[ERROR] Using "make" command #443
Comments
The VirtualBox causes the problem [1], you can try to use gcc:10.2 with Docker. [1]. virtual box is the problem limiting the avx instructions. |
|
As the Makefile no longer sets specific instruction set options, but uses |
|
See issue comment for disable AVX2 : #1655 (comment) |
Deadsg
pushed a commit
to Deadsg/llama.cpp
that referenced
this issue
Dec 19, 2023
…terial-9.1.18 Bump mkdocs-material from 9.1.17 to 9.1.18
# for free
to join this conversation on GitHub.
Already have an account?
# to comment
Hello evryone,
I have an issue when i run "make" cmd :
I use Ubuntu 22.04 in VirtualBox
Make version : GNU Make 4.3
Here the return of cmd
I llama.cpp build info: I UNAME_S: Linux I UNAME_P: x86_64 I UNAME_M: x86_64 I CFLAGS: -I. -O3 -DNDEBUG -std=c11 -fPIC -pthread -mavx -mavx2 -msse3 I CXXFLAGS: -I. -I./examples -O3 -DNDEBUG -std=c++11 -fPIC -pthread I LDFLAGS: I CC: cc (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0 I CXX: g++ (Ubuntu 11.3.0-1ubuntu1~22.04) 11.3.0 cc -I. -O3 -DNDEBUG -std=c11 -fPIC -pthread -mavx -mavx2 -msse3 -c ggml.c -o ggml.o In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:101, from ggml.c:158: ggml.c: In function ‘ggml_vec_dot_f16’: /usr/lib/gcc/x86_64-linux-gnu/11/include/f16cintrin.h:52:1: error: inlining failed in call to ‘always_inline’ ‘_mm256_cvtph_ps’: target specific option mismatch 52 | _mm256_cvtph_ps (__m128i __A) | ^~~~~~~~~~~~~~~ ggml.c:936:33: note: called from here 936 | #define GGML_F32Cx8_LOAD(x) _mm256_cvtph_ps(_mm_loadu_si128((__m128i *)(x))) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ggml.c:946:37: note: in expansion of macro ‘GGML_F32Cx8_LOAD’ 946 | #define GGML_F16_VEC_LOAD(p, i) GGML_F32Cx8_LOAD(p) | ^~~~~~~~~~~~~~~~ ggml.c:1340:21: note: in expansion of macro ‘GGML_F16_VEC_LOAD’ 1340 | ay[j] = GGML_F16_VEC_LOAD(y + i + j*GGML_F16_EPR, j); | ^~~~~~~~~~~~~~~~~ In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:101, from ggml.c:158: /usr/lib/gcc/x86_64-linux-gnu/11/include/f16cintrin.h:52:1: error: inlining failed in call to ‘always_inline’ ‘_mm256_cvtph_ps’: target specific option mismatch 52 | _mm256_cvtph_ps (__m128i __A) | ^~~~~~~~~~~~~~~ ggml.c:936:33: note: called from here 936 | #define GGML_F32Cx8_LOAD(x) _mm256_cvtph_ps(_mm_loadu_si128((__m128i *)(x))) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ggml.c:946:37: note: in expansion of macro ‘GGML_F32Cx8_LOAD’ 946 | #define GGML_F16_VEC_LOAD(p, i) GGML_F32Cx8_LOAD(p) | ^~~~~~~~~~~~~~~~ ggml.c:1339:21: note: in expansion of macro ‘GGML_F16_VEC_LOAD’ 1339 | ax[j] = GGML_F16_VEC_LOAD(x + i + j*GGML_F16_EPR, j); | ^~~~~~~~~~~~~~~~~ In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:101, from ggml.c:158: /usr/lib/gcc/x86_64-linux-gnu/11/include/f16cintrin.h:52:1: error: inlining failed in call to ‘always_inline’ ‘_mm256_cvtph_ps’: target specific option mismatch 52 | _mm256_cvtph_ps (__m128i __A) | ^~~~~~~~~~~~~~~ ggml.c:936:33: note: called from here 936 | #define GGML_F32Cx8_LOAD(x) _mm256_cvtph_ps(_mm_loadu_si128((__m128i *)(x))) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ggml.c:946:37: note: in expansion of macro ‘GGML_F32Cx8_LOAD’ 946 | #define GGML_F16_VEC_LOAD(p, i) GGML_F32Cx8_LOAD(p) | ^~~~~~~~~~~~~~~~ ggml.c:1339:21: note: in expansion of macro ‘GGML_F16_VEC_LOAD’ 1339 | ax[j] = GGML_F16_VEC_LOAD(x + i + j*GGML_F16_EPR, j); | ^~~~~~~~~~~~~~~~~ In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:101, from ggml.c:158: /usr/lib/gcc/x86_64-linux-gnu/11/include/f16cintrin.h:52:1: error: inlining failed in call to ‘always_inline’ ‘_mm256_cvtph_ps’: target specific option mismatch 52 | _mm256_cvtph_ps (__m128i __A) | ^~~~~~~~~~~~~~~ ggml.c:936:33: note: called from here 936 | #define GGML_F32Cx8_LOAD(x) _mm256_cvtph_ps(_mm_loadu_si128((__m128i *)(x))) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ggml.c:946:37: note: in expansion of macro ‘GGML_F32Cx8_LOAD’ 946 | #define GGML_F16_VEC_LOAD(p, i) GGML_F32Cx8_LOAD(p) | ^~~~~~~~~~~~~~~~ ggml.c:1340:21: note: in expansion of macro ‘GGML_F16_VEC_LOAD’ 1340 | ay[j] = GGML_F16_VEC_LOAD(y + i + j*GGML_F16_EPR, j); | ^~~~~~~~~~~~~~~~~ In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:101, from ggml.c:158: /usr/lib/gcc/x86_64-linux-gnu/11/include/f16cintrin.h:52:1: error: inlining failed in call to ‘always_inline’ ‘_mm256_cvtph_ps’: target specific option mismatch 52 | _mm256_cvtph_ps (__m128i __A) | ^~~~~~~~~~~~~~~ ggml.c:936:33: note: called from here 936 | #define GGML_F32Cx8_LOAD(x) _mm256_cvtph_ps(_mm_loadu_si128((__m128i *)(x))) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ggml.c:946:37: note: in expansion of macro ‘GGML_F32Cx8_LOAD’ 946 | #define GGML_F16_VEC_LOAD(p, i) GGML_F32Cx8_LOAD(p) | ^~~~~~~~~~~~~~~~ ggml.c:1339:21: note: in expansion of macro ‘GGML_F16_VEC_LOAD’ 1339 | ax[j] = GGML_F16_VEC_LOAD(x + i + j*GGML_F16_EPR, j); | ^~~~~~~~~~~~~~~~~ In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:101, from ggml.c:158: /usr/lib/gcc/x86_64-linux-gnu/11/include/f16cintrin.h:52:1: error: inlining failed in call to ‘always_inline’ ‘_mm256_cvtph_ps’: target specific option mismatch 52 | _mm256_cvtph_ps (__m128i __A) | ^~~~~~~~~~~~~~~ ggml.c:936:33: note: called from here 936 | #define GGML_F32Cx8_LOAD(x) _mm256_cvtph_ps(_mm_loadu_si128((__m128i *)(x))) | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ggml.c:946:37: note: in expansion of macro ‘GGML_F32Cx8_LOAD’ 946 | #define GGML_F16_VEC_LOAD(p, i) GGML_F32Cx8_LOAD(p) | ^~~~~~~~~~~~~~~~ ggml.c:1340:21: note: in expansion of macro ‘GGML_F16_VEC_LOAD’ 1340 | ay[j] = GGML_F16_VEC_LOAD(y + i + j*GGML_F16_EPR, j); | ^~~~~~~~~~~~~~~~~ make: *** [Makefile:221 : ggml.o] Erreur 1The text was updated successfully, but these errors were encountered: