Implementation notes: amd64, h3neo, crypto_encode/653xfreeze3

Computer: h3neo
Microarchitecture: amd64; K10 45nm (100f63)
Architecture: amd64
CPU ID: AuthenticAMD-00100f63-078bfbff
SUPERCOP version: 20231107
Operation: crypto_encode
Primitive: 653xfreeze3

Time	Object size	Test size	Implementation	Compiler	Benchmark date	SUPERCOP version
2065	1905 0 0	14773 804 960	`ref`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231108	20231107
2502	503 0 0	12499 844 896	`ref`	`clang_-mcpu=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230624	20230530
2506	343 0 0	12179 844 896	`ref`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230624	20230530
2508	343 0 0	11259 844 896	`ref`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230624	20230530
2653	96 0 0	9533 836 896	`ref`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230624	20230530
2685	105 0 0	10179 844 896	`ref`	`clang_-march=native_-O_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230624	20230530
3305	138 0 0	11677 804 960	`ref`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231108	20231107
3522	131 0 0	9936 780 928	`ref`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231108	20231107
3956	140 0 0	10900 796 960	`ref`	`gcc_-march=native_-mtune=native_-O_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231108	20231107

Compiler output

Implementation: avx
Security model: constbranchindex
Compiler: clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

encode.c: encode.c:15:19: error: always_inline function '_mm256_loadu_si256' requires target feature 'avx', but would be inlined into function 'crypto_encode_653xfreeze3_avx_constbranchindex' that is compiled without support for 'avx'
encode.c: __m256i x = _mm256_loadu_si256((__m256i *) r);
encode.c: ^
encode.c: encode.c:15:19: error: AVX vector return of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
encode.c: encode.c:16:41: error: always_inline function '_mm256_set1_epi16' requires target feature 'avx', but would be inlined into function 'crypto_encode_653xfreeze3_avx_constbranchindex' that is compiled without support for 'avx'
encode.c: __m256i y = _mm256_mulhrs_epi16(x,_mm256_set1_epi16(10923));
encode.c: ^
encode.c: encode.c:16:41: error: AVX vector return of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
encode.c: encode.c:16:19: error: always_inline function '_mm256_mulhrs_epi16' requires target feature 'avx2', but would be inlined into function 'crypto_encode_653xfreeze3_avx_constbranchindex' that is compiled without support for 'avx2'
encode.c: __m256i y = _mm256_mulhrs_epi16(x,_mm256_set1_epi16(10923));
encode.c: ^
encode.c: encode.c:16:19: error: AVX vector argument of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
encode.c: encode.c:17:11: error: always_inline function '_mm256_sub_epi16' requires target feature 'avx2', but would be inlined into function 'crypto_encode_653xfreeze3_avx_constbranchindex' that is compiled without support for 'avx2'
encode.c: x = _mm256_sub_epi16(x,y);
encode.c: ^
encode.c: encode.c:17:11: error: AVX vector argument of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
encode.c: encode.c:18:11: error: always_inline function '_mm256_add_epi16' requires target feature 'avx2', but would be inlined into function 'crypto_encode_653xfreeze3_avx_constbranchindex' that is compiled without support for 'avx2'
encode.c: y = _mm256_add_epi16(y,y);
encode.c: ^
encode.c: encode.c:18:11: error: AVX vector argument of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
encode.c: encode.c:19:11: error: always_inline function '_mm256_sub_epi16' requires target feature 'avx2', but would be inlined into function 'crypto_encode_653xfreeze3_avx_constbranchindex' that is compiled without support for 'avx2'
encode.c: x = _mm256_sub_epi16(x,y);
encode.c: ^
encode.c: encode.c:19:11: error: AVX vector argument of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
encode.c: encode.c:20:20: error: '__builtin_ia32_vextractf128_si256' needs target feature avx
encode.c: ...

Number of similar (compiler,implementation) pairs: 5, namely:

Compiler	Implementations
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx
clang -march=native -O -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx
clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx

Compiler output

Implementation: avx
Security model: constbranchindex
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE

encode.c: encode.c: In function 'crypto_encode_653xfreeze3_avx_constbranchindex':
encode.c: encode.c:15:15: warning: AVX vector return without AVX enabled changes the ABI [-Wpsabi]
encode.c: 15 | __m256i x = _mm256_loadu_si256((__m256i *) r);
encode.c: | ^
encode.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:43,
encode.c: from encode.c:1:
encode.c: /usr/lib/gcc/x86_64-linux-gnu/11/include/avxintrin.h:530:1: error: inlining failed in call to 'always_inline' '_mm256_extractf128_si256': target specific option mismatch
encode.c: 530 | _mm256_extractf128_si256 (__m256i __X, const int __N)
encode.c: | ^~~~~~~~~~~~~~~~~~~~~~~~
encode.c: encode.c:21:20: note: called from here
encode.c: 21 | __m128i x1 = _mm256_extractf128_si256(x,1);
encode.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
encode.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:43,
encode.c: from encode.c:1:
encode.c: /usr/lib/gcc/x86_64-linux-gnu/11/include/avxintrin.h:530:1: error: inlining failed in call to 'always_inline' '_mm256_extractf128_si256': target specific option mismatch
encode.c: 530 | _mm256_extractf128_si256 (__m256i __X, const int __N)
encode.c: | ^~~~~~~~~~~~~~~~~~~~~~~~
encode.c: encode.c:20:20: note: called from here
encode.c: 20 | __m128i x0 = _mm256_extractf128_si256(x,0);
encode.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
encode.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:47,
encode.c: from encode.c:1:
encode.c: /usr/lib/gcc/x86_64-linux-gnu/11/include/avx2intrin.h:808:1: error: inlining failed in call to 'always_inline' '_mm256_sub_epi16': target specific option mismatch
encode.c: 808 | _mm256_sub_epi16 (__m256i __A, __m256i __B)
encode.c: | ^~~~~~~~~~~~~~~~
encode.c: ...

Number of similar (compiler,implementation) pairs: 4, namely:

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE	avx
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv -fPIC -fPIE	avx
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv -fPIC -fPIE	avx
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv -fPIC -fPIE	avx