You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
did anyone get this to work on AMD GPUs on Linux? I tried various things, including the hipconvonvert scripts, but there seems to be some mismatch between headers and the compilation runs into errors like (the actual output is a lot longer):
/mnt/ssd/Segment-and-Track-Anything/Pytorch-Correlation-extension/Correlation_Module/correlation_hip_kernel.hip:4:17: error: expected namespace name
using namespace torch;
^
/mnt/ssd/Segment-and-Track-Anything/Pytorch-Correlation-extension/Correlation_Module/correlation_hip_kernel.hip:25:11: error: use of undeclared identifier 'RestrictPtrTraits'
const TensorAcc4R rInput1,
^
/mnt/ssd/Segment-and-Track-Anything/Pytorch-Correlation-extension/Correlation_Module/correlation_hip_kernel.hip:14:55: note: expanded from macro 'TensorAcc4R'
#define TensorAcc4R PackedTensorAccessor32<scalar_t,4,RestrictPtrTraits>
^
/mnt/ssd/Segment-and-Track-Anything/Pytorch-Correlation-extension/Correlation_Module/correlation_hip_kernel.hip:26:11: error: use of undeclared identifier 'RestrictPtrTraits'
const TensorAcc4R rInput2,
^
/mnt/ssd/Segment-and-Track-Anything/Pytorch-Correlation-extension/Correlation_Module/correlation_hip_kernel.hip:14:55: note: expanded from macro 'TensorAcc4R'
#define TensorAcc4R PackedTensorAccessor32<scalar_t,4,RestrictPtrTraits>
^
/mnt/ssd/Segment-and-Track-Anything/Pytorch-Correlation-extension/Correlation_Module/correlation_hip_kernel.hip:27:5: error: use of undeclared identifier 'RestrictPtrTraits'
TensorAcc5R output,
^
This is on torch 2.0.1+rocm5.4.2.
The text was updated successfully, but these errors were encountered:
Hi,
did anyone get this to work on AMD GPUs on Linux? I tried various things, including the hipconvonvert scripts, but there seems to be some mismatch between headers and the compilation runs into errors like (the actual output is a lot longer):
This is on
torch 2.0.1+rocm5.4.2
.The text was updated successfully, but these errors were encountered: