Skip to content

Define device-only functions as overlays, so sysimages load #751

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

Open
maleadt wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Define device-only functions as overlays, so sysimages load #751

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
6 changes: 2 additions & 4 deletions Project.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,6 @@ AbstractFFTs = "621f4979-c628-5d54-868e-fcf4e3e8185c"
Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
CEnum = "fa961155-64e5-5f13-b03f-caf6b980ea82"
Crayons = "a8cc5b0e-0ffa-5ad4-8c14-923d3ee1735f"
ExprTools = "e2ba6199-217a-4e67-a87a-7c52f15ade04"
GPUArrays = "0c68f7d7-f131-5f86-a1c3-88cf8149b2d7"
GPUCompiler = "61eb1bfa-7361-4325-ad38-22787b887f55"
GPUToolbox = "096a3bc2-3ced-46d0-87f4-dd12716f4bfc"
Expand Down Expand Up @@ -41,10 +40,9 @@ AbstractFFTs = "1.5.0"
Adapt = "4"
CEnum = "0.5, 1"
Crayons = "4"
ExprTools = "0.1"
GPUArrays = "12.1"
GPUCompiler = "2.18"
GPUToolbox = "3.4"
GPUToolbox = "3.5"
IntelITT = "0.2"
KernelAbstractions = "0.10"
KernelInterface = "0.4"
Expand All @@ -54,7 +52,7 @@ PTI_jll = "1.2"
PrecompileTools = "1"
PrettyTables = "3"
Preferences = "1"
SPIRVIntrinsics = "1.5.1"
SPIRVIntrinsics = "1.6"
SPIRV_LLVM_Backend_jll = "23"
SPIRV_LLVM_Translator_jll = "23"
SPIRV_Tools_jll = "2025.4.0"
Expand Down
10 changes: 5 additions & 5 deletions src/device/runtime.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@ KernelState(exception_report::UInt64=UInt64(0), exception_claim::UInt64=UInt64(0
KernelState(reinterpret(LLVMPtr{UInt8, AS.Function}, C_NULL),
exception_report, exception_claim)

@inline @generated kernel_state() = GPUCompiler.kernel_state_value(KernelState)
@device_function @inline @generated kernel_state() = GPUCompiler.kernel_state_value(KernelState)


## dynamic memory allocation
Expand All @@ -41,7 +41,7 @@ const HEAP_ALIGNMENT = 16
# Two 64-bit words keep the payload aligned to HEAP_ALIGNMENT.
const HEAP_HEADER = 2 * sizeof(Csize_t)

function malloc(sz::Csize_t)
@device_function function malloc(sz::Csize_t)
heap = kernel_state().heap
heap == reinterpret(LLVMPtr{UInt8, AS.Function}, C_NULL) && return C_NULL

Expand Down Expand Up @@ -157,7 +157,7 @@ end
# Claim the report, returning whether this work-item owns it. Idempotent, so that every
# reporting function can check ownership with it. The owner sets the report's status before
# exiting, which keeps work-items with the same id in later kernels from claiming it too.
@inline function claim_report!()
@device_function @inline function claim_report!()
claim = kernel_state().exception_claim
claim == 0 && return false
# A relaxed claim suffices: the host only reads the report once the kernel has completed.
Expand All @@ -179,7 +179,7 @@ string_address(str::CString) = reinterpret(UInt64, str)

# Record the exception's name and reason, unless one was recorded already: `@gputhrow` comes
# first, and knows a more specific name than GPUCompiler derives from the throw.
@inline function record_exception!(name::CString, reason::Union{CString, Nothing}=nothing)
@device_function @inline function record_exception!(name::CString, reason::Union{CString, Nothing}=nothing)
reporting_call(claim_report!) || return false
name_ptr = report_field(Val(:name))
if unsafe_load(name_ptr) == 0
Expand Down Expand Up @@ -213,7 +213,7 @@ function report_exception_frame(idx, func, file, line)
return
end

function signal_exception()
@device_function function signal_exception()
# At debug level 0 this is the only reporting function, and claims the report.
reporting_call(claim_report!) && unsafe_store!(report_field(Val(:status)), 1 % Int32)
return
Expand Down
2 changes: 0 additions & 2 deletions src/oneAPI.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,8 +6,6 @@ using Adapt
using GPUCompiler
import GPUToolbox

import ExprTools

using SpecialFunctions

import Preferences
Expand Down
1 change: 1 addition & 0 deletions test/Project.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,7 @@ Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
FFTW = "7a1cc6ca-52ef-59f5-83cd-3a7055c09341"
GPUArrays = "0c68f7d7-f131-5f86-a1c3-88cf8149b2d7"
GPUToolbox = "096a3bc2-3ced-46d0-87f4-dd12716f4bfc"
InteractiveUtils = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
JLD2 = "033835bb-8acc-5ee8-8aae-3f567f8a3819"
KernelAbstractions = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
Expand Down
6 changes: 6 additions & 0 deletions test/runtests.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,12 @@ for name in keys(TestSuite.tests)
testsuite["gpuarrays/$name"] = :(TestSuite.tests[$name](oneArray))
end

# sysimage builds compile methods for the host, which must not reach device-only code
testsuite["host leaks"] = quote
import GPUToolbox
@test isempty(GPUToolbox.Overlays.host_leaks(oneAPI))
end

args = parse_args(ARGS)

# Optional: spread test workers across all available GPUs, one worker per device
Expand Down
Loading