Skip to content
Merged
Changes from all commits
Commits
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
49 changes: 49 additions & 0 deletions test/test.jl
Original file line number Diff line number Diff line change
Expand Up @@ -331,5 +331,54 @@ function unittest_testsuite(Backend, backend_str, backend_mod, BackendArrayT; sk
end
end

# from https://github.com/JuliaGPU/KernelAbstractions.jl/issues/652
@kernel function unaliased_accumulate!(output, input, n)
i, j = @index(Global, NTuple)

for k in j:n
output[i, j] += input[i, k]
end
end

@kernel function unaliased_accumulate_local!(output, input, n)
i, j = @index(Global, NTuple)

# Use local accumulator
sum_val = zero(eltype(output))
for k in j:n
sum_val += input[i, k]
end
output[i, j] = sum_val
end

@testset "unaliased accumulate" begin
backend = Backend()
N = 8
M = 5

input = Float32[i + k for i in 1:M, k in 1:N]

reference = zeros(Float32, M, N)
for i in 1:M
for j in 1:N
for k in j:N
reference[i, j] += input[i, k]
end
end
end

# Allocate device arrays
input = adapt(backend, input)
output = KernelAbstractions.zeros(backend, Float32, M, N)

# Perform accumulation
unaliased_accumulate!(backend)(output, input, N; ndrange = size(output))
@test adapt(Array, output) == reference

fill!(output, 0)
unaliased_accumulate_local!(backend)(output, input, N; ndrange = size(output))
@test adapt(Array, output) == reference
end

return
end
Loading