[build] Compilation error on build: k_quants.h:26:15: error: expected declaration specifiers or '...' before 'sizeof'

[ya-ds-web]

Prerequisites

Check out the master branch. Run: make LLAMA_CUBLAS=1

Expected Behavior

build succeed

Current Behavior

getting compilation error like:

I llama.cpp build info: I UNAME_S: Linux I UNAME_P: unknown I UNAME_M: x86_64 I CFLAGS: -I. -O3 -std=c11 -fPIC -DNDEBUG -Wall -Wextra -Wpedantic -Wcast-qual -Wdouble-promotion -Wshadow -Wstrict-prototypes -Wpointer-arith -pthread -march=native -mtune=native -DGGML_USE_K_QUANTS -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I/targets/x86_64-linux/include I CXXFLAGS: -I. -I./examples -O3 -std=c++11 -fPIC -DNDEBUG -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wno-multichar -pthread -march=native -mtune=native -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I/targets/x86_64-linux/include I LDFLAGS: -lcublas -lculibos -lcudart -lcublasLt -lpthread -ldl -lrt -L/usr/local/cuda/lib64 -L/opt/cuda/lib64 -L/targets/x86_64-linux/lib I CC: cc (conda-forge gcc 12.2.0-19) 12.2.0 I CXX: g++ (conda-forge gcc 12.2.0-19) 12.2.0

cc -I. -O3 -std=c11 -fPIC -DNDEBUG -Wall -Wextra -Wpedantic -Wcast-qual -Wdouble-promotion -Wshadow -Wstrict-prototypes -Wpointer-arith -pthread -march=native -mtune=native -DGGML_USE_K_QUANTS -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I/targets/x86_64-linux/include -c ggml.c -o ggml.o In file included from ggml.c:7: k_quants.h:26:15: error: expected declaration specifiers or '...' before 'sizeof' 26 | static_assert(sizeof(block_q2_K) == 2sizeof(ggml_fp16_t) + QK_K/16 + QK_K/4, "wrong q2_K block size/padding"); | ^~~~~~ k_quants.h:26:79: error: expected declaration specifiers or '...' before string constant 26 | static_assert(sizeof(block_q2_K) == 2sizeof(ggml_fp16_t) + QK_K/16 + QK_K/4, "wrong q2_K block size/padding"); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ k_quants.h:38:15: error: expected declaration specifiers or '...' before 'sizeof' 38 | static_assert(sizeof(block_q3_K) == sizeof(ggml_fp16_t) + QK_K / 4 + 11 * QK_K / 64, "wrong q3_K block size/padding"); | ^~~~~~ k_quants.h:38:86: error: expected declaration specifiers or '...' before string constant 38 | static_assert(sizeof(block_q3_K) == sizeof(ggml_fp16_t) + QK_K / 4 + 11 * QK_K / 64, "wrong q3_K block size/padding"); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ k_quants.h:50:15: error: expected declaration specifiers or '...' before 'sizeof' 50 | static_assert(sizeof(block_q4_K) == 2sizeof(ggml_fp16_t) + 3QK_K/64 + QK_K/2, "wrong q4_K block size/padding"); | ^~~~~~ k_quants.h:50:81: error: expected declaration specifiers or '...' before string constant 50 | static_assert(sizeof(block_q4_K) == 2sizeof(ggml_fp16_t) + 3QK_K/64 + QK_K/2, "wrong q4_K block size/padding"); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ k_quants.h:63:15: error: expected declaration specifiers or '...' before 'sizeof' 63 | static_assert(sizeof(block_q5_K) == 2sizeof(ggml_fp16_t) + 3QK_K/64 + QK_K/2 + QK_K/8, "wrong q5_K block size/padding"); | ^~~~~~ k_quants.h:63:90: error: expected declaration specifiers or '...' before string constant 63 | static_assert(sizeof(block_q5_K) == 2sizeof(ggml_fp16_t) + 3QK_K/64 + QK_K/2 + QK_K/8, "wrong q5_K block size/padding"); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ k_quants.h:75:15: error: expected declaration specifiers or '...' before 'sizeof' 75 | static_assert(sizeof(block_q6_K) == sizeof(ggml_fp16_t) + QK_K / 16 + 3QK_K/4, "wrong q6_K block size/padding"); | ^~~~~~ k_quants.h:75:81: error: expected declaration specifiers or '...' before string constant 75 | static_assert(sizeof(block_q6_K) == sizeof(ggml_fp16_t) + QK_K / 16 + 3QK_K/4, "wrong q6_K block size/padding"); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ k_quants.h:83:15: error: expected declaration specifiers or '...' before 'sizeof' 83 | static_assert(sizeof(block_q8_K) == sizeof(float) + QK_K + QK_K/16sizeof(int16_t), "wrong q8_K block size/padding"); | ^~~~~~ k_quants.h:83:85: error: expected declaration specifiers or '...' before string constant 83 | static_assert(sizeof(block_q8_K) == sizeof(float) + QK_K + QK_K/16sizeof(int16_t), "wrong q8_K block size/padding"); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ make: *** [Makefile:249: ggml.o] Error 1

Environment and Context

I am using fedora 38 (Linux 6.2.15-300.fc38.x86_64) inside an kvm vm gcc and gxx versions can be seen from the otput above I CC: cc (conda-forge gcc 12.2.0-19) 12.2.0 I CXX: g++ (conda-forge gcc 12.2.0-19) 12.2.0

[marcom]

Replacing static_assert with _Static_assert in k_quants.h fixed the compile for me on x86_64-linux-gnu using gcc-8.1.0.

[ya-ds-web]

thank you for the work around, I wonder how can I idetify the correct c compiler I can use to build it wthould the modification ?

[marcom]

According to https://github.com/ggerganov/llama.cpp/actions/runs/5304488298/jobs/9600839945#step:4:14 it compiles with gcc-8.4.0 and g++-9.4.0 on ubuntu 20.04 (Focal Fossa). You can see the successful CI builds by clicking on the little 'CI' badge at the top of the README in github.

[Sciumo]

Replacing static_assert with _Static_assert in k_quants.h

Confirm that this hack is still required as of 7/5/23 31cfbb1

[github-actions[bot]]

This issue was closed because it has been inactive for 14 days since being marked as stale.