Skip to content

Some cudadrv tests #2684

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
maleadt merged 2 commits into from
Mar 10, 2025
Merged

Some cudadrv tests #2684

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
dbb2215
Some cudadrv tests
kshyatt Mar 8, 2025
855c8e4
Add another idempotent texture method and tests
kshyatt Mar 9, 2025
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
10 changes: 10 additions & 0 deletions test/base/texture.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,6 +51,8 @@ end
texarr1D = CuTextureArray(d_a1D)
tex1D = CuTexture(texarr1D)
@test fetch_all(tex1D) == d_a1D
array_mem = texarr1D.mem
@test startswith(sprint(show, array_mem), "$testheight-element ArrayMemory")

texarr2D = CuTextureArray(d_a2D)
tex2D = CuTexture(texarr2D)
Expand All @@ -59,6 +61,14 @@ end
texarr3D = CuTextureArray(d_a3D)
tex3D = CuTexture(texarr3D)
@test fetch_all(tex3D) == d_a3D

# test the underlying ArrayMemory too
array_mem = texarr3D.mem
@test_throws ErrorException("Opaque array memory does not have a definite size") sizeof(array_mem)
@test size(array_mem) == size(d_a3D)
@test length(array_mem) == length(d_a3D)
@test ndims(array_mem) == ndims(d_a3D)

end

@testset "CuTextureArray(::Array)" begin
Expand Down
35 changes: 23 additions & 12 deletions test/core/cudadrv.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -435,8 +435,15 @@ nb = sizeof(data)
typed_pointer(buf::Union{CUDA.DeviceMemory, CUDA.UnifiedMemory}, T) = convert(CuPtr{T}, buf)
typed_pointer(buf::CUDA.HostMemory, T) = convert(Ptr{T}, buf)

# allocations and copies
for srcTy in [CUDA.DeviceMemory, CUDA.HostMemory, CUDA.UnifiedMemory],
@testset "showing" begin
for (Ty, str) in zip([CUDA.DeviceMemory, CUDA.HostMemory, CUDA.UnifiedMemory], ("DeviceMemory", "HostMemory", "UnifiedMemory"))
dummy = CUDA.alloc(Ty, 0)
@test startswith(sprint(show, dummy), str)
CUDA.free(dummy)
end
end

@testset "allocations and copies, src $srcTy dst $dstTy" for srcTy in [CUDA.DeviceMemory, CUDA.HostMemory, CUDA.UnifiedMemory],
dstTy in [CUDA.DeviceMemory, CUDA.HostMemory, CUDA.UnifiedMemory]

dummy = CUDA.alloc(srcTy, 0)
Expand Down Expand Up @@ -472,6 +479,7 @@ for srcTy in [CUDA.DeviceMemory, CUDA.HostMemory, CUDA.UnifiedMemory],

# test device with context in which pointer was allocated.
@test device(typed_pointer(src, T)) == device()
@test context(typed_pointer(src, T)) == context()
if !memory_pools_supported(device())
# NVIDIA bug #3319609
@test context(typed_pointer(src, T)) == context()
Expand All @@ -495,17 +503,15 @@ for srcTy in [CUDA.DeviceMemory, CUDA.HostMemory, CUDA.UnifiedMemory],
CUDA.free(dst)
end

# pointer attributes
let
@testset "pointer attributes" begin
src = CUDA.alloc(CUDA.DeviceMemory, nb)

attribute!(typed_pointer(src, T), CUDA.POINTER_ATTRIBUTE_SYNC_MEMOPS, 0)

CUDA.free(src)
end

# asynchronous operations
let
@testset "asynchronous operations" begin
src = CUDA.alloc(CUDA.DeviceMemory, nb)

unsafe_copyto!(typed_pointer(src, T), pointer(data), N; async=true)
Expand All @@ -515,8 +521,7 @@ let
CUDA.free(src)
end

# pinned memory
let
@testset "pinned memory" begin
# create a pinned and mapped buffer
src = CUDA.alloc(CUDA.HostMemory, nb, CUDA.MEMHOSTALLOC_DEVICEMAP)

Expand Down Expand Up @@ -547,10 +552,17 @@ if attribute(device(), CUDA.DEVICE_ATTRIBUTE_HOST_REGISTER_SUPPORTED) != 0
@test ref == data

CUDA.unregister(src)

# with a RefValue
src = Ref{T}(T(42))
CUDA.pin(src)
cpu_ptr = Base.unsafe_convert(Ptr{T}, src)
ref = Array{T}(undef, 1)
unsafe_copyto!(pointer(ref), cpu_ptr, 1)
@test ref == [T(42)]
end

# unified memory
let
@testset "unified memory" begin
src = CUDA.alloc(CUDA.UnifiedMemory, nb)

@test_throws BoundsError CUDA.prefetch(src, 2*nb; device=CUDA.DEVICE_CPU)
Expand All @@ -571,8 +583,7 @@ let
CUDA.free(src)
end

# 3d memcpy
let
@testset "3d memcpy" begin
# TODO: use cuMemAllocPitch (and put pitch in buffer?) to actually get benefit from this

data = collect(reshape(1:27, 3, 3, 3))
Expand Down