-
Notifications
You must be signed in to change notification settings - Fork 18
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.
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
top-level continuation 'name' got free def #80
Comments
sorry. I can't reproduce:
I'm using |
The hash was on the wrong branch, I've updated my previous post. |
I have the same (or similar) issue on master: With everything in Debug mode (thorin, impala, runtime, and LLVM), the assertion is triggered when compiling the CPU version of the traversal. The GPU version works (compiling and running). The behaviour in Release mode (thorin, impala, runtime, and LLVM) is different: Both versions compile, but the GPU version crashes during execution. To trigger this bug, use the master branch used at the time of this writing, which is: |
The issue goes away when using the lower2cff_pe branch of thorin, and compiling thorin in Debug mode.
|
This assertion is a known issue. Impala and Thorin have to be built both in Release mode or both in Debug mode. |
With both Thorin and Impala in Release, Impala seg-faults when compiling the traversal for old gpus, i.e. specifically when replacing the nvvm_ldg4_i32 in line 93 of mapping_gpu.impala. Replacing any combination of the nvvm_ldg4_f32 intrinsics works. |
We get this assertion when compiling the traversal for GPU (nvvm & amdgpu).
When commenting out unused code in the
intrinsics_amdgpu.impala
(amdgcn_atomic_* lines 81 - 93), the code compiles fornvvm
andamdgpu
asGPU_BACKEND
.Branches:
The text was updated successfully, but these errors were encountered: