Skip to content

Restrict native shuffle ops to Float32, Int32, and UInt32 for Metal #7

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
epilliat merged 1 commit into from
May 6, 2026
Merged

Restrict native shuffle ops to Float32, Int32, and UInt32 for Metal #7

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 2 additions & 7 deletions ext/Metal/shuffle_vote.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ const SHFL_DISPATCH = Dict(
Down => :(Metal.simd_shuffle_down)
)

for T in (:Float32, :Int32, :UInt32, :Float16, :Int16, :UInt16)
for T in (:Float32, :Int32, :UInt32)
for (direction, metal_fname) in SHFL_DISPATCH
@eval begin
Base.Experimental.@overlay Metal.method_table @inline _shfl(::Type{$direction}, mask, val::$T, src) =
Expand All @@ -21,13 +21,8 @@ end
# Will update when they are part of official release
# =======================================================================================================
simd_shuffle_map = ((Float32, "f32"),
(Float16, "f16"),
(Int32, "s.i32"),
(UInt32, "u.i32"),
(Int16, "s.i16"),
(UInt16, "u.i16"),
(Int8, "s.i8"),
(UInt8, "u.i8"))
(UInt32, "u.i32"))

for (jltype, suffix) in simd_shuffle_map
@eval begin
Expand Down
19 changes: 18 additions & 1 deletion test/tests/shfl.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,12 @@ struct ComplexType
z::Float64
end

struct ComplexTypeMetal
x::Int32
y::SubType
z::Float16
end

# ─────────────────────────────────────────────────────────────────────────────
# Helpers
# ─────────────────────────────────────────────────────────────────────────────
Expand Down Expand Up @@ -126,7 +132,18 @@ end
end
end

if TEST_BACKEND != "metal"
if TEST_BACKEND == "metal"
@testset "Up ComplexTypeMetal" begin
src = to_device([ComplexTypeMetal(i, SubType(Float16(i), UInt8(i)), Float16(i)) for i in 1:warpsz])
dst = to_device([ComplexTypeMetal(0, SubType(Float16(0), UInt8(0)),Float16(0.0)) for _ in 1:warpsz])
launch(shfl_up_kernel, dst, src, 1; ndrange=warpsz)
result = from_device(dst)
@test result[1] == ComplexTypeMetal(1, SubType(Float16(1), UInt8(1)), Float16(1.0))
for i in 2:warpsz
@test result[i] == ComplexTypeMetal(i - 1, SubType(Float16(i - 1), UInt8(i - 1)), Float16(i - 1))
end
end
else
@testset "Up ComplexType" begin
src = to_device([ComplexType(i, SubType(Float16(i), UInt8(i)), Float64(i)) for i in 1:warpsz])
dst = to_device([ComplexType(0, SubType(Float16(0), UInt8(0)), 0.0) for _ in 1:warpsz])
Expand Down
Loading