Skip to content

Adding a skeleton for an AMDGPU extension. #86

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
merged 1 commit into from
Mar 30, 2025
Merged

Adding a skeleton for an AMDGPU extension. #86

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
5 changes: 4 additions & 1 deletion Project.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,25 +18,28 @@ Tensors = "48a634ad-e948-5137-8d70-aa71f2a747f4"

[weakdeps]
Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
AMDGPU = "21141c5a-9bdb-4563-92ae-f87d6854732e"
CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
Exodus = "f57ae99e-f805-4780-bdca-96e224be1e5a"

[extensions]
FiniteElementContainersAdaptExt = "Adapt"
FiniteElementContainersAMDGPUExt = ["Adapt", "AMDGPU"]
FiniteElementContainersCUDAExt = ["Adapt", "CUDA"]
FiniteElementContainersExodusExt = "Exodus"

[compat]
AcceleratedKernels = "0.3"
Adapt = "3, 4"
AMDGPU = "1"
Aqua = "0.8"
Atomix = "1"
CUDA = "5"
DocStringExtensions = "0.9"
Exodus = "0.13"
JET = "0.9"
Krylov = "0.9"
KernelAbstractions = "0.9"
Krylov = "0.9"
LinearAlgebra = "1"
Parameters = "0.12"
Reexport = "1"
Expand Down
26 changes: 26 additions & 0 deletions ext/FiniteElementContainersAMDGPUExt.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
module FiniteElementContainersAMDGPUExt

using Adapt
using AMDGPU
using FiniteElementContainers
using KernelAbstractions

FiniteElementContainers.gpu(x) = adapt_structure(ROCArray, x)

Check warning on line 8 in ext/FiniteElementContainersAMDGPUExt.jl

View check run for this annotation

Codecov / codecov/patch

ext/FiniteElementContainersAMDGPUExt.jl#L8 

Added line #L8 was not covered by tests

function AMDGPU.rocSPARSE.ROCSparseMatrixCSC(asm::SparseMatrixAssembler)

Check warning on line 10 in ext/FiniteElementContainersAMDGPUExt.jl

View check run for this annotation

Codecov / codecov/patch

ext/FiniteElementContainersAMDGPUExt.jl#L10 

Added line #L10 was not covered by tests
# TODO Not sure what the AMD Backend is called in KernelAbstractions
# I couldn't quite figure it out. This assert statement below though
# would be good for error checking and device consistency.
# @assert typeof(get_backend(asm)) <: CUDABackend "Assembler is not on a CUDA device"
@assert length(asm.pattern.cscnzval) > 0 "Need to assemble the assembler once with SparseArrays.sparse!(assembler)"
@assert all(x -> x != zero(eltype(asm.pattern.cscnzval)), asm.pattern.cscnzval) "Need to assemble the assembler once with SparseArrays.sparse!(assembler)"
n_dofs = FiniteElementContainers.num_unknowns(asm.dof)
return AMDGPU.rocSPARSE.ROCSparseMatrixCSC(

Check warning on line 18 in ext/FiniteElementContainersAMDGPUExt.jl

View check run for this annotation

Codecov / codecov/patch

ext/FiniteElementContainersAMDGPUExt.jl#L15-L18 

Added lines #L15 - L18 were not covered by tests
asm.pattern.csccolptr,
asm.pattern.cscrowval,
asm.pattern.cscnzval,
(n_dofs, n_dofs)
)
end

end # module
Loading