On Tuesday, 30 May 2017 at 18:06:56 UTC, Walter Bright wrote:
On 5/30/2017 5:12 AM, Nicholas Wilson wrote:
Ah, isn't English wonderful. I guess Walter is suffering the
inverse of the Calvin & Hobbes "Verbing nouns weirds the
language", nouning verbs does weird the language, but only to
those who aren't used to that particular nouning of the verb.
Just to clarify, I find that "Compute" is not evocative of
"GPU". I read "CUDA by Example" a couple years ago, and even
downloaded the CUDA SDK and compiled/ran a simple program on a
graphics card. But I never noticed that "Compute" had anything
specific to do with GPU programming.
I fear the conversation will go like this, like it has for me:
W: What's DCompute?
N: Enables GPU programming with D
W: Cool! I can use D to program GPUs!
The problem with the first conversation is W may just move on
to the next topic rather than investigate what DCompute is.
OK, I get that a portion of the community seems to think that
dcompute is either too generic, is too easily passed over by the
fact that it doesn't mention GPUs or otherwise could be named
I am more inclined to be persuaded by the fact that everybody
that has actually done GPU programming has said that it makes
sense to them.
But can we please reduce the bike shedding, that was not the
point of this announcement, which was to bring together people
interested in contributing to either the runtime libraries, LDC
or LLVM and become familiar with goals, expected route of
development and infrastructure.
Perhaps there will be scope for renaming if/when this also
includes graphics when either OpenCL is merged into the Vulkan
API or Petar Kirov gets Vulkan SPIRV generation going on LLVM,
but for now the name stays.