Skip to content

Avoid boxing the views captured by the threaded BLAS multiplication #94

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
lkdvos merged 2 commits into from
Oct 14, 2026
Merged

Avoid boxing the views captured by the threaded BLAS multiplication #94

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
80e18f5
Avoid boxing the views captured by the threaded BLAS multiplication
lkdvos Oct 13, 2026
4a045f2
Test that BLAS multiplication with StridedView does not allocate
lkdvos Oct 14, 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
2 changes: 1 addition & 1 deletion src/linalg.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -148,7 +148,7 @@ function _threaded_blas_mul!(
end
nthreads1 = nthreads >> 1
nthreads2 = nthreads - nthreads1
t = Threads.@spawn _threaded_blas_mul!(Chead, Ahead, CA, Bhead, CB, α, β, nthreads1)
t = Threads.@spawn _threaded_blas_mul!($Chead, $Ahead, CA, $Bhead, CB, α, β, nthreads1)
_threaded_blas_mul!(Ctail, Atail, CA, Btail, CB, α, β, nthreads2)
wait(t)
return C
Expand Down
15 changes: 15 additions & 0 deletions test/blasmultests.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,3 +96,18 @@ end
C1 = StridedView(copy(C))
@test mul!(C, A, B, α, β) ≈ mul!(C1, A1, B1, α, β)
end

# small enough to never be split over threads, such that no tasks need to be spawned
blasmul_allocations(C, A, B, α, β) = @allocated mul!(C, A, B, α, β)
@testset "Matrix multiplication with StridedView does not allocate: $T" for T in (Float64, ComplexF64)
d = 8
A = StridedView(rand(T, (d, d)))
B = StridedView(rand(T, (d, d)))
C = StridedView(rand(T, (d, d)))
α = rand(T)
β = rand(T)
for op1 in (identity, conj, transpose, adjoint), op2 in (identity, conj, transpose, adjoint)
blasmul_allocations(C, op1(A), op2(B), α, β) # compile
@test blasmul_allocations(C, op1(A), op2(B), α, β) == 0
end
end
Loading