Implementation notes: amd64, h8bobcat, crypto_core/multsntrup1013

Computer: h8bobcat
Microarchitecture: amd64; Bobcat (500f10)
Architecture: amd64
CPU ID: AuthenticAMD-00500f20-178bfbff
SUPERCOP version: 20240107
Operation: crypto_core
Primitive: multsntrup1013

Time	Object size	Test size	Implementation	Compiler	Benchmark date	SUPERCOP version
1415187	6400 0 0	20406 776 832	`ref`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231215	20231212
2740822	2020 0 0	15484 816 760	`ref`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20231215	20231212
2740840	2020 0 0	14340 816 760	`ref`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20231215	20231212
2782222	2085 0 0	15764 816 760	`ref`	`clang_-mcpu=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20231215	20231212
5361093	556 0 0	11553 752 800	`ref`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231215	20231212
5492287	611 0 0	11310 808 760	`ref`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20231215	20231212
5703572	697 0 0	13254 776 832	`ref`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231215	20231212
5769265	666 0 0	11972 816 760	`ref`	`clang_-march=native_-O_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20231215	20231212
6012550	651 0 0	12845 768 832	`ref`	`gcc_-march=native_-mtune=native_-O_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231215	20231212

Compiler output

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

mult1024.c: mult1024.c:307:7: error: always_inline function '_mm256_set1_epi16' requires target feature 'avx', but would be inlined into function 'crypto_core_multsntrup1013_avx_constbranchindex' that is compiled without support for 'avx'
mult1024.c: x = const_x16(0);
mult1024.c: ^
mult1024.c: mult1024.c:10:19: note: expanded from macro 'const_x16'
mult1024.c: #define const_x16 _mm256_set1_epi16
mult1024.c: ^
mult1024.c: mult1024.c:307:7: error: AVX vector return of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
mult1024.c: mult1024.c:10:19: note: expanded from macro 'const_x16'
mult1024.c: #define const_x16 _mm256_set1_epi16
mult1024.c: ^
mult1024.c: mult1024.c:308:36: error: always_inline function '_mm256_storeu_si256' requires target feature 'avx', but would be inlined into function 'crypto_core_multsntrup1013_avx_constbranchindex' that is compiled without support for 'avx'
mult1024.c: for (i = p&~15;i < 1024;i += 16) store_x16(&f[i],x);
mult1024.c: ^
mult1024.c: mult1024.c:9:24: note: expanded from macro 'store_x16'
mult1024.c: #define store_x16(p,v) _mm256_storeu_si256((int16x16 *) (p),(v))
mult1024.c: ^
mult1024.c: mult1024.c:308:36: error: AVX vector argument of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
mult1024.c: mult1024.c:9:24: note: expanded from macro 'store_x16'
mult1024.c: #define store_x16(p,v) _mm256_storeu_si256((int16x16 *) (p),(v))
mult1024.c: ^
mult1024.c: mult1024.c:309:36: error: always_inline function '_mm256_storeu_si256' requires target feature 'avx', but would be inlined into function 'crypto_core_multsntrup1013_avx_constbranchindex' that is compiled without support for 'avx'
mult1024.c: for (i = p&~15;i < 1024;i += 16) store_x16(&g[i],x);
mult1024.c: ^
mult1024.c: mult1024.c:9:24: note: expanded from macro 'store_x16'
mult1024.c: #define store_x16(p,v) _mm256_storeu_si256((int16x16 *) (p),(v))
mult1024.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

mult1024.c: mult1024.c: In function 'squeeze_7177_x16':
mult1024.c: mult1024.c:19:1: warning: AVX vector return without AVX enabled changes the ABI [-Wpsabi]
mult1024.c: 19 | {
mult1024.c: | ^
mult1024.c: mult1024.c: In function 'squeeze_7681_x16':
mult1024.c: mult1024.c:23:24: note: the ABI for passing parameters with 32-byte alignment has changed in GCC 4.6
mult1024.c: 23 | static inline int16x16 squeeze_7681_x16(int16x16 x)
mult1024.c: | ^~~~~~~~~~~~~~~~
mult1024.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:47,
mult1024.c: from mult1024.c:2:
mult1024.c: mult1024.c: In function 'squeeze_7177_x16':
mult1024.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
mult1024.c: 808 | _mm256_sub_epi16 (__m256i __A, __m256i __B)
mult1024.c: | ^~~~~~~~~~~~~~~~
mult1024.c: mult1024.c:12:17: note: called from here
mult1024.c: 12 | #define sub_x16 _mm256_sub_epi16
mult1024.c: | ^
mult1024.c: mult1024.c:20:10: note: in expansion of macro 'sub_x16'
mult1024.c: 20 | return sub_x16(x,mullo_x16(mulhrs_x16(x,const_x16(4)),const_x16(7177)));
mult1024.c: | ^~~~~~~
mult1024.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:47,
mult1024.c: from mult1024.c:2:
mult1024.c: /usr/lib/gcc/x86_64-linux-gnu/11/include/avx2intrin.h:553:1: error: inlining failed in call to 'always_inline' '_mm256_mullo_epi16': target specific option mismatch
mult1024.c: 553 | _mm256_mullo_epi16 (__m256i __A, __m256i __B)
mult1024.c: | ^~~~~~~~~~~~~~~~~~
mult1024.c: ...

Number of similar (compiler,implementation) pairs: 8, 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
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE	avx800
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv -fPIC -fPIE	avx800
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv -fPIC -fPIE	avx800
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv -fPIC -fPIE	avx800

Compiler output

Implementation: avx800
Security model: constbranchindex
Compiler: clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

mult1024.c: mult1024.c:307:7: error: always_inline function '_mm256_set1_epi16' requires target feature 'avx', but would be inlined into function 'crypto_core_multsntrup1013_avx800_constbranchindex' that is compiled without support for 'avx'
mult1024.c: x = const_x16(0);
mult1024.c: ^
mult1024.c: mult1024.c:10:19: note: expanded from macro 'const_x16'
mult1024.c: #define const_x16 _mm256_set1_epi16
mult1024.c: ^
mult1024.c: mult1024.c:307:7: error: AVX vector return of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
mult1024.c: mult1024.c:10:19: note: expanded from macro 'const_x16'
mult1024.c: #define const_x16 _mm256_set1_epi16
mult1024.c: ^
mult1024.c: mult1024.c:308:36: error: always_inline function '_mm256_storeu_si256' requires target feature 'avx', but would be inlined into function 'crypto_core_multsntrup1013_avx800_constbranchindex' that is compiled without support for 'avx'
mult1024.c: for (i = p&~15;i < 1024;i += 16) store_x16(&f[i],x);
mult1024.c: ^
mult1024.c: mult1024.c:9:24: note: expanded from macro 'store_x16'
mult1024.c: #define store_x16(p,v) _mm256_storeu_si256((int16x16 *) (p),(v))
mult1024.c: ^
mult1024.c: mult1024.c:308:36: error: AVX vector argument of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
mult1024.c: mult1024.c:9:24: note: expanded from macro 'store_x16'
mult1024.c: #define store_x16(p,v) _mm256_storeu_si256((int16x16 *) (p),(v))
mult1024.c: ^
mult1024.c: mult1024.c:309:36: error: always_inline function '_mm256_storeu_si256' requires target feature 'avx', but would be inlined into function 'crypto_core_multsntrup1013_avx800_constbranchindex' that is compiled without support for 'avx'
mult1024.c: for (i = p&~15;i < 1024;i += 16) store_x16(&g[i],x);
mult1024.c: ^
mult1024.c: mult1024.c:9:24: note: expanded from macro 'store_x16'
mult1024.c: #define store_x16(p,v) _mm256_storeu_si256((int16x16 *) (p),(v))
mult1024.c: ...

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

Compiler	Implementations
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx800
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx800
clang -march=native -O -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx800
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx800
clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	avx800