, though I will acknowledge that I felt Vulkan was better than OpenCL.

hgs3 2y ago
> I would also like to point out that CUDA and rocM are also considerably easier to write than the Vulkan equivalents I agree that Vulkan has more boilerplate, but I think that's because Vulkan is an open standard and makes fewer assumptions which means the API has more ceremony. For example Vulkan might be implemented for custom embedded hardware; its design helps give the manufacturer more leeway. CUDA and ROCm were designed by Nvidia and AMD for their own hardware and so they can bake in low-level assumptions which means a more streamlined API.
- tombert 2y ago
  Well ROCm is more or less designed to be portable between both; that's not strictly true but it's fairly similar to CUDA so I don't know how much of that boilerplate is reduced by the fact that it's AMD specific. Your point is fair though; Vulkan kind of runs on almost-literally everything made in the last nine years or so, so it can assume basically zero about any kind of underlying OS stuff. Still, Vulkan has convinced me that I have no desire to work that low-level of GPU programming, as I had zero fun playing with it. I guess I like having one layer up of abstraction; I'm happy enough to write CUDA, and I'm happy enough to write anything higher level than that.
garaetjjte 2y ago
>I do wish that there was a more consumer-friendly API announced in addition to Vulkan though Maybe WebGPU could be that?

hgs3 2y ago

> I would also like to point out that CUDA and rocM are also considerably easier to write than the Vulkan equivalents I agree that Vulkan has more boilerplate, but I think that's because Vulkan is an open standard and makes fewer assumptions which means the API has more ceremony. For example Vulkan might be implemented for custom embedded hardware; its design helps give the manufacturer more leeway. CUDA and ROCm were designed by Nvidia and AMD for their own hardware and so they can bake in low-level assumptions which means a more streamlined API.

tombert 2y ago
Well ROCm is more or less designed to be portable between both; that's not strictly true but it's fairly similar to CUDA so I don't know how much of that boilerplate is reduced by the fact that it's AMD specific. Your point is fair though; Vulkan kind of runs on almost-literally everything made in the last nine years or so, so it can assume basically zero about any kind of underlying OS stuff. Still, Vulkan has convinced me that I have no desire to work that low-level of GPU programming, as I had zero fun playing with it. I guess I like having one layer up of abstraction; I'm happy enough to write CUDA, and I'm happy enough to write anything higher level than that.

garaetjjte 2y ago

>I do wish that there was a more consumer-friendly API announced in addition to Vulkan though Maybe WebGPU could be that?