KernelIntrinsics API

[christiangnrd]

The goal is to allow for kernels to be written without relying on KernelAbstractions macros

See #562 for initial discussion

@vchuravy @maleadt

[github-actions]

Your PR requires formatting changes to meet the project's style guidelines.
Please consider running Runic (git runic main) to apply these changes.

Click here to view the suggested changes.

diff --git a/src/KernelAbstractions.jl b/src/KernelAbstractions.jl
index 7b626618..1e27573a 100644
--- a/src/KernelAbstractions.jl
+++ b/src/KernelAbstractions.jl
@@ -247,7 +247,7 @@ macro localmem(T, dims)
     # id = gensym("static_shmem")
 
     return quote
-        $SharedMemory($(esc(T)), Val($(esc(dims))))#, Val($(QuoteNode(id))))
+        $SharedMemory($(esc(T)), Val($(esc(dims)))) #, Val($(QuoteNode(id))))
     end
 end
 
diff --git a/src/intrinsics.jl b/src/intrinsics.jl
index 37cbba88..4aaedd24 100644
--- a/src/intrinsics.jl
+++ b/src/intrinsics.jl
@@ -63,7 +63,7 @@ Declare memory that is local to a workgroup.
     ```
     As well as the on-device functionality.
 """
-localmemory(::Type{T}, dims) where T = localmemory(T, Val(dims))
+localmemory(::Type{T}, dims) where {T} = localmemory(T, Val(dims))
 # @inline localmemory(::Type{T}, dims::Val{Dims}) where {T, Dims} = localmemory(T, dims, Val(gensym("static_shmem")))
 
 """
diff --git a/test/intrinsics.jl b/test/intrinsics.jl
index e58b58a6..a586de08 100644
--- a/test/intrinsics.jl
+++ b/test/intrinsics.jl
@@ -20,7 +20,7 @@ end
 
 
 function intrinsics_testsuite(backend, AT)
-    @testset "KernelIntrinsics Tests" begin
+    return @testset "KernelIntrinsics Tests" begin
         @testset "Basic intrinsics functionality" begin
 
             # Test with small kernel

[vchuravy]

Can you rebase?

[vchuravy]

Needs docs.

[vchuravy]

@christiangnrd do you think we need a lower-level kernel launch interface? Otherwise the three-dimensional indices would be superflous.

[christiangnrd]

do you think we need a lower-level kernel launch interface?

I'd been thinking about that and I think so. Would something that, assuming you wrote the whole kernel with KernelIntrinsics, could be used interchangeably with the backend's kernel launch macro (@cuda, @metal, etc.) make sense/be feasible?