Compile bug: make CC='/opt/AMD/aocc-compiler-5.0.0/bin/clang' CXX='/opt/AMD/aocc-compiler-5.0.0/bin/clang++' AMD_ZEN4_BLIS_5=1 GGML_CUDA=1 GGML_CUDA_FORCE_CUBLAS=1 GML_CUDA_F16=1 GGML_CUDA_FORCE_MMQ=1 CUDA_USE_TENSOR_CORES=1 GGML_RPC=1

[KarlHeinzMali]

Git commit

Build 4160

Which operating systems do you know to be affected?

Linux

GGML backends

BLAS, CPU, CUDA, RPC

Steps to Reproduce

Can't compile build 4160 with GGML_RPC=1

First Bad Commit

No response

Relevant log output

/opt/AMD/aocc-compiler-5.0.0/bin/clang -Iggml/include -Iggml/src -Iinclude -Isrc -Icommon -D_XOPEN_SOURCE=600 -D_GNU_SOURCE -DNDEBUG -DGGML_USE_CPU_AARCH64 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_USE_AMX -DGGML_USE_RPC -DGGML_USE_CUDA -DGGML_CUDA_USE_GRAPHS -I/usr/local/cuda/include -I/usr/local/cuda/targets/x86_64-linux/include  -std=c11   -fPIC -O3 -g -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes -Werror=implicit-int -Werror=implicit-function-declaration -pthread -march=native -mtune=native -fopenmp -Wunreachable-code -Wdouble-promotion  -MMD -c ggml/src/ggml-cpu/ggml-cpu-quants.c -o ggml/src/ggml-cpu/ggml-cpu-quants.o
/opt/AMD/aocc-compiler-5.0.0/bin/clang++ -std=c++11 -fPIC -O3 -g -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wmissing-declarations -Wmissing-noreturn -pthread -fopenmp  -march=native -mtune=native -Wunreachable-code -Wmissing-prototypes -Wextra-semi -Iggml/include -Iggml/src -Iinclude -Isrc -Icommon -D_XOPEN_SOURCE=600 -D_GNU_SOURCE -DNDEBUG -DGGML_USE_CPU_AARCH64 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_USE_AMX -DGGML_USE_RPC -DGGML_USE_CUDA -DGGML_CUDA_USE_GRAPHS -I/usr/local/cuda/include -I/usr/local/cuda/targets/x86_64-linux/include  -MMD -c ggml/src/ggml-cpu/llamafile/sgemm.cpp -o ggml/src/ggml-cpu/llamafile/sgemm.o
ggml/src/ggml-cpu/llamafile/sgemm.cpp:1871:11: warning: code will never be executed [-Wunreachable-code]
 1871 |     (void)m;
      |           ^
1 warning generated.
/opt/AMD/aocc-compiler-5.0.0/bin/clang++ -std=c++11 -fPIC -O3 -g -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wmissing-declarations -Wmissing-noreturn -pthread -fopenmp  -march=native -mtune=native -Wunreachable-code -Wmissing-prototypes -Wextra-semi -Iggml/include -Iggml/src -Iinclude -Isrc -Icommon -D_XOPEN_SOURCE=600 -D_GNU_SOURCE -DNDEBUG -DGGML_USE_CPU_AARCH64 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_USE_AMX -DGGML_USE_RPC -DGGML_USE_CUDA -DGGML_CUDA_USE_GRAPHS -I/usr/local/cuda/include -I/usr/local/cuda/targets/x86_64-linux/include  -MMD -c ggml/src/ggml-amx/ggml-amx.cpp -o ggml/src/ggml-amx/ggml-amx.o
/opt/AMD/aocc-compiler-5.0.0/bin/clang++ -std=c++11 -fPIC -O3 -g -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wmissing-declarations -Wmissing-noreturn -pthread -fopenmp  -march=native -mtune=native -Wunreachable-code -Wmissing-prototypes -Wextra-semi -Iggml/include -Iggml/src -Iinclude -Isrc -Icommon -D_XOPEN_SOURCE=600 -D_GNU_SOURCE -DNDEBUG -DGGML_USE_CPU_AARCH64 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_USE_AMX -DGGML_USE_RPC -DGGML_USE_CUDA -DGGML_CUDA_USE_GRAPHS -I/usr/local/cuda/include -I/usr/local/cuda/targets/x86_64-linux/include  -MMD -c ggml/src/ggml-amx/mmq.cpp -o ggml/src/ggml-amx/mmq.o
make: *** No rule to make target 'ggml/src/ggml-rpc.o', needed by 'libllava.a'.  Stop.

[slaren]

It should work with cmake.

[KarlHeinzMali]

Works with cmake, but performance is partial poor with same flags like make.