Skip to content

Refactor(GPU): switch to per-stream synchronization in CUDA and ROCm backend #699

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

Closed
victorcamaraa wants to merge 4 commits into from
Closed

Refactor(GPU): switch to per-stream synchronization in CUDA and ROCm backend #699

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
ce8f4ad
refactor: change CUDA and ROCm backends to per-stream syncronization
victorcamaraa Apr 15, 2026
b464070
add teste cuda
victorcamaraa Apr 16, 2026
96d48db
add ROCm test, both tests were written by claude
victorcamaraa Apr 16, 2026
0682040
Remove both test files
victorcamaraa Apr 16, 2026
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
4 changes: 2 additions & 2 deletions ext/CUDAExt.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,7 +104,7 @@ end

function _sync_with_context(x::Union{Dagger.Processor,Dagger.MemorySpace})
with_context(x) do
CUDA.synchronize()
CUDA.synchronize(stream())
end
end
function sync_with_context(x::Union{Dagger.Processor,Dagger.MemorySpace})
Expand Down Expand Up @@ -391,7 +391,7 @@ Dagger.gpu_with_device(f, proc::CuArrayDeviceProc) =
CUDA.device!(f, proc.device)
function Dagger.gpu_synchronize(proc::CuArrayDeviceProc)
with_context(proc) do
CUDA.synchronize()
CUDA.synchronize(stream())
end
end
function Dagger.gpu_synchronize(::Val{:CUDA})
Expand Down
4 changes: 2 additions & 2 deletions ext/ROCExt.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -98,7 +98,7 @@ end

function _sync_with_context(x::Union{Dagger.Processor,Dagger.MemorySpace})
with_context(x) do
AMDGPU.synchronize()
AMDGPU.synchronize(stream())
end
end
function sync_with_context(x::Union{Dagger.Processor,Dagger.MemorySpace})
Expand Down Expand Up @@ -364,7 +364,7 @@ Dagger.gpu_with_device(f, proc::ROCArrayDeviceProc) =
AMDGPU.device!(f, AMDGPU.devices()[proc.device_id])
function Dagger.gpu_synchronize(proc::ROCArrayDeviceProc)
with_context(proc) do
AMDGPU.synchronize()
AMDGPU.synchronize(stream())
end
end
function Dagger.gpu_synchronize(::Val{:ROC})
Expand Down
Loading