Implementation notes: amd64, gemini, crypto_kem/bikel3

Computer: gemini
Microarchitecture: amd64; Goldmont Plus (706a8)
Architecture: amd64
CPU ID: GenuineIntel-000706a8-bfebfbff
SUPERCOP version: 20240107
Operation: crypto_kem
Primitive: bikel3

Time	Object size	Test size	Implementation	Compiler	Benchmark date	SUPERCOP version
26595572	32807 56 4	53412 884 1724	`T:aes-ni-and-pclmul`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
36185431	22315 56 4	40660 884 1724	`T:aes-ni-and-pclmul`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
37852757	78542 56 4	98032 852 1756	`T:aes-ni-and-pclmul`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
45120422	15378 56 4	31722 876 1724	`T:aes-ni-and-pclmul`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
51766237	43618 56 4	61304 852 1756	`T:aes-ni-and-pclmul`	`gcc_-march=native_-mtune=native_-O_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
51828400	16627 56 4	34138 876 1724	`T:aes-ni-and-pclmul`	`clang_-march=native_-O_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
52078815	45195 56 4	63464 852 1756	`T:aes-ni-and-pclmul`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
60433817	29375 56 4	45648 844 1724	`T:aes-ni-and-pclmul`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
61942614	32906 48 4	52778 932 1724	`T:portable`	`clang_-mcpu=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
62447951	29181 48 4	50474 932 1724	`T:portable`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
62503935	32385 56 4	53004 884 1724	`T:aes-ni-only`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
69147191	104178 56 4	123704 852 1756	`T:aes-ni-only`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
72060320	19446 48 4	38394 932 1724	`T:portable`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
76094635	100153 48 4	120211 916 1756	`T:portable`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
77484595	21893 56 4	40188 884 1724	`T:aes-ni-only`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
92595538	14354 56 4	30722 876 1724	`T:aes-ni-only`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
97104960	12129 48 4	29136 924 1724	`T:portable`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
108025895	41608 48 4	60483 916 1756	`T:portable`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
108285638	44359 56 4	62624 852 1756	`T:aes-ni-only`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
109307044	16221 56 4	33738 876 1724	`T:aes-ni-only`	`clang_-march=native_-O_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
109908959	40296 48 4	58659 916 1756	`T:portable`	`gcc_-march=native_-mtune=native_-O_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
111730703	29058 56 4	45288 844 1724	`T:aes-ni-only`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
113058305	26851 48 4	43715 908 1724	`T:portable`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212
117453212	14105 48 4	32264 924 1724	`T:portable`	`clang_-march=native_-O_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240102	20231212
119691533	42519 56 4	60208 852 1756	`T:aes-ni-only`	`gcc_-march=native_-mtune=native_-O_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20240102	20231212

Compiler output

Implementation: T:aes-ni-and-pclmul
Security model: timingleaks
Compiler: clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

aes.c: aes.c:9:4: error: "This code requries support for AES_NI and SSSE3"
aes.c: # error "This code requries support for AES_NI and SSSE3"
aes.c: ^
aes.c: 1 error generated.

Number of similar (compiler,implementation) pairs: 6, namely:

Compiler	Implementations
clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:aes-ni-and-pclmul T:aes-ni-only T:avx2 T:avx512 T:avx512-vpclmul T:ches2021

Compiler output

Implementation: T:avx2
Security model: timingleaks
Compiler: clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

decode.c: In file included from decode.c:39:
decode.c: ./gf2x.h:22:10: error: always_inline function '_mm256_loadu_si256' requires target feature 'avx', but would be inlined into function 'gf2x_mod_add' that is compiled without support for 'avx'
decode.c: va = LOAD(&a_qwords[i]);
decode.c: ^
decode.c: ./x86_64_intrinsic.h:30:27: note: expanded from macro 'LOAD'
decode.c: # define LOAD(mem) _mm256_loadu_si256((const void *)(mem))
decode.c: ^
decode.c: In file included from decode.c:39:
decode.c: ./gf2x.h:22:10: error: AVX vector return of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
decode.c: ./x86_64_intrinsic.h:30:27: note: expanded from macro 'LOAD'
decode.c: # define LOAD(mem) _mm256_loadu_si256((const void *)(mem))
decode.c: ^
decode.c: In file included from decode.c:39:
decode.c: ./gf2x.h:23:10: error: always_inline function '_mm256_loadu_si256' requires target feature 'avx', but would be inlined into function 'gf2x_mod_add' that is compiled without support for 'avx'
decode.c: vb = LOAD(&b_qwords[i]);
decode.c: ^
decode.c: ./x86_64_intrinsic.h:30:27: note: expanded from macro 'LOAD'
decode.c: # define LOAD(mem) _mm256_loadu_si256((const void *)(mem))
decode.c: ^
decode.c: In file included from decode.c:39:
decode.c: ./gf2x.h:23:10: error: AVX vector return of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
decode.c: ./x86_64_intrinsic.h:30:27: note: expanded from macro 'LOAD'
decode.c: # define LOAD(mem) _mm256_loadu_si256((const void *)(mem))
decode.c: ^
decode.c: In file included from decode.c:39:
decode.c: ...

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

Compiler	Implementations
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx2
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx2
clang -march=native -O -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx2
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx2

Compiler output

Implementation: T:avx2
Security model: timingleaks
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE

decode.c: In file included from decode.c:39:
decode.c: gf2x.h: In function 'gf2x_mod_add':
decode.c: gf2x.h:22:8: warning: AVX vector return without AVX enabled changes the ABI [-Wpsabi]
decode.c: 22 | va = LOAD(&a_qwords[i]);
decode.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/10/include/immintrin.h:51,
decode.c: from x86_64_intrinsic.h:20,
decode.c: from defs.h:103,
decode.c: from bike_defs.h:10,
decode.c: from types.h:13,
decode.c: from decode.h:10,
decode.c: from decode.c:37:
decode.c: /usr/lib/gcc/x86_64-linux-gnu/10/include/avxintrin.h:926:1: error: inlining failed in call to 'always_inline' '_mm256_storeu_si256': target specific option mismatch
decode.c: 926 | _mm256_storeu_si256 (__m256i_u *__P, __m256i __A)
decode.c: | ^~~~~~~~~~~~~~~~~~~
decode.c: In file included from defs.h:103,
decode.c: from bike_defs.h:10,
decode.c: from types.h:13,
decode.c: from decode.h:10,
decode.c: from decode.c:37:
decode.c: x86_64_intrinsic.h:31:27: note: called from here
decode.c: 31 | # define STORE(mem, reg) _mm256_storeu_si256((void *)(mem), (reg))
decode.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
decode.c: gf2x.h:25:5: note: in expansion of macro 'STORE'
decode.c: 25 | STORE(&c_qwords[i], va ^ vb);
decode.c: | ^~~~~
decode.c: ...

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

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx2
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx2
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx2
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx2

Compiler output

Implementation: T:avx512
Security model: timingleaks
Compiler: clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

decode.c: In file included from decode.c:39:
decode.c: ./gf2x.h:22:10: error: always_inline function '_mm512_loadu_si512' requires target feature 'avx512f', but would be inlined into function 'gf2x_mod_add' that is compiled without support for 'avx512f'
decode.c: va = LOAD(&a_qwords[i]);
decode.c: ^
decode.c: ./x86_64_intrinsic.h:40:27: note: expanded from macro 'LOAD'
decode.c: # define LOAD(mem) _mm512_loadu_si512((mem))
decode.c: ^
decode.c: In file included from decode.c:39:
decode.c: ./gf2x.h:22:10: error: AVX vector return of type '__m512i' (vector of 8 'long long' values) without 'avx512f' enabled changes the ABI
decode.c: ./x86_64_intrinsic.h:40:27: note: expanded from macro 'LOAD'
decode.c: # define LOAD(mem) _mm512_loadu_si512((mem))
decode.c: ^
decode.c: In file included from decode.c:39:
decode.c: ./gf2x.h:23:10: error: always_inline function '_mm512_loadu_si512' requires target feature 'avx512f', but would be inlined into function 'gf2x_mod_add' that is compiled without support for 'avx512f'
decode.c: vb = LOAD(&b_qwords[i]);
decode.c: ^
decode.c: ./x86_64_intrinsic.h:40:27: note: expanded from macro 'LOAD'
decode.c: # define LOAD(mem) _mm512_loadu_si512((mem))
decode.c: ^
decode.c: In file included from decode.c:39:
decode.c: ./gf2x.h:23:10: error: AVX vector return of type '__m512i' (vector of 8 'long long' values) without 'avx512f' enabled changes the ABI
decode.c: ./x86_64_intrinsic.h:40:27: note: expanded from macro 'LOAD'
decode.c: # define LOAD(mem) _mm512_loadu_si512((mem))
decode.c: ^
decode.c: In file included from decode.c:39:
decode.c: ...

Number of similar (compiler,implementation) pairs: 8, namely:

Compiler	Implementations
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512
clang -march=native -O -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512-vpclmul
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512-vpclmul
clang -march=native -O -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512-vpclmul
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512-vpclmul

Compiler output

Implementation: T:avx512
Security model: timingleaks
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE

decode.c: In file included from decode.c:39:
decode.c: gf2x.h: In function 'gf2x_mod_add':
decode.c: gf2x.h:22:8: warning: AVX512F vector return without AVX512F enabled changes the ABI [-Wpsabi]
decode.c: 22 | va = LOAD(&a_qwords[i]);
decode.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/10/include/immintrin.h:55,
decode.c: from x86_64_intrinsic.h:20,
decode.c: from defs.h:103,
decode.c: from bike_defs.h:10,
decode.c: from types.h:13,
decode.c: from decode.h:10,
decode.c: from decode.c:37:
decode.c: /usr/lib/gcc/x86_64-linux-gnu/10/include/avx512fintrin.h:6429:1: error: inlining failed in call to 'always_inline' '_mm512_storeu_si512': target specific option mismatch
decode.c: 6429 | _mm512_storeu_si512 (void *__P, __m512i __A)
decode.c: | ^~~~~~~~~~~~~~~~~~~
decode.c: In file included from defs.h:103,
decode.c: from bike_defs.h:10,
decode.c: from types.h:13,
decode.c: from decode.h:10,
decode.c: from decode.c:37:
decode.c: x86_64_intrinsic.h:41:27: note: called from here
decode.c: 41 | # define STORE(mem, reg) _mm512_storeu_si512((mem), (reg))
decode.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
decode.c: gf2x.h:25:5: note: in expansion of macro 'STORE'
decode.c: 25 | STORE(&c_qwords[i], va ^ vb);
decode.c: | ^~~~~
decode.c: ...

Number of similar (compiler,implementation) pairs: 8, namely:

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512-vpclmul
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512-vpclmul
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512-vpclmul
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512-vpclmul

Compiler output

Implementation: T:ches2021
Security model: timingleaks
Compiler: clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

decode.c: decode.c:303:20: error: '__builtin_ia32_vec_ext_v4di' needs target feature avx
decode.c: printf("%.16llX", _mm256_extract_epi64(v, 0));
decode.c: ^
decode.c: /usr/lib/llvm-11/lib/clang/11.0.1/include/avxintrin.h:2024:14: note: expanded from macro '_mm256_extract_epi64'
decode.c: (long long)__builtin_ia32_vec_ext_v4di((__v4di)(__m256i)(X), (int)(N))
decode.c: ^
decode.c: decode.c:304:21: error: '__builtin_ia32_vec_ext_v4di' needs target feature avx
decode.c: printf(" %.16llX", _mm256_extract_epi64(v, 1));
decode.c: ^
decode.c: /usr/lib/llvm-11/lib/clang/11.0.1/include/avxintrin.h:2024:14: note: expanded from macro '_mm256_extract_epi64'
decode.c: (long long)__builtin_ia32_vec_ext_v4di((__v4di)(__m256i)(X), (int)(N))
decode.c: ^
decode.c: decode.c:305:21: error: '__builtin_ia32_vec_ext_v4di' needs target feature avx
decode.c: printf(" %.16llX", _mm256_extract_epi64(v, 2));
decode.c: ^
decode.c: /usr/lib/llvm-11/lib/clang/11.0.1/include/avxintrin.h:2024:14: note: expanded from macro '_mm256_extract_epi64'
decode.c: (long long)__builtin_ia32_vec_ext_v4di((__v4di)(__m256i)(X), (int)(N))
decode.c: ^
decode.c: decode.c:306:23: error: '__builtin_ia32_vec_ext_v4di' needs target feature avx
decode.c: printf(" %.16llX\n", _mm256_extract_epi64(v, 3));
decode.c: ^
decode.c: /usr/lib/llvm-11/lib/clang/11.0.1/include/avxintrin.h:2024:14: note: expanded from macro '_mm256_extract_epi64'
decode.c: (long long)__builtin_ia32_vec_ext_v4di((__v4di)(__m256i)(X), (int)(N))
decode.c: ^
decode.c: 4 errors generated.

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

Compiler	Implementations
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:ches2021
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:ches2021
clang -march=native -O -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:ches2021
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:ches2021

Compiler output

Implementation: T:ches2021
Security model: timingleaks
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE

decode.c: decode.c: In function 'dump1':
decode.c: decode.c:301:6: note: the ABI for passing parameters with 32-byte alignment has changed in GCC 4.6
decode.c: 301 | void dump1(__m256i v)
decode.c: | ^~~~~
decode.c: decode.c: In function 'find_err1':
decode.c: decode.c:351:11: warning: AVX vector return without AVX enabled changes the ABI [-Wpsabi]
decode.c: 351 | buf[j] = LOAD(&syndrome->qw[4 * j]);
decode.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/10/include/immintrin.h:51,
decode.c: from x86_64_intrinsic.h:20,
decode.c: from defs.h:106,
decode.c: from bike_defs.h:10,
decode.c: from types.h:15,
decode.c: from decode.h:17,
decode.c: from decode.c:39:
decode.c: gf2x.h: In function 'gf2x_mod_add':
decode.c: /usr/lib/gcc/x86_64-linux-gnu/10/include/avxintrin.h:926:1: error: inlining failed in call to 'always_inline' '_mm256_storeu_si256': target specific option mismatch
decode.c: 926 | _mm256_storeu_si256 (__m256i_u *__P, __m256i __A)
decode.c: | ^~~~~~~~~~~~~~~~~~~
decode.c: In file included from defs.h:106,
decode.c: from bike_defs.h:10,
decode.c: from types.h:15,
decode.c: from decode.h:17,
decode.c: from decode.c:39:
decode.c: x86_64_intrinsic.h:31:27: note: called from here
decode.c: 31 | # define STORE(mem, reg) _mm256_storeu_si256((void *)(mem), (reg))
decode.c: ...

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

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:ches2021
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:ches2021
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:ches2021
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:ches2021