Compilation Failure: Incompatible Use of AVX-512 Intrinsics in ggml-aarch64.c

[WanliZhong]

CPU Infomation:

Architecture:        x86_64
CPU op-mode(s):      32-bit, 64-bit
Byte Order:          Little Endian
CPU(s):              56
On-line CPU(s) list: 0-55
Thread(s) per core:  2
Core(s) per socket:  14
Socket(s):           2
NUMA node(s):        2
Vendor ID:           GenuineIntel
CPU family:          6
Model:               85
Model name:          Intel(R) Xeon(R) Gold 6132 CPU @ 2.60GHz
Stepping:            4
CPU MHz:             999.647
CPU max MHz:         3700.0000
CPU min MHz:         1000.0000
BogoMIPS:            5200.00
Virtualization:      VT-x
L1d cache:           32K
L1i cache:           32K
L2 cache:            1024K
L3 cache:            19712K
NUMA node0 CPU(s):   0-13,28-41
NUMA node1 CPU(s):   14-27,42-55
Flags:               fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single pti intel_ppin ssbd mba ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req pku ospke md_clear flush_l1d arch_capabilities

Build options:

cmake 
-DCMAKE_BUILD_TYPE=Debug
-DGGML_CUDA_F16:BOOL=ON
-DGGML_CUDA:BOOL=ON
-DGGML_CUDA_FA_ALL_QUANTS:BOOL=ON
-DCUDAToolkit_ROOT=/home/wanli/cuda-12.1
-DCUDAToolkit_LIBRARY_DIR=/home/wanli/cuda-12.1/lib64
-DCMAKE_CUDA_COMPILER=/home/wanli/cuda-12.1/bin/nvcc
-DGGML_CUDA_FORCE_CUBLAS=1
-DGGML_CUDA_FORCE_MMQ=1

Error as below:

====================[ Build | test-backend-ops | 8GPU-Debug ]===================
/home/wanli/software/cmake-3.29.9-linux-x86_64/bin/cmake --build /home/wanli/workspace/llama.cpp/cmake-build-8gpu-debug --target test-backend-ops -j 8
[  0%] Built target build_info
[  0%] Building C object ggml/src/CMakeFiles/ggml.dir/ggml-aarch64.c.o
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c: In function ‘ggml_gemm_q4_0_8x8_q8_0’:
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:47:54: error: implicit declaration of function ‘_mm256_set_m128i’; did you mean ‘_mm256_set_epi8’? [-Werror=implicit-function-declaration]
 #define GGML_F32Cx8x2_LOAD(x, y)     _mm512_cvtph_ps(_mm256_set_m128i(_mm_loadu_si128((const __m128i *)(y)), _mm_loadu_si128((const __m128i *)(x))))
                                                      ^
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:2607:50: note: in expansion of macro ‘GGML_F32Cx8x2_LOAD’
                     const __m512 col_scale_f32 = GGML_F32Cx8x2_LOAD(b_ptr_0[b].d, b_ptr_1[b].d);
                                                  ^~~~~~~~~~~~~~~~~~
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:47:54: error: incompatible type for argument 1 of ‘_mm512_cvtph_ps’
 #define GGML_F32Cx8x2_LOAD(x, y)     _mm512_cvtph_ps(_mm256_set_m128i(_mm_loadu_si128((const __m128i *)(y)), _mm_loadu_si128((const __m128i *)(x))))
                                                      ^
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:2607:50: note: in expansion of macro ‘GGML_F32Cx8x2_LOAD’
                     const __m512 col_scale_f32 = GGML_F32Cx8x2_LOAD(b_ptr_0[b].d, b_ptr_1[b].d);
                                                  ^~~~~~~~~~~~~~~~~~
In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:45:0,
                 from /home/wanli/workspace/llama.cpp/ggml/src/ggml-cpu-impl.h:426,
                 from /home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:11:
/usr/lib/gcc/x86_64-linux-gnu/7/include/avx512fintrin.h:12514:1: note: expected ‘__m256i {aka __vector(4) long long int}’ but argument is of type ‘int’
 _mm512_cvtph_ps (__m256i __A)
 ^~~~~~~~~~~~~~~
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:48:54: error: incompatible type for argument 1 of ‘_mm512_cvtph_ps’
 #define GGML_F32Cx16_REPEAT_LOAD(x)  _mm512_cvtph_ps(_mm256_set_m128i(x, x))
                                                      ^
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:2698:54: note: in expansion of macro ‘GGML_F32Cx16_REPEAT_LOAD’
                         const __m512 row_scale_f32 = GGML_F32Cx16_REPEAT_LOAD(row_scale_f16);
                                                      ^~~~~~~~~~~~~~~~~~~~~~~~
In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:45:0,
                 from /home/wanli/workspace/llama.cpp/ggml/src/ggml-cpu-impl.h:426,
                 from /home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:11:
/usr/lib/gcc/x86_64-linux-gnu/7/include/avx512fintrin.h:12514:1: note: expected ‘__m256i {aka __vector(4) long long int}’ but argument is of type ‘int’
 _mm512_cvtph_ps (__m256i __A)
 ^~~~~~~~~~~~~~~
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:47:54: error: incompatible type for argument 1 of ‘_mm512_cvtph_ps’
 #define GGML_F32Cx8x2_LOAD(x, y)     _mm512_cvtph_ps(_mm256_set_m128i(_mm_loadu_si128((const __m128i *)(y)), _mm_loadu_si128((const __m128i *)(x))))
                                                      ^
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:2801:50: note: in expansion of macro ‘GGML_F32Cx8x2_LOAD’
                     const __m512 col_scale_f32 = GGML_F32Cx8x2_LOAD(b_ptr_0[b].d, b_ptr_1[b].d);
                                                  ^~~~~~~~~~~~~~~~~~
In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:45:0,
                 from /home/wanli/workspace/llama.cpp/ggml/src/ggml-cpu-impl.h:426,
                 from /home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:11:
/usr/lib/gcc/x86_64-linux-gnu/7/include/avx512fintrin.h:12514:1: note: expected ‘__m256i {aka __vector(4) long long int}’ but argument is of type ‘int’
 _mm512_cvtph_ps (__m256i __A)
 ^~~~~~~~~~~~~~~
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:48:54: error: incompatible type for argument 1 of ‘_mm512_cvtph_ps’
 #define GGML_F32Cx16_REPEAT_LOAD(x)  _mm512_cvtph_ps(_mm256_set_m128i(x, x))
                                                      ^
/home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:2889:50: note: in expansion of macro ‘GGML_F32Cx16_REPEAT_LOAD’
                     const __m512 row_scale_f32 = GGML_F32Cx16_REPEAT_LOAD(row_scale_f16);
                                                  ^~~~~~~~~~~~~~~~~~~~~~~~
In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:45:0,
                 from /home/wanli/workspace/llama.cpp/ggml/src/ggml-cpu-impl.h:426,
                 from /home/wanli/workspace/llama.cpp/ggml/src/ggml-aarch64.c:11:
/usr/lib/gcc/x86_64-linux-gnu/7/include/avx512fintrin.h:12514:1: note: expected ‘__m256i {aka __vector(4) long long int}’ but argument is of type ‘int’
 _mm512_cvtph_ps (__m256i __A)
 ^~~~~~~~~~~~~~~
cc1: some warnings being treated as errors
make[3]: *** [ggml/src/CMakeFiles/ggml.dir/build.make:2350: ggml/src/CMakeFiles/ggml.dir/ggml-aarch64.c.o] Error 1
make[2]: *** [CMakeFiles/Makefile2:1591: ggml/src/CMakeFiles/ggml.dir/all] Error 2
make[1]: *** [CMakeFiles/Makefile2:2113: tests/CMakeFiles/test-backend-ops.dir/rule] Error 2
make: *** [Makefile:777: test-backend-ops] Error 2

[slaren]

Try using a newer version of gcc.

[WanliZhong]

It work, Thanks