mpanoop

icon indicating an email [email protected]

icon company Intel Corporation icon location Hillsboro, OR icon location SW Enabling and Optimization Engineering Manager at Intel. Primary focus on Compilers (for CPU and GPU) and GPU compute stack.

Results 4 comments of mpanoop

LLVM with AMD backend fails to link binary object

Looks like the original command is trying to build the final executable which includes native device code for gfx900. Initial set of clang-offload-bundler is unbundling the host code and feeding...

LLVM with AMD backend fails to link binary object

@jgtong , could you please post the output of the following command for reference here: `/home/jaytong/sycl_workspace/llvm/build/bin/clang++ -### -O3 -fsycl -fsycl-targets=amdgcn-amd-amdhsa -Xsycl-target-backend --offload-arch=gfx908 -Wall -Wno-unknown-pragmas -Wextra -stdlib=libstdc++ -fcolor-diagnostics -Qunused-arguments -O2 -g...

LLVM with AMD backend fails to link binary object

Thanks @jgtong. Looks like the clang-offload-bundler in the malformed llvm-foreach command (which is failing) should be invoked outside of llvm-foreach command considering the sequence of commands which follows after this...

SYCLomatic README missing a pre-requisite and a Useful link

Building SYCLomatic from source doesn't need CUDA headers but using SYCLomatic to migrate the CUDA application to SYCL needs CUDA headers. Can we call this out explicitly in README?