Skip to content

Gate the pinned-memory system-scope atomic test on host native atomics #3271

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 1 commit into from
Sep 14, 2026
+10 −6
Merged

Gate the pinned-memory system-scope atomic test on host native atomics #3271

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
16 changes: 10 additions & 6 deletions test/core/device/intrinsics/atomics.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -621,12 +621,16 @@ if system_scope_supported
@test Array(a) == [1024, 2048 - 1024]

# Smoke-test host-pinned memory. Synchronization precedes the host read; this
# does not test concurrent CPU/GPU atomicity.
counter = Int32[0]
a = unsafe_wrap(CuArray{Int32,1,CUDA.HostMemory}, counter)
@cuda threads=1024 add_kernel(a, Val(:system))
synchronize()
@test counter[1] == 1024
# does not test concurrent CPU/GPU atomicity. System-scope atomics on pinned
# memory are only atomic at system scope when the host supports native atomics
# (e.g., NVLink C2C); compute-sanitizer flags them as invalid otherwise.
if attribute(device(), CUDA.DEVICE_ATTRIBUTE_HOST_NATIVE_ATOMIC_SUPPORTED) == 1
counter = Int32[0]
a = unsafe_wrap(CuArray{Int32,1,CUDA.HostMemory}, counter)
@cuda threads=1024 add_kernel(a, Val(:system))
synchronize()
@test counter[1] == 1024
end
end
end

Expand Down