Implementation notes: amd64, hertz, crypto_hash/blake512

Computer: hertz
Microarchitecture: amd64; Zen 4 (a60f12)
Architecture: amd64
CPU ID: AuthenticAMD-00a60f12-178bfbff
SUPERCOP version: 20240425
Operation: crypto_hash
Primitive: blake512

Time	Object size	Test size	Implementation	Compiler	Benchmark date	SUPERCOP version
8570	14556 0 0	24800 780 936	`regs`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
8582	15383 0 0	33055 828 968	`bswap`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
8597	14556 0 0	24800 780 936	`bswap`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
8605	14583 0 0	26168 820 968	`bswap`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
8612	16854 0 0	28597 804 968	`regs`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
8677	15415 0 0	33199 828 968	`bswap`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
8688	15418 0 0	29237 804 1032	`bswap`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
8708	10982 0 0	28758 828 968	`sse41`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
8714	15195 0 0	26941 804 968	`bswap`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
8728	10181 0 0	21792 820 968	`sse41`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
8735	10966 0 0	28630 828 968	`sse41`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
8837	29513 0 0	43405 804 1032	`sphlib`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
8856	26059 0 0	43911 828 968	`sphlib`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
8926	28962 0 0	40797 804 968	`sphlib`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
8972	26098 0 0	37744 820 968	`sphlib`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
8994	16893 0 0	30709 804 1032	`regs`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
9188	11516 0 0	29254 828 968	`ssse3`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9206	11484 0 0	29110 828 968	`ssse3`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9235	10685 0 0	22272 820 968	`ssse3`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9346	26059 0 0	43799 828 968	`sphlib`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9367	11518 0 0	29190 828 968	`sse2s`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9368	11550 0 0	29334 828 968	`sse2s`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9386	10719 0 0	22352 820 968	`sse2s`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9499	11234 0 0	28886 828 968	`sse2`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9534	11266 0 0	29030 828 968	`sse2`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9570	10435 0 0	22032 820 968	`sse2`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9633	4861 0 0	22614 828 968	`ref`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9650	4829 0 0	22470 828 968	`ref`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
9991	13910 0 0	27709 804 1032	`ssse3`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10099	13886 0 0	25677 804 968	`ssse3`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10150	11521 0 0	25293 804 1032	`sse41`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10157	11505 0 0	23261 804 968	`sse41`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10182	11147 0 0	21408 780 936	`sse41`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10268	13782 0 0	27549 804 1032	`sse2s`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10270	13758 0 0	25517 804 968	`sse2s`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10541	18600 0 0	36358 828 968	`regs`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
10702	18568 0 0	36214 828 968	`regs`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
10709	17769 0 0	29376 820 968	`regs`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
10732	24536 0 0	34872 780 936	`sphlib`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10866	14990 0 0	26749 804 968	`sse2`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
10946	15014 0 0	28781 804 1032	`sse2`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
11221	13399 0 0	23656 780 936	`sse2`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
11377	12557 0 0	22808 780 936	`ssse3`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
11395	12656 0 0	22920 780 936	`sse2s`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
11512	7650 0 0	17976 780 936	`sphlib-small`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
11873	7258 0 0	18928 820 968	`sphlib-small`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
11877	9039 0 0	26839 828 968	`sphlib-small`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
11947	9039 0 0	26951 828 968	`sphlib-small`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
12143	3706 0 0	15312 820 968	`ref`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
12740	3339 0 0	13576 780 936	`ref`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
14009	11074 0 0	22909 804 968	`sphlib-small`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
14041	11609 0 0	25501 804 1032	`sphlib-small`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
14463	6942 0 0	18717 804 968	`ref`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
14493	7029 0 0	20813 804 1032	`ref`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
16141	14958 0 0	32639 828 968	`sandy`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
16148	14990 0 0	32783 828 968	`sandy`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
16189	14158 0 0	25752 820 968	`sandy`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20240429	20240425
16502	15022 0 0	25264 780 936	`sandy`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
16706	15642 0 0	29461 804 1032	`sandy`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217
16723	15419 0 0	27165 804 968	`sandy`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20231219	20231217

Compiler output

Implementation: sse41
Security model: constbranchindex
Compiler: clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

hash.c: In file included from hash.c:8:
hash.c: ./rounds.h:8:10: warning: '_mm_roti_epi64' macro redefined [-Wmacro-redefined]
hash.c: 8 | #define _mm_roti_epi64(x, c) \
hash.c: | ^
hash.c: /usr/lib/llvm-18/lib/clang/18/include/xopintrin.h:236:9: note: previous definition is here
hash.c: 236 | #define _mm_roti_epi64(A, N) \
hash.c: | ^
hash.c: 1 warning generated.

Number of similar (compiler,implementation) pairs: 3, namely:

Compiler	Implementations
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	sse41
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	sse41
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	sse41

Compiler output

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

hash.c: hash.c:81:8: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: 81 | m0 = BSWAP64(m0);
hash.c: | ^
hash.c: ./rounds.h:13:21: note: expanded from macro 'BSWAP64'
hash.c: 13 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^
hash.c: hash.c:82:8: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: 82 | m1 = BSWAP64(m1);
hash.c: | ^
hash.c: ./rounds.h:13:21: note: expanded from macro 'BSWAP64'
hash.c: 13 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^
hash.c: hash.c:83:8: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: 83 | m2 = BSWAP64(m2);
hash.c: | ^
hash.c: ./rounds.h:13:21: note: expanded from macro 'BSWAP64'
hash.c: 13 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^
hash.c: hash.c:84:8: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: 84 | m3 = BSWAP64(m3);
hash.c: | ^
hash.c: ./rounds.h:13:21: note: expanded from macro 'BSWAP64'
hash.c: 13 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^
hash.c: hash.c:85:8: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: ...

Number of similar (compiler,implementation) pairs: 3, namely:

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

Compiler output

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

hash.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/13/include/x86intrin.h:38,
hash.c: from hash.c:5:
hash.c: /usr/lib/gcc/x86_64-linux-gnu/13/include/xopintrin.h: In function 'blake512_compress':
hash.c: /usr/lib/gcc/x86_64-linux-gnu/13/include/xopintrin.h:272:1: error: inlining failed in call to 'always_inline' '_mm_roti_epi64': target specific option mismatch
hash.c: 272 | _mm_roti_epi64(__m128i __A, const int __B)
hash.c: | ^~~~~~~~~~~~~~
hash.c: In file included from hash.c:8:
hash.c: rounds.h:825:11: note: called from here
hash.c: 825 | row2h = _mm_roti_epi64(row2h, -11); \
hash.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~
hash.c: rounds.h:867:3: note: in expansion of macro 'G2'
hash.c: 867 | G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
hash.c: | ^~
hash.c: hash.c:132:3: note: in expansion of macro 'ROUND'
hash.c: 132 | ROUND(15);
hash.c: | ^~~~~
hash.c: /usr/lib/gcc/x86_64-linux-gnu/13/include/xopintrin.h:272:1: error: inlining failed in call to 'always_inline' '_mm_roti_epi64': target specific option mismatch
hash.c: 272 | _mm_roti_epi64(__m128i __A, const int __B)
hash.c: | ^~~~~~~~~~~~~~
hash.c: rounds.h:824:11: note: called from here
hash.c: 824 | row2l = _mm_roti_epi64(row2l, -11); \
hash.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~
hash.c: rounds.h:867:3: note: in expansion of macro 'G2'
hash.c: 867 | G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
hash.c: | ^~
hash.c: ...

Number of similar (compiler,implementation) pairs: 3, namely:

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

Compiler output

Implementation: xop-2
Security model: constbranchindex
Compiler: clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

hash.c: hash.c:92:15: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: 92 | m.u128[0] = BSWAP64(m.u128[0]);
hash.c: | ^
hash.c: ./rounds.h:15:21: note: expanded from macro 'BSWAP64'
hash.c: 15 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^
hash.c: hash.c:93:15: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: 93 | m.u128[1] = BSWAP64(m.u128[1]);
hash.c: | ^
hash.c: ./rounds.h:15:21: note: expanded from macro 'BSWAP64'
hash.c: 15 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^
hash.c: hash.c:94:15: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: 94 | m.u128[2] = BSWAP64(m.u128[2]);
hash.c: | ^
hash.c: ./rounds.h:15:21: note: expanded from macro 'BSWAP64'
hash.c: 15 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^
hash.c: hash.c:95:15: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: 95 | m.u128[3] = BSWAP64(m.u128[3]);
hash.c: | ^
hash.c: ./rounds.h:15:21: note: expanded from macro 'BSWAP64'
hash.c: 15 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^
hash.c: hash.c:96:15: error: always_inline function '_mm_perm_epi8' requires target feature 'xop', but would be inlined into function 'blake512_compress' that is compiled without support for 'xop'
hash.c: ...

Number of similar (compiler,implementation) pairs: 3, namely:

Compiler	Implementations
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	xop-2
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	xop-2
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	xop-2

Compiler output

Implementation: xop-2
Security model: constbranchindex
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE

hash.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/13/include/x86intrin.h:38,
hash.c: from hash.c:5:
hash.c: /usr/lib/gcc/x86_64-linux-gnu/13/include/xopintrin.h: In function 'blake512_compress':
hash.c: /usr/lib/gcc/x86_64-linux-gnu/13/include/xopintrin.h:218:1: error: inlining failed in call to 'always_inline' '_mm_perm_epi8': target specific option mismatch
hash.c: 218 | _mm_perm_epi8(__m128i __A, __m128i __B, __m128i __C)
hash.c: | ^~~~~~~~~~~~~
hash.c: In file included from hash.c:8:
hash.c: rounds.h:15:28: note: called from here
hash.c: 15 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
hash.c: hash.c:99:15: note: in expansion of macro 'BSWAP64'
hash.c: 99 | m.u128[7] = BSWAP64(m.u128[7]);
hash.c: | ^~~~~~~
hash.c: /usr/lib/gcc/x86_64-linux-gnu/13/include/xopintrin.h:218:1: error: inlining failed in call to 'always_inline' '_mm_perm_epi8': target specific option mismatch
hash.c: 218 | _mm_perm_epi8(__m128i __A, __m128i __B, __m128i __C)
hash.c: | ^~~~~~~~~~~~~
hash.c: rounds.h:15:28: note: called from here
hash.c: 15 | #define BSWAP64(x) _mm_perm_epi8((x),(x),u8to64)
hash.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
hash.c: hash.c:98:15: note: in expansion of macro 'BSWAP64'
hash.c: 98 | m.u128[6] = BSWAP64(m.u128[6]);
hash.c: | ^~~~~~~
hash.c: /usr/lib/gcc/x86_64-linux-gnu/13/include/xopintrin.h:218:1: error: inlining failed in call to 'always_inline' '_mm_perm_epi8': target specific option mismatch
hash.c: 218 | _mm_perm_epi8(__m128i __A, __m128i __B, __m128i __C)
hash.c: | ^~~~~~~~~~~~~
hash.c: ...

Number of similar (compiler,implementation) pairs: 3, namely:

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE	xop-2
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv -fPIC -fPIE	xop-2
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv -fPIC -fPIE	xop-2