Skip to content

[pocl] Back @private Scratchpad with GPUCompiler.alloca #714

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

Draft
vchuravy wants to merge 2 commits into
base: main
Choose a base branch
from
+6 −5
Draft

[pocl] Back @private Scratchpad with GPUCompiler.alloca #714

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
c33f8bb
remove StaticArrays
vchuravy Jun 22, 2026
0b58f9b
[pocl] Use GPUCompiler.alloca for Scratchpad instead of MArray
vchuravy Jun 22, 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 Project.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,7 +36,7 @@ SparseArraysExt = "SparseArrays"
Adapt = "0.4, 1.0, 2.0, 3.0, 4"
Atomix = "0.1, 1"
EnzymeCore = "0.7, 0.8.1"
GPUCompiler = "1.13.3"
GPUCompiler = "1.23"
InteractiveUtils = "1.6"
LLVM = "9.4.1"
LinearAlgebra = "1.6"
Expand Down
1 change: 0 additions & 1 deletion src/KernelAbstractions.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,6 @@ import PrecompileTools
import Atomix: @atomic, @atomicswap, @atomicreplace

using MacroTools
using StaticArrays
using Adapt

"""
Expand Down
8 changes: 5 additions & 3 deletions src/pocl/backend.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,8 +7,6 @@ using ..POCL: device, clconvert, clfunction
import KernelAbstractions as KA
import KernelAbstractions.KernelIntrinsics as KI

import StaticArrays

import Adapt


Expand Down Expand Up @@ -222,7 +220,11 @@ end
end

@device_override @inline function KA.Scratchpad(ctx, ::Type{T}, ::Val{Dims}) where {T, Dims}
StaticArrays.MArray{KA.__size(Dims), T}(undef)
# private per-workitem scratch: a stack `alloca` (lowered by GPUCompiler) wrapped in a
# device array. the slot lives in OpenCL "Function" storage (LLVM addrspace 0), which is
# where the SPIR-V target places allocas.
ptr = POCL.GPUCompiler.alloca(T, Val(prod(Dims)))
CLDeviceArray(Dims, reinterpret(POCL.LLVMPtr{T, POCL.AS.Function}, ptr))
end


Expand Down
Loading