Implementation notes: amd64, hydra4, crypto_sign/luov863256pc

Computer: hydra4
Microarchitecture: amd64; K10 32nm (300f10)
Architecture: amd64
CPU ID: AuthenticAMD-00300f10-178bfbff
SUPERCOP version: 20240625
Operation: crypto_sign
Primitive: luov863256pc

Compiler output

LUOV.c: LUOV.c:38:19: error: '__builtin_ia32_permdi256' needs target feature avx2
LUOV.c:                         __m256i rrrr = _mm256_permute4x64_epi64(_mm256_loadu_si256((__m256i *)&Q1[col++]),0);
LUOV.c:                                        ^
LUOV.c: /usr/lib/llvm-11/lib/clang/11.0.1/include/avx2intrin.h:818:12: note: expanded from macro '_mm256_permute4x64_epi64'
LUOV.c:   (__m256i)__builtin_ia32_permdi256((__v4di)(__m256i)(V), (int)(M))
LUOV.c:            ^
LUOV.c: LUOV.c:38:44: error: always_inline function '_mm256_loadu_si256' requires target feature 'avx', but would be inlined into function 'calculateQ2' that is compiled without support for 'avx'
LUOV.c:                         __m256i rrrr = _mm256_permute4x64_epi64(_mm256_loadu_si256((__m256i *)&Q1[col++]),0);
LUOV.c:                                                                 ^
LUOV.c: LUOV.c:38:44: error: AVX vector return of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
LUOV.c: LUOV.c:43:69: error: always_inline function '_mm256_setzero_pd' requires target feature 'avx', but would be inlined into function 'calculateQ2' that is compiled without support for 'avx'
LUOV.c:                                 *((__m256i *)&TempMat[i][k*8+4]) ^=  (__m256i) _mm256_blendv_pd(_mm256_setzero_pd(),(__m256d) rrrr,(__m256d)TJ);
LUOV.c:                                                                                                 ^
LUOV.c: LUOV.c:43:69: error: AVX vector return of type '__m256d' (vector of 4 'double' values) without 'avx' enabled changes the ABI
LUOV.c: LUOV.c:43:52: error: always_inline function '_mm256_blendv_pd' requires target feature 'avx', but would be inlined into function 'calculateQ2' that is compiled without support for 'avx'
LUOV.c:                                 *((__m256i *)&TempMat[i][k*8+4]) ^=  (__m256i) _mm256_blendv_pd(_mm256_setzero_pd(),(__m256d) rrrr,(__m256d)TJ);
LUOV.c:                                                                                ^
LUOV.c: LUOV.c:43:52: error: AVX vector argument of type '__m256d' (vector of 4 'double' values) without 'avx' enabled changes the ABI
LUOV.c: LUOV.c:44:10: error: always_inline function '_mm256_slli_epi64' requires target feature 'avx2', but would be inlined into function 'calculateQ2' that is compiled without support for 'avx2'
LUOV.c:                                 TJ = _mm256_slli_epi64(TJ,4);
LUOV.c:                                      ^
LUOV.c: LUOV.c:44:10: error: AVX vector argument of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
LUOV.c: LUOV.c:46:66: error: always_inline function '_mm256_setzero_pd' requires target feature 'avx', but would be inlined into function 'calculateQ2' that is compiled without support for 'avx'
LUOV.c:                                 *((__m256i *)&TempMat[i][k*8]) ^= (__m256i) _mm256_blendv_pd(_mm256_setzero_pd(),(__m256d) rrrr,(__m256d)TJ);
LUOV.c:                                                                                              ^
LUOV.c: ...
Number of similar (implementation,compiler) pairs: 5, namely:
ImplementationCompiler
T:avx2clang -march=native -O2 -fwrapv -Qunused-arguments -fPIC -fPIE -gdwarf-4 -Wall (Debian_Clang_11.0.1)
T:avx2clang -march=native -O3 -fwrapv -Qunused-arguments -fPIC -fPIE -gdwarf-4 -Wall (Debian_Clang_11.0.1)
T:avx2clang -march=native -O -fwrapv -Qunused-arguments -fPIC -fPIE -gdwarf-4 -Wall (Debian_Clang_11.0.1)
T:avx2clang -march=native -Os -fwrapv -Qunused-arguments -fPIC -fPIE -gdwarf-4 -Wall (Debian_Clang_11.0.1)
T:avx2clang -mcpu=native -O3 -fwrapv -Qunused-arguments -fPIC -fPIE -gdwarf-4 -Wall (Debian_Clang_11.0.1)

Compiler output

LUOV.c: LUOV.c: In function 'calculateQ2':
LUOV.c: LUOV.c:38:12: warning: AVX vector return without AVX enabled changes the ABI [-Wpsabi]
LUOV.c:    38 |    __m256i rrrr = _mm256_permute4x64_epi64(_mm256_loadu_si256((__m256i *)&Q1[col++]),0);
LUOV.c:       |            ^~~~
LUOV.c: In file included from LinearAlgebra.h:9,
LUOV.c:                  from LUOV.h:14,
LUOV.c:                  from LUOV.c:1:
LUOV.c: AVX_Operations.h: In function 'scalarMul_ct':
LUOV.c: AVX_Operations.h:529:6: note: the ABI for passing parameters with 32-byte alignment has changed in GCC 4.6
LUOV.c:   529 | void scalarMul_ct(__m256i *Out, __m256i A, FELT b){
LUOV.c:       |      ^~~~~~~~~~~~
LUOV.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/10/include/immintrin.h:53,
LUOV.c:                  from LUOV.h:8,
LUOV.c:                  from LUOV.c:1:
LUOV.c: AVX_Operations.h: In function 'addScalarProductAVX':
LUOV.c: /usr/lib/gcc/x86_64-linux-gnu/10/include/avx2intrin.h:186:1: error: inlining failed in call to 'always_inline' '_mm256_andnot_si256': target specific option mismatch
LUOV.c:   186 | _mm256_andnot_si256 (__m256i __A, __m256i __B)
LUOV.c:       | ^~~~~~~~~~~~~~~~~~~
LUOV.c: In file included from LinearAlgebra.h:9,
LUOV.c:                  from LUOV.h:14,
LUOV.c:                  from LUOV.c:1:
LUOV.c: AVX_Operations.h:73:9: note: called from here
LUOV.c:    73 |  avx2 = _mm256_andnot_si256(avx2,aa);
LUOV.c:       |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
LUOV.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/10/include/immintrin.h:53,
LUOV.c: ...
Number of similar (implementation,compiler) pairs: 4, namely:
ImplementationCompiler
T:avx2gcc -march=native -mtune=native -O2 -fwrapv -fPIC -fPIE -gdwarf-4 -Wall (10.2.1_20210110)
T:avx2gcc -march=native -mtune=native -O3 -fwrapv -fPIC -fPIE -gdwarf-4 -Wall (10.2.1_20210110)
T:avx2gcc -march=native -mtune=native -O -fwrapv -fPIC -fPIE -gdwarf-4 -Wall (10.2.1_20210110)
T:avx2gcc -march=native -mtune=native -Os -fwrapv -fPIC -fPIE -gdwarf-4 -Wall (10.2.1_20210110)