Skip to content

CUDA: fix Float64 add and float swap on device #81

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鈥檒l occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
vchuravy merged 1 commit into from
Sep 8, 2026
Merged

CUDA: fix Float64 add and float swap on device #81

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
11 changes: 6 additions & 5 deletions ext/AtomixCUDAExt.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,7 +48,7 @@ end
@inline modify_native!(ptr::LLVMPtr{Float32}, ::typeof(-), x) = CUDA.atomic_sub!(ptr, x)
# Float64 atomic add needs compute capability 6.0; use compare-and-swap below that.
@inline function modify_native!(ptr::LLVMPtr{Float64}, ::typeof(+), x)
if CUDA.compute_capability() >= v"6.0"
if CUDA.compute_capability().major >= 6
CUDA.atomic_add!(ptr, x)
else
modify_cas!(ptr, +, x)
Expand All @@ -59,10 +59,11 @@ end
# swap: exchange floats through their integer representation
@inline modify_native!(ptr::LLVMPtr{<:NativeInt}, ::typeof(right), x) =
CUDA.atomic_xchg!(ptr, x)
@inline function modify_native!(ptr::LLVMPtr{T,A}, ::typeof(right), x) where {T<:NativeFloat,A}
I = CUDA.inttype(T)
old = CUDA.atomic_xchg!(reinterpret(LLVMPtr{I,A}, ptr), reinterpret(I, x))
return reinterpret(T, old)
for (T, I) in [Float32 => UInt32, Float64 => UInt64]
@eval @inline function modify_native!(ptr::LLVMPtr{$T,A}, ::typeof(right), x) where {A}
old = CUDA.atomic_xchg!(reinterpret(LLVMPtr{$I,A}, ptr), reinterpret($I, x))
return reinterpret($T, old)
end
end

# everything else (float min/max, arbitrary functions): compare-and-swap loop
Expand Down
22 changes: 20 additions & 2 deletions test/test_atomix_cuda.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,7 +108,7 @@ end


@testset "AtomixCUDAExt:test_float" begin
A = CUDA.CuVector(Float32[1, 1, 1, 1, 1, 0])
A = CUDA.CuVector(Float32[1, 1, 1, 1, 1, 0, 0, 1])
cuda() do
GC.@preserve A begin
@atomic A[1] += 1.5f0
Expand All @@ -118,7 +118,25 @@ end
# no native instruction: compare-and-swap loop
pre, post = @atomic A[5] * 4f0
A[6] = pre + post
A[7] = @atomicswap A[8] = 8f0
end
end
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5]
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5, 1, 8]
end


@testset "AtomixCUDAExt:test_float64" begin
A = CUDA.CuVector(Float64[1, 1, 1, 1, 1, 0, 0, 1])
cuda() do
GC.@preserve A begin
@atomic A[1] += 1.5
@atomic A[2] -= 0.5
@atomic max(A[3], 3.0)
@atomic min(A[4], -1.0)
pre, post = @atomic A[5] * 4.0
A[6] = pre + post
A[7] = @atomicswap A[8] = 8.0
end
end
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5, 1, 8]
end
5 changes: 3 additions & 2 deletions test/test_atomix_metal.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,7 +124,7 @@ end


@testset "AtomixMetalExt:test_float" begin
A = Metal.MtlVector(Float32[1, 1, 1, 1, 1, 0])
A = Metal.MtlVector(Float32[1, 1, 1, 1, 1, 0, 0, 1])
metal() do
GC.@preserve A begin
@atomic A[1] += 1.5f0
Expand All @@ -134,7 +134,8 @@ end
# no native instruction: compare-and-swap loop
pre, post = @atomic A[5] * 4f0
A[6] = pre + post
A[7] = @atomicswap A[8] = 8f0
end
end
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5]
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5, 1, 8]
end
5 changes: 3 additions & 2 deletions test/test_atomix_oneapi.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,7 +108,7 @@ end


@testset "AtomixoneAPIExt:test_float" begin
A = oneAPI.oneVector(Float32[1, 1, 1, 1, 1, 0])
A = oneAPI.oneVector(Float32[1, 1, 1, 1, 1, 0, 0, 1])
oneapi() do
GC.@preserve A begin
@atomic A[1] += 1.5f0
Expand All @@ -118,7 +118,8 @@ end
# no native instruction: compare-and-swap loop
pre, post = @atomic A[5] * 4f0
A[6] = pre + post
A[7] = @atomicswap A[8] = 8f0
end
end
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5]
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5, 1, 8]
end
22 changes: 20 additions & 2 deletions test/test_atomix_opencl.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,7 +108,7 @@ end


@testset "AtomixOpenCLExt:test_float" begin
A = OpenCL.CLArray(Float32[1, 1, 1, 1, 1, 0])
A = OpenCL.CLArray(Float32[1, 1, 1, 1, 1, 0, 0, 1])
opencl() do
GC.@preserve A begin
@atomic A[1] += 1.5f0
Expand All @@ -118,7 +118,25 @@ end
# no native instruction: compare-and-swap loop
pre, post = @atomic A[5] * 4f0
A[6] = pre + post
A[7] = @atomicswap A[8] = 8f0
end
end
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5]
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5, 1, 8]
end


@testset "AtomixOpenCLExt:test_float64" begin
A = OpenCL.CLArray(Float64[1, 1, 1, 1, 1, 0, 0, 1])
opencl() do
GC.@preserve A begin
@atomic A[1] += 1.5
@atomic A[2] -= 0.5
@atomic max(A[3], 3.0)
@atomic min(A[4], -1.0)
pre, post = @atomic A[5] * 4.0
A[6] = pre + post
A[7] = @atomicswap A[8] = 8.0
end
end
@test collect(A) == [2.5, 0.5, 3, -1, 4, 5, 1, 8]
end
Loading