\> at C:/workdir/src\aotcompile.cpp:1647 [inlined]
> jl\_dump\_native\_impl at C:/workdir/src\aotcompile.cpp:1793
> ijl\_write\_compiler\_output at C:/workdir/src\precompile.c:177
> ijl\_atexit\_hook at C:/workdir/src\init.c:285
> jl\_repl\_entrypoint at C:/workdir/src\jlapi.c:1060
> mainCRTStartup at C:/workdir/cli\loader\_exe.c:58
> BaseThreadInitThunk at C:\WINDOWS\System32\KERNEL32.DLL (unknown line)
> RtlUserThreadStart at C:\WINDOWS\SYSTEM32\ntdll.dll (unknown line)
> Allocations: 1616528984 (Pool: 1616519201; Big: 9783); GC: 281
> ✖ [10m:10s] PackageCompiler: compiling incremental system image
> ERROR: failed process: Process(`'C:\Users\karlo\.julia\juliaup\julia-1.11.3+0.x64.w64.mingw32\bin\julia.exe' --color=yes --startup-file=no --pkgimages=no --cpu-target=native --compile=all '--sysimage=C:\Users\karlo\.julia\juliaup\julia-1.11.3+0.x64.w64.mingw32\lib\julia\sys.dll' '--project=C:\Users\karlo\.julia\environments\v1.11' '--output-o=C:\Users\karlo\AppData\Local\Temp\jl_KrgDWpWrgR-o.a' 'C:\Users\karlo\AppData\Local\Temp\jl_yhIfmFWHaX'`, ProcessExited(3)) [3]
>
> Stacktrace:
> [1] pipeline\_error
> @ .\process.jl:598 [inlined]
> [2] run(::Cmd; wait::Bool)
> @ Base .\process.jl:513
> [3] run
> @ .\process.jl:510 [inlined]
> [4] #20
> @ C:\Users\karlo.julia\packages\PackageCompiler\UbaS4\ext\TerminalSpinners.jl:157 [inlined]
> [5] spin(f::PackageCompiler.var"#20#22"{Cmd}, s::PackageCompiler.TerminalSpinners.Spinner{Base.TTY})
> @ PackageCompiler.TerminalSpinners C:\Users\karlo.julia\packages\PackageCompiler\UbaS4\ext\TerminalSpinners.jl:164
> [6] macro expansion
> @ C:\Users\karlo.julia\packages\PackageCompiler\UbaS4\ext\TerminalSpinners.jl:157 [inlined]
> [7] create\_sysimg\_object\_file(object\_file::String, packages::Vector{…}, packages\_sysimg::Set{…}; project::String, base\_sysimage::String, precompile\_execution\_file::Vector{…}, precompile\_statements\_file::Vector{…}, cpu\_target::String, script::Nothing, sysimage\_build\_args::Cmd, extra\_precompiles::String, incremental::Bool, import\_into\_main::Bool)
> @ PackageCompiler C:\Users\karlo.julia\packages\PackageCompiler\UbaS4\src\PackageCompiler.jl:130
> [8] create\_sysimg\_object\_file
> @ C:\Users\karlo.julia\packages\PackageCompiler\UbaS4\src\PackageCompiler.jl:319 [inlined]
> [9] create\_sysimage(packages::Vector{…}; sysimage\_path::String, project::String, precompile\_execution\_file::Vector{…}, precompile\_statements\_file::Vector{…}, incremental::Bool, filter\_stdlibs::Bool, cpu\_target::String, script::Nothing, sysimage\_build\_args::Cmd, include\_transitive\_dependencies::Bool, base\_sysimage::Nothing, julia\_init\_c\_file::Nothing, julia\_init\_h\_file::Nothing, version::Nothing, soname::Nothing, compat\_level::String, extra\_precompiles::String, import\_into\_main::Bool)
> @ PackageCompiler C:\Users\karlo.julia\packages\PackageCompiler\UbaS4\src\PackageCompiler.jl:639
> [10] top-level scope
> @ REPL[4]:1
> Some type information was truncated. Use `show(err)` to see complete types.
> julia\> versioninfo()
> Julia Version 1.11.3
> Commit d63adeda50 (2025-01-21 19:42 UTC)
> Build Info:
> Official [https://julialang.org/](https://julialang.org/) release
> Platform Info:
> OS: Windows (x86\_64-w64-mingw32)
> CPU: 32 × AMD Ryzen 9 5950X 16-Core Processor
> WORD\_SIZE: 64
> LLVM: libLLVM-16.0.6 (ORCJIT, znver3)
> Threads: 1 default, 0 interactive, 1 GC (on 32 virtual cores)
> julia\> CUDA.versioninfo()
> CUDA runtime 12.6, artifact installation
> CUDA driver 12.4
> NVIDIA driver 552.44.0
>
> CUDA libraries:
>
> - CUBLAS: 12.6.4
> - CURAND: 10.3.7
> - CUFFT: 11.3.0
> - CUSOLVER: 11.7.1
> - CUSPARSE: 12.5.4
> - CUPTI: 2024.3.2 (API 24.0.0)
> - NVML: 12.0.0+552.44
>
> Julia packages:
>
> - CUDA: 5.6.1
> - CUDA\_Driver\_jll: 0.10.4+0
> - CUDA\_Runtime\_jll: 0.15.5+0
>
> Toolchain:
>
> - Julia: 1.11.3
> - LLVM: 16.0.6
>
> 1 device:
> 0: NVIDIA GeForce RTX 2070 (sm\_75, 6.667 GiB / 8.000 GiB available)
---
### Author:  [@maleadt](https://discourse.julialang.org/u/maleadt)
#### Post date: [March 3, 2025, 7:37am UTC](https://discourse.julialang.org/t/packagecompiler-fails-to-create-app-for-madnlpgpu-examodels-cudss-linear-solver/125998/6 "2025-03-03T07:37:19Z")
This is likely [JIT session error triggered by precompilation · Issue #611 · JuliaGPU/GPUCompiler.jl · GitHub](https://github.com/JuliaGPU/GPUCompiler.jl/issues/611), or at least related to that (likely a Julia issue where GPU-only intrinsics leak into the native cache; cc @gbaraldi).
---
### Author:  [@KSepetanc](https://discourse.julialang.org/u/KSepetanc)
#### Post date: [May 26, 2025, 8:58pm UTC](https://discourse.julialang.org/t/packagecompiler-fails-to-create-app-for-madnlpgpu-examodels-cudss-linear-solver/125998/7 "2025-05-26T20:58:55Z")
Any way to revive this? Got no feedback from maintainers. This issue blocks compilation of anything CUDA.jl related (using either PackageCompiler or PrecompileTools).
---
### Author:  [@KSepetanc](https://discourse.julialang.org/u/KSepetanc)
#### Post date: [May 31, 2025, 10:26pm UTC](https://discourse.julialang.org/t/packagecompiler-fails-to-create-app-for-madnlpgpu-examodels-cudss-linear-solver/125998/8 "2025-05-31T22:26:19Z")
Not sure if meaningful in any way, but noticed that the following command also fails:
> ccall(“llvm.nvvm.cp.async.commit.group”, llvmcall, Cvoid, ())
> ERROR: LLVM error: Cannot select: intrinsic %llvm.nvvm.cp.async.commit.group
> Stacktrace:
> [1] handle\_error(reason::Cstring)
> @ LLVM ~/.julia/packages/LLVM/2JPxT/src/core/context.jl:194
Also, `llvm.nvvm.membar.sys` is now formally depreciated (again not sure if useful info).
---
### Author:  [@maleadt](https://discourse.julialang.org/u/maleadt)
#### Post date: [June 2, 2025, 1:25pm UTC](https://discourse.julialang.org/t/packagecompiler-fails-to-create-app-for-madnlpgpu-examodels-cudss-linear-solver/125998/9 "2025-06-02T13:25:47Z")
> [@KSepetanc](#):
>
> Not sure if meaningful in any way, but noticed that the following command also fails:
>
> > ccall(“llvm.nvvm.cp.async.commit.group”, llvmcall, Cvoid, ())
That’s expected; you’re calling a GPU-only intrinsic from a CPU context here. It also illustrates the issue at hand here: During static image generation (which is invoked by PackageCompiler.jl, but also by precompilation), GPU-only intrinsics “leak” into the CPU cache, despite only being accessible via a specific method overlay table that makes it impossible to directly call into GPU code from a CPU context. It’s likely a bug in Julia itself, e.g., losing the provenance of these functions and treating them as CPU-compatible.
---
### Author:  [@KSepetanc](https://discourse.julialang.org/u/KSepetanc)
#### Post date: [June 20, 2025, 9:24pm UTC](https://discourse.julialang.org/t/packagecompiler-fails-to-create-app-for-madnlpgpu-examodels-cudss-linear-solver/125998/10 "2025-06-20T21:24:26Z")
Could Julia v1.12 trim option help with “leak” issue?
It filters only proven reachable commands, so could it avoid the leak?
---
### Author:  [@KSepetanc](https://discourse.julialang.org/u/KSepetanc)
#### Post date: [June 28, 2025, 4:02pm UTC](https://discourse.julialang.org/t/packagecompiler-fails-to-create-app-for-madnlpgpu-examodels-cudss-linear-solver/125998/11 "2025-06-28T16:02:09Z")
I have more info on this. Besides compile=all that fails as described in reports above, I have tried default compile=yes.
On both Linux and Windows it completes without reported error for my gpu “app”. However, generated executable file works only with Windows workflow. When everything (create\_app command and executable) is run on Linux, create\_app finishes, but resulting executable when runs in loop quickly consuming all ram (and I have 64 GB). In the code there are no such loops that could do that and this is some compile error not registered during compiling. I don’t have much to copy here as there is no reported error. In Windows, executable runs as expected.
---
### Author:  [@maleadt](https://discourse.julialang.org/u/maleadt)
#### Post date: [June 30, 2025, 7:26am UTC](https://discourse.julialang.org/t/packagecompiler-fails-to-create-app-for-madnlpgpu-examodels-cudss-linear-solver/125998/12 "2025-06-30T07:26:00Z")
That could be [PackageCompiler CUDA app hangs at runtime · Issue #2696 · JuliaGPU/CUDA.jl · GitHub](https://github.com/JuliaGPU/CUDA.jl/issues/2696), try [PackageCompiler CUDA app hangs at runtime · Issue #2696 · JuliaGPU/CUDA.jl · GitHub](https://github.com/JuliaGPU/CUDA.jl/issues/2696#issuecomment-2984259471)