Implementation notes: x86, samba, crypto_kem/kyber1024

Computer: samba
Architecture: x86
CPU ID: GenuineIntel-000506e3-bfebfbff
SUPERCOP version: 20190803
Operation: crypto_kem
Primitive: kyber1024

Time	Implementation	Compiler	Benchmark date	SUPERCOP version
1480881	`ref`	`gcc -funroll-loops -m32 -march=pentium-m -O3 -fomit-frame-pointer`	20190810	20190803
1525601	`ref`	`gcc -funroll-loops -m32 -march=pentium-m -O2 -fomit-frame-pointer`	20190810	20190803
1529954	`ref`	`gcc -m32 -march=core2 -msse4.1 -O3 -fomit-frame-pointer`	20190810	20190803
1534627	`ref`	`gcc -m32 -march=pentium-m -O3 -fomit-frame-pointer`	20190810	20190803
1539149	`ref`	`gcc -m32 -march=core2 -msse4 -O3 -fomit-frame-pointer`	20190810	20190803
1550508	`ref`	`gcc -m32 -march=corei7-avx -O2 -fomit-frame-pointer`	20190810	20190803
1552554	`ref`	`gcc -m32 -march=core2 -O3 -fomit-frame-pointer`	20190810	20190803
1558880	`ref`	`gcc -m32 -march=core-avx2 -O2 -fomit-frame-pointer`	20190810	20190803
1559095	`ref`	`gcc -m32 -march=native -mtune=native -O3 -fomit-frame-pointer`	20190810	20190803
1559218	`ref`	`gcc -m32 -march=corei7 -O3 -fomit-frame-pointer`	20190810	20190803
1561679	`ref`	`gcc -m32 -march=core-avx-i -O2 -fomit-frame-pointer`	20190810	20190803
1566581	`ref`	`gcc -m32 -march=native -mtune=native -O2 -fomit-frame-pointer`	20190810	20190803
1568294	`ref`	`gcc -m32 -march=core-avx2 -O3 -fomit-frame-pointer`	20190810	20190803
1583988	`ref`	`gcc -m32 -march=native -mtune=native -Os -fomit-frame-pointer`	20190810	20190803
1603906	`ref`	`gcc -m32 -march=core-avx2 -Os -fomit-frame-pointer`	20190810	20190803
1609651	`ref`	`gcc -m32 -march=pentium-m -O2 -fomit-frame-pointer`	20190810	20190803
1613760	`ref`	`gcc -m32 -march=core2 -msse4.1 -O2 -fomit-frame-pointer`	20190810	20190803
1614415	`ref`	`gcc -funroll-loops -m32 -march=pentium-m -O -fomit-frame-pointer`	20190810	20190803
1616387	`ref`	`gcc -m32 -march=corei7 -O2 -fomit-frame-pointer`	20190810	20190803
1624760	`ref`	`gcc -m32 -march=core-avx-i -Os -fomit-frame-pointer`	20190810	20190803
1626962	`ref`	`gcc -funroll-loops -m32 -march=k8 -O -fomit-frame-pointer`	20190810	20190803
1627443	`ref`	`gcc -m32 -march=core2 -msse4 -Os -fomit-frame-pointer`	20190810	20190803
1632479	`ref`	`gcc -m32 -march=core2 -msse4 -O2 -fomit-frame-pointer`	20190810	20190803
1635276	`ref`	`gcc -m32 -march=corei7-avx -Os -fomit-frame-pointer`	20190810	20190803
1640246	`ref`	`gcc -m32 -march=corei7 -Os -fomit-frame-pointer`	20190810	20190803
1643916	`ref`	`gcc -m32 -march=core2 -msse4.1 -Os -fomit-frame-pointer`	20190810	20190803
1655583	`ref`	`gcc -funroll-loops -m32 -march=barcelona -O -fomit-frame-pointer`	20190810	20190803
1659012	`ref`	`gcc -m32 -march=nocona -Os -fomit-frame-pointer`	20190810	20190803
1662582	`ref`	`gcc -m32 -march=corei7-avx -O3 -fomit-frame-pointer`	20190810	20190803
1669483	`ref`	`gcc -funroll-loops -m32 -march=pentium-m -Os -fomit-frame-pointer`	20190810	20190803
1673552	`ref`	`gcc -m32 -march=core2 -Os -fomit-frame-pointer`	20190810	20190803
1673627	`ref`	`gcc -m32 -march=pentium-m -Os -fomit-frame-pointer`	20190810	20190803
1673863	`ref`	`gcc -m32 -march=prescott -Os -fomit-frame-pointer`	20190810	20190803
1678433	`ref`	`gcc -m32 -march=core-avx-i -O3 -fomit-frame-pointer`	20190810	20190803
1681744	`ref`	`gcc -m32 -march=pentium4 -Os -fomit-frame-pointer`	20190810	20190803
1697893	`ref`	`gcc -funroll-loops -m32 -march=nocona -Os -fomit-frame-pointer`	20190810	20190803
1702447	`ref`	`gcc -funroll-loops -m32 -march=pentium4 -Os -fomit-frame-pointer`	20190810	20190803
1702867	`ref`	`gcc -m32 -march=core2 -O2 -fomit-frame-pointer`	20190810	20190803
1718077	`ref`	`gcc -funroll-loops -m32 -march=prescott -Os -fomit-frame-pointer`	20190810	20190803
1725271	`ref`	`gcc -m32 -march=core-avx2 -O -fomit-frame-pointer`	20190810	20190803
1729123	`ref`	`gcc -m32 -march=k8 -O -fomit-frame-pointer`	20190810	20190803
1735434	`ref`	`gcc -m32 -march=native -mtune=native -O -fomit-frame-pointer`	20190810	20190803
1737248	`ref`	`gcc -m32 -march=pentium-m -O -fomit-frame-pointer`	20190810	20190803
1744835	`ref`	`gcc -funroll-loops -m32 -march=prescott -O3 -fomit-frame-pointer`	20190810	20190803
1749703	`ref`	`gcc -funroll-loops -m32 -march=nocona -O3 -fomit-frame-pointer`	20190810	20190803
1756149	`ref`	`gcc -funroll-loops -m32 -march=nocona -O2 -fomit-frame-pointer`	20190810	20190803
1760540	`ref`	`gcc -funroll-loops -m32 -march=prescott -O -fomit-frame-pointer`	20190810	20190803
1765170	`ref`	`gcc -funroll-loops -m32 -march=prescott -O2 -fomit-frame-pointer`	20190810	20190803
1778778	`ref`	`gcc -m32 -march=corei7 -O -fomit-frame-pointer`	20190810	20190803
1781719	`ref`	`gcc -m32 -march=prescott -O3 -fomit-frame-pointer`	20190810	20190803
1787903	`ref`	`gcc -m32 -march=barcelona -O -fomit-frame-pointer`	20190810	20190803
1789133	`ref`	`gcc -funroll-loops -m32 -march=nocona -O -fomit-frame-pointer`	20190810	20190803
1794152	`ref`	`gcc -m32 -march=core2 -O -fomit-frame-pointer`	20190810	20190803
1801285	`ref`	`gcc -m32 -march=core2 -msse4.1 -O -fomit-frame-pointer`	20190810	20190803
1804207	`ref`	`gcc -funroll-loops -m32 -march=athlon -O3 -fomit-frame-pointer`	20190810	20190803
1806046	`ref`	`gcc -m32 -march=core-avx-i -O -fomit-frame-pointer`	20190810	20190803
1806063	`ref`	`gcc -m32 -march=core2 -msse4 -O -fomit-frame-pointer`	20190810	20190803
1823319	`ref`	`gcc -m32 -march=corei7-avx -O -fomit-frame-pointer`	20190810	20190803
1824971	`ref`	`gcc -m32 -march=nocona -O3 -fomit-frame-pointer`	20190810	20190803
1828897	`ref`	`gcc -funroll-loops -m32 -march=pentium2 -O -fomit-frame-pointer`	20190810	20190803
1830659	`ref`	`gcc -funroll-loops -m32 -O3 -fomit-frame-pointer`	20190810	20190803
1846670	`ref`	`gcc -funroll-loops -m32 -march=pentium3 -O -fomit-frame-pointer`	20190810	20190803
1857235	`ref`	`gcc -funroll-loops -m32 -march=pentium4 -O3 -fomit-frame-pointer`	20190810	20190803
1862847	`ref`	`gcc -funroll-loops -m32 -march=pentium4 -O2 -fomit-frame-pointer`	20190810	20190803
1868077	`ref`	`gcc -funroll-loops -m32 -march=pentiumpro -O -fomit-frame-pointer`	20190810	20190803
1870673	`ref`	`gcc -funroll-loops -m32 -march=k6 -O3 -fomit-frame-pointer`	20190810	20190803
1874068	`ref`	`gcc -funroll-loops -m32 -march=k6 -O -fomit-frame-pointer`	20190810	20190803
1880216	`ref`	`gcc -m32 -march=athlon -O3 -fomit-frame-pointer`	20190810	20190803
1884337	`ref`	`gcc -funroll-loops -m32 -march=pentium2 -O3 -fomit-frame-pointer`	20190810	20190803
1887563	`ref`	`gcc -funroll-loops -m32 -O -fomit-frame-pointer`	20190810	20190803
1888530	`ref`	`gcc -funroll-loops -m32 -march=k6-3 -O3 -fomit-frame-pointer`	20190810	20190803
1889355	`ref`	`gcc -funroll-loops -m32 -march=k6-2 -O -fomit-frame-pointer`	20190810	20190803
1892423	`ref`	`gcc -funroll-loops -m32 -march=pentium4 -O -fomit-frame-pointer`	20190810	20190803
1893553	`ref`	`gcc -funroll-loops -m32 -march=k6-3 -O -fomit-frame-pointer`	20190810	20190803
1894171	`ref`	`gcc -funroll-loops -m32 -march=pentium3 -O3 -fomit-frame-pointer`	20190810	20190803
1896726	`ref`	`gcc -funroll-loops -m32 -march=athlon -O2 -fomit-frame-pointer`	20190810	20190803
1899440	`ref`	`gcc -m32 -march=nocona -O2 -fomit-frame-pointer`	20190810	20190803
1906256	`ref`	`gcc -m32 -march=pentium4 -O3 -fomit-frame-pointer`	20190810	20190803
1910080	`ref`	`gcc -m32 -march=nocona -O -fomit-frame-pointer`	20190810	20190803
1910417	`ref`	`gcc -m32 -march=prescott -O2 -fomit-frame-pointer`	20190810	20190803
1910717	`ref`	`gcc -funroll-loops -m32 -march=athlon -O -fomit-frame-pointer`	20190810	20190803
1912551	`ref`	`gcc -funroll-loops -m32 -march=k6-2 -O3 -fomit-frame-pointer`	20190810	20190803
1914032	`ref`	`gcc -funroll-loops -m32 -march=pentiumpro -O3 -fomit-frame-pointer`	20190810	20190803
1917164	`ref`	`gcc -funroll-loops -m32 -march=k6 -O2 -fomit-frame-pointer`	20190810	20190803
1919232	`ref`	`gcc -funroll-loops -m32 -O2 -fomit-frame-pointer`	20190810	20190803
1923861	`ref`	`gcc -m32 -march=prescott -O -fomit-frame-pointer`	20190810	20190803
1924649	`ref`	`gcc -funroll-loops -m32 -march=k6-3 -O2 -fomit-frame-pointer`	20190810	20190803
1927099	`ref`	`gcc -m32 -O3 -fomit-frame-pointer`	20190810	20190803
1931520	`ref`	`gcc -m32 -march=k6-3 -O3 -fomit-frame-pointer`	20190810	20190803
1932302	`ref`	`gcc -funroll-loops -m32 -march=k6 -Os -fomit-frame-pointer`	20190810	20190803
1932952	`ref`	`gcc -funroll-loops -m32 -march=pentium3 -O2 -fomit-frame-pointer`	20190810	20190803
1935430	`ref`	`gcc -funroll-loops -m32 -march=k6-2 -Os -fomit-frame-pointer`	20190810	20190803
1938723	`ref`	`gcc -funroll-loops -m32 -march=pentiumpro -O2 -fomit-frame-pointer`	20190810	20190803
1943200	`ref`	`gcc -m32 -march=pentium3 -O3 -fomit-frame-pointer`	20190810	20190803
1948238	`ref`	`gcc -funroll-loops -m32 -march=k6-3 -Os -fomit-frame-pointer`	20190810	20190803
1951288	`ref`	`gcc -funroll-loops -m32 -march=k6-2 -O2 -fomit-frame-pointer`	20190810	20190803
1952594	`ref`	`gcc -m32 -march=k6-2 -Os -fomit-frame-pointer`	20190810	20190803
1957151	`ref`	`gcc -m32 -march=k6-3 -Os -fomit-frame-pointer`	20190810	20190803
1958720	`ref`	`gcc -funroll-loops -m32 -march=pentium2 -O2 -fomit-frame-pointer`	20190810	20190803
1966379	`ref`	`gcc -m32 -march=k6 -Os -fomit-frame-pointer`	20190810	20190803
1968736	`ref`	`gcc -m32 -march=k6-2 -O3 -fomit-frame-pointer`	20190810	20190803
1969012	`ref`	`gcc -funroll-loops -m32 -Os -fomit-frame-pointer`	20190810	20190803
1972069	`ref`	`gcc -m32 -march=athlon -O -fomit-frame-pointer`	20190810	20190803
1973351	`ref`	`gcc -m32 -march=pentium2 -O -fomit-frame-pointer`	20190810	20190803
1977722	`ref`	`gcc -funroll-loops -m32 -march=i486 -Os -fomit-frame-pointer`	20190810	20190803
1979395	`ref`	`gcc -m32 -march=i386 -Os -fomit-frame-pointer`	20190810	20190803
1980941	`ref`	`gcc -funroll-loops -m32 -march=pentium -Os -fomit-frame-pointer`	20190810	20190803
1981387	`ref`	`gcc -funroll-loops -m32 -march=athlon -Os -fomit-frame-pointer`	20190810	20190803
1982110	`ref`	`gcc -m32 -march=i486 -Os -fomit-frame-pointer`	20190810	20190803
1984340	`ref`	`gcc -funroll-loops -m32 -march=pentium-mmx -Os -fomit-frame-pointer`	20190810	20190803
1987796	`ref`	`gcc -m32 -march=k6 -O3 -fomit-frame-pointer`	20190810	20190803
1990424	`ref`	`gcc -funroll-loops -m32 -march=i386 -Os -fomit-frame-pointer`	20190810	20190803
1990692	`ref`	`gcc -m32 -march=pentiumpro -O -fomit-frame-pointer`	20190810	20190803
1993636	`ref`	`gcc -m32 -march=pentium-mmx -Os -fomit-frame-pointer`	20190810	20190803
1995640	`ref`	`gcc -m32 -Os -fomit-frame-pointer`	20190810	20190803
1996114	`ref`	`gcc -m32 -march=pentiumpro -O3 -fomit-frame-pointer`	20190810	20190803
1996688	`ref`	`gcc -m32 -march=pentium2 -O3 -fomit-frame-pointer`	20190810	20190803
2002373	`ref`	`gcc -m32 -march=pentium -Os -fomit-frame-pointer`	20190810	20190803
2009909	`ref`	`gcc -m32 -march=pentium4 -O -fomit-frame-pointer`	20190810	20190803
2010204	`ref`	`gcc -m32 -march=pentium3 -O -fomit-frame-pointer`	20190810	20190803
2013931	`ref`	`gcc -m32 -march=pentium3 -Os -fomit-frame-pointer`	20190810	20190803
2018572	`ref`	`gcc -m32 -march=pentium2 -Os -fomit-frame-pointer`	20190810	20190803
2022117	`ref`	`gcc -m32 -march=athlon -Os -fomit-frame-pointer`	20190810	20190803
2024528	`ref`	`gcc -m32 -march=pentiumpro -Os -fomit-frame-pointer`	20190810	20190803
2030028	`ref`	`gcc -m32 -march=pentium4 -O2 -fomit-frame-pointer`	20190810	20190803
2031549	`ref`	`gcc -funroll-loops -m32 -march=pentiumpro -Os -fomit-frame-pointer`	20190810	20190803
2034123	`ref`	`gcc -funroll-loops -m32 -march=pentium2 -Os -fomit-frame-pointer`	20190810	20190803
2040583	`ref`	`gcc -m32 -march=athlon -O2 -fomit-frame-pointer`	20190810	20190803
2042903	`ref`	`gcc -m32 -O2 -fomit-frame-pointer`	20190810	20190803
2047409	`ref`	`gcc -funroll-loops -m32 -march=pentium3 -Os -fomit-frame-pointer`	20190810	20190803
2056143	`ref`	`gcc -m32 -O -fomit-frame-pointer`	20190810	20190803
2081373	`ref`	`gcc -m32 -march=k6 -O -fomit-frame-pointer`	20190810	20190803
2085407	`ref`	`gcc -m32 -march=k6-3 -O -fomit-frame-pointer`	20190810	20190803
2086228	`ref`	`gcc -m32 -march=pentiumpro -O2 -fomit-frame-pointer`	20190810	20190803
2087121	`ref`	`gcc -m32 -march=pentium2 -O2 -fomit-frame-pointer`	20190810	20190803
2091990	`ref`	`gcc -funroll-loops -m32 -march=pentium-mmx -O -fomit-frame-pointer`	20190810	20190803
2096570	`ref`	`gcc -m32 -march=pentium3 -O2 -fomit-frame-pointer`	20190810	20190803
2097716	`ref`	`gcc -m32 -march=k6-2 -O -fomit-frame-pointer`	20190810	20190803
2107069	`ref`	`gcc -funroll-loops -m32 -march=i486 -O -fomit-frame-pointer`	20190810	20190803
2117518	`ref`	`gcc -funroll-loops -m32 -march=i486 -O3 -fomit-frame-pointer`	20190810	20190803
2122482	`ref`	`gcc -funroll-loops -m32 -march=pentium -O -fomit-frame-pointer`	20190810	20190803
2124304	`ref`	`gcc -m32 -march=k6 -O2 -fomit-frame-pointer`	20190810	20190803
2128887	`ref`	`gcc -funroll-loops -m32 -march=i386 -O3 -fomit-frame-pointer`	20190810	20190803
2131067	`ref`	`gcc -m32 -march=k6-2 -O2 -fomit-frame-pointer`	20190810	20190803
2133338	`ref`	`gcc -m32 -march=k6-3 -O2 -fomit-frame-pointer`	20190810	20190803
2144270	`ref`	`gcc -funroll-loops -m32 -march=i386 -O -fomit-frame-pointer`	20190810	20190803
2152482	`ref`	`gcc -funroll-loops -m32 -march=i386 -O2 -fomit-frame-pointer`	20190810	20190803
2159151	`ref`	`gcc -funroll-loops -m32 -march=i486 -O2 -fomit-frame-pointer`	20190810	20190803
2210008	`ref`	`gcc -m32 -march=i386 -O3 -fomit-frame-pointer`	20190810	20190803
2241230	`ref`	`gcc -m32 -march=i486 -O3 -fomit-frame-pointer`	20190810	20190803
2249482	`ref`	`gcc -m32 -march=pentium -O -fomit-frame-pointer`	20190810	20190803
2259325	`ref`	`gcc -m32 -march=pentium-mmx -O -fomit-frame-pointer`	20190810	20190803
2285000	`ref`	`gcc -m32 -march=i386 -O -fomit-frame-pointer`	20190810	20190803
2294797	`ref`	`gcc -m32 -march=i486 -O -fomit-frame-pointer`	20190810	20190803
2299752	`ref`	`gcc -funroll-loops -m32 -march=pentium-mmx -O3 -fomit-frame-pointer`	20190810	20190803
2305498	`ref`	`gcc -funroll-loops -m32 -march=pentium -O3 -fomit-frame-pointer`	20190810	20190803
2313195	`ref`	`gcc -m32 -march=i386 -O2 -fomit-frame-pointer`	20190810	20190803
2314343	`ref`	`gcc -m32 -march=i486 -O2 -fomit-frame-pointer`	20190810	20190803
2349137	`ref`	`gcc -funroll-loops -m32 -march=pentium-mmx -O2 -fomit-frame-pointer`	20190810	20190803
2376923	`ref`	`gcc -funroll-loops -m32 -march=pentium -O2 -fomit-frame-pointer`	20190810	20190803
2412351	`ref`	`gcc -m32 -march=pentium-mmx -O3 -fomit-frame-pointer`	20190810	20190803
2418823	`ref`	`gcc -m32 -march=pentium -O3 -fomit-frame-pointer`	20190810	20190803
2460234	`ref`	`gcc -m32 -march=pentium -O2 -fomit-frame-pointer`	20190810	20190803
2463855	`ref`	`gcc -m32 -march=pentium-mmx -O2 -fomit-frame-pointer`	20190810	20190803
3533075	`ref`	`gcc -funroll-loops -m32 -march=barcelona -O3 -fomit-frame-pointer`	20190810	20190803
3594102	`ref`	`gcc -m32 -march=barcelona -O3 -fomit-frame-pointer`	20190810	20190803
3640216	`ref`	`gcc -funroll-loops -m32 -march=k8 -O2 -fomit-frame-pointer`	20190810	20190803
3661726	`ref`	`gcc -m32 -march=barcelona -Os -fomit-frame-pointer`	20190810	20190803
3673619	`ref`	`gcc -funroll-loops -m32 -march=barcelona -O2 -fomit-frame-pointer`	20190810	20190803
3676625	`ref`	`gcc -funroll-loops -m32 -march=barcelona -Os -fomit-frame-pointer`	20190810	20190803
3717881	`ref`	`gcc -funroll-loops -m32 -march=k8 -O3 -fomit-frame-pointer`	20190810	20190803
3751873	`ref`	`gcc -m32 -march=k8 -O3 -fomit-frame-pointer`	20190810	20190803
3822411	`ref`	`gcc -m32 -march=k8 -Os -fomit-frame-pointer`	20190810	20190803
3835565	`ref`	`gcc -funroll-loops -m32 -march=k8 -Os -fomit-frame-pointer`	20190810	20190803
3852036	`ref`	`gcc -m32 -march=barcelona -O2 -fomit-frame-pointer`	20190810	20190803
3884782	`ref`	`gcc -m32 -march=k8 -O2 -fomit-frame-pointer`	20190810	20190803

Compiler output

Implementation: crypto_kem/kyber1024/avx2
Compiler: gcc -funroll-loops -m32 -O2 -fomit-frame-pointer

Number of similar (compiler,implementation) pairs: 156, namely:

Compiler	Implementations
gcc -funroll-loops -m32 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=athlon -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=athlon -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=athlon -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=athlon -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=barcelona -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=barcelona -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=barcelona -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=barcelona -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=i386 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=i386 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=i386 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=i386 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=i486 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=i486 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=i486 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=i486 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6-2 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6-2 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6-2 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6-2 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6-3 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6-3 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6-3 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6-3 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k6 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k8 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k8 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k8 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=k8 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=nocona -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=nocona -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=nocona -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=nocona -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium-m -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium-m -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium-m -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium-m -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium-mmx -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium-mmx -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium-mmx -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium-mmx -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium2 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium2 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium2 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium2 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium3 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium3 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium3 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium3 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium4 -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium4 -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium4 -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium4 -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentium -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentiumpro -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentiumpro -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentiumpro -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=pentiumpro -Os -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=prescott -O2 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=prescott -O3 -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=prescott -O -fomit-frame-pointer	avx2
gcc -funroll-loops -m32 -march=prescott -Os -fomit-frame-pointer	avx2
gcc -m32 -O2 -fomit-frame-pointer	avx2
gcc -m32 -O3 -fomit-frame-pointer	avx2
gcc -m32 -O -fomit-frame-pointer	avx2
gcc -m32 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=athlon -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=athlon -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=athlon -O -fomit-frame-pointer	avx2
gcc -m32 -march=athlon -Os -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -O -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -msse4.1 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -msse4.1 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -msse4.1 -O -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -msse4.1 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -msse4 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -msse4 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -msse4 -O -fomit-frame-pointer	avx2
gcc -m32 -march=core2 -msse4 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=corei7 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=corei7 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=corei7 -O -fomit-frame-pointer	avx2
gcc -m32 -march=corei7 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=i386 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=i386 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=i386 -O -fomit-frame-pointer	avx2
gcc -m32 -march=i386 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=i486 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=i486 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=i486 -O -fomit-frame-pointer	avx2
gcc -m32 -march=i486 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=k6-2 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=k6-2 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=k6-2 -O -fomit-frame-pointer	avx2
gcc -m32 -march=k6-2 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=k6-3 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=k6-3 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=k6-3 -O -fomit-frame-pointer	avx2
gcc -m32 -march=k6-3 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=k6 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=k6 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=k6 -O -fomit-frame-pointer	avx2
gcc -m32 -march=k6 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=k8 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=k8 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=k8 -O -fomit-frame-pointer	avx2
gcc -m32 -march=k8 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=nocona -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=nocona -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=nocona -O -fomit-frame-pointer	avx2
gcc -m32 -march=nocona -Os -fomit-frame-pointer	avx2
gcc -m32 -march=pentium-m -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium-m -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium-m -O -fomit-frame-pointer	avx2
gcc -m32 -march=pentium-m -Os -fomit-frame-pointer	avx2
gcc -m32 -march=pentium-mmx -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium-mmx -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium-mmx -O -fomit-frame-pointer	avx2
gcc -m32 -march=pentium-mmx -Os -fomit-frame-pointer	avx2
gcc -m32 -march=pentium2 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium2 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium2 -O -fomit-frame-pointer	avx2
gcc -m32 -march=pentium2 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=pentium3 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium3 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium3 -O -fomit-frame-pointer	avx2
gcc -m32 -march=pentium3 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=pentium4 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium4 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium4 -O -fomit-frame-pointer	avx2
gcc -m32 -march=pentium4 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=pentium -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=pentium -O -fomit-frame-pointer	avx2
gcc -m32 -march=pentium -Os -fomit-frame-pointer	avx2
gcc -m32 -march=pentiumpro -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=pentiumpro -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=pentiumpro -O -fomit-frame-pointer	avx2
gcc -m32 -march=pentiumpro -Os -fomit-frame-pointer	avx2
gcc -m32 -march=prescott -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=prescott -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=prescott -O -fomit-frame-pointer	avx2
gcc -m32 -march=prescott -Os -fomit-frame-pointer	avx2

Compiler output

Implementation: crypto_kem/kyber1024/avx2
Compiler: gcc -m32 -march=barcelona -O2 -fomit-frame-pointer

KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c: In function 'KeccakP1600times4_AddLanesAll':
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:135:40: warning: AVX vector return without AVX enabled changes the ABI [-Wpsabi]
KeccakP-1600-times4-SIMD256.c: #define Xor_In4( argIndex ) lanes0 = LOAD256u( curData0[argIndex]),\
KeccakP-1600-times4-SIMD256.c: ^
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:146:9: note: in expansion of macro 'Xor_In4'
KeccakP-1600-times4-SIMD256.c: Xor_In4( 0 );
KeccakP-1600-times4-SIMD256.c: ^~~~~~~
KeccakP-1600-times4-SIMD256.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:43:0,
KeccakP-1600-times4-SIMD256.c: from KeccakP-1600-times4-SIMD256.c:21:
KeccakP-1600-times4-SIMD256.c: /usr/lib/gcc/x86_64-linux-gnu/7/include/avx2intrin.h:913:1: error: inlining failed in call to always_inline '_mm256_xor_si256': target specific option mismatch
KeccakP-1600-times4-SIMD256.c: _mm256_xor_si256 (__m256i __A, __m256i __B)
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~~~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:55:41: note: called from here
KeccakP-1600-times4-SIMD256.c: #define XOReq256(a, b) a = _mm256_xor_si256(a, b)
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~~~~~~~~~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:143:33: note: in expansion of macro 'XOReq256'
KeccakP-1600-times4-SIMD256.c: XOReq256( stateAsLanes[argIndex+3], lanes3 )
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:149:9: note: in expansion of macro 'Xor_In4'
KeccakP-1600-times4-SIMD256.c: Xor_In4( 12 );
KeccakP-1600-times4-SIMD256.c: ^~~~~~~
KeccakP-1600-times4-SIMD256.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:43:0,
KeccakP-1600-times4-SIMD256.c: from KeccakP-1600-times4-SIMD256.c:21:
KeccakP-1600-times4-SIMD256.c: /usr/lib/gcc/x86_64-linux-gnu/7/include/avx2intrin.h:913:1: error: inlining failed in call to always_inline '_mm256_xor_si256': target specific option mismatch
KeccakP-1600-times4-SIMD256.c: _mm256_xor_si256 (__m256i __A, __m256i __B)
KeccakP-1600-times4-SIMD256.c: ...
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c: In function 'KeccakP1600times4_AddLanesAll':
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:135:40: warning: AVX vector return without AVX enabled changes the ABI [-Wpsabi]
KeccakP-1600-times4-SIMD256.c: #define Xor_In4( argIndex ) lanes0 = LOAD256u( curData0[argIndex]),\
KeccakP-1600-times4-SIMD256.c: ^
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:146:9: note: in expansion of macro 'Xor_In4'
KeccakP-1600-times4-SIMD256.c: Xor_In4( 0 );
KeccakP-1600-times4-SIMD256.c: ^~~~~~~
KeccakP-1600-times4-SIMD256.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:43:0,
KeccakP-1600-times4-SIMD256.c: from KeccakP-1600-times4-SIMD256.c:21:
KeccakP-1600-times4-SIMD256.c: /usr/lib/gcc/x86_64-linux-gnu/7/include/avx2intrin.h:913:1: error: inlining failed in call to always_inline '_mm256_xor_si256': target specific option mismatch
KeccakP-1600-times4-SIMD256.c: _mm256_xor_si256 (__m256i __A, __m256i __B)
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~~~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:55:41: note: called from here
KeccakP-1600-times4-SIMD256.c: #define XOReq256(a, b) a = _mm256_xor_si256(a, b)
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~~~~~~~~~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:143:33: note: in expansion of macro 'XOReq256'
KeccakP-1600-times4-SIMD256.c: XOReq256( stateAsLanes[argIndex+3], lanes3 )
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:149:9: note: in expansion of macro 'Xor_In4'
KeccakP-1600-times4-SIMD256.c: Xor_In4( 12 );
KeccakP-1600-times4-SIMD256.c: ^~~~~~~
KeccakP-1600-times4-SIMD256.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:43:0,
KeccakP-1600-times4-SIMD256.c: from KeccakP-1600-times4-SIMD256.c:21:
KeccakP-1600-times4-SIMD256.c: /usr/lib/gcc/x86_64-linux-gnu/7/include/avx2intrin.h:913:1: error: inlining failed in call to always_inline '_mm256_xor_si256': target specific option mismatch
KeccakP-1600-times4-SIMD256.c: _mm256_xor_si256 (__m256i __A, __m256i __B)
KeccakP-1600-times4-SIMD256.c: ...

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

Compiler	Implementations
gcc -m32 -march=barcelona -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=barcelona -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=barcelona -O -fomit-frame-pointer	avx2
gcc -m32 -march=barcelona -Os -fomit-frame-pointer	avx2

Compiler output

Implementation: crypto_kem/kyber1024/avx2
Compiler: gcc -m32 -march=core-avx-i -O2 -fomit-frame-pointer

KeccakP-1600-times4-SIMD256.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:43:0,
KeccakP-1600-times4-SIMD256.c: from KeccakP-1600-times4-SIMD256.c:21:
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c: In function 'KeccakP1600times4_AddLanesAll':
KeccakP-1600-times4-SIMD256.c: /usr/lib/gcc/x86_64-linux-gnu/7/include/avx2intrin.h:913:1: error: inlining failed in call to always_inline '_mm256_xor_si256': target specific option mismatch
KeccakP-1600-times4-SIMD256.c: _mm256_xor_si256 (__m256i __A, __m256i __B)
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~~~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:55:41: note: called from here
KeccakP-1600-times4-SIMD256.c: #define XOReq256(a, b) a = _mm256_xor_si256(a, b)
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~~~~~~~~~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:143:33: note: in expansion of macro 'XOReq256'
KeccakP-1600-times4-SIMD256.c: XOReq256( stateAsLanes[argIndex+3], lanes3 )
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:149:9: note: in expansion of macro 'Xor_In4'
KeccakP-1600-times4-SIMD256.c: Xor_In4( 12 );
KeccakP-1600-times4-SIMD256.c: ^~~~~~~
KeccakP-1600-times4-SIMD256.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/7/include/immintrin.h:43:0,
KeccakP-1600-times4-SIMD256.c: from KeccakP-1600-times4-SIMD256.c:21:
KeccakP-1600-times4-SIMD256.c: /usr/lib/gcc/x86_64-linux-gnu/7/include/avx2intrin.h:913:1: error: inlining failed in call to always_inline '_mm256_xor_si256': target specific option mismatch
KeccakP-1600-times4-SIMD256.c: _mm256_xor_si256 (__m256i __A, __m256i __B)
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~~~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:55:41: note: called from here
KeccakP-1600-times4-SIMD256.c: #define XOReq256(a, b) a = _mm256_xor_si256(a, b)
KeccakP-1600-times4-SIMD256.c: ^~~~~~~~~~~~~~~~~~~~~~
KeccakP-1600-times4-SIMD256.c: KeccakP-1600-times4-SIMD256.c:142:33: note: in expansion of macro 'XOReq256'
KeccakP-1600-times4-SIMD256.c: XOReq256( stateAsLanes[argIndex+2], lanes2 ),\
KeccakP-1600-times4-SIMD256.c: ...

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

Compiler	Implementations
gcc -m32 -march=core-avx-i -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=core-avx-i -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=core-avx-i -O -fomit-frame-pointer	avx2
gcc -m32 -march=core-avx-i -Os -fomit-frame-pointer	avx2
gcc -m32 -march=corei7-avx -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=corei7-avx -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=corei7-avx -O -fomit-frame-pointer	avx2
gcc -m32 -march=corei7-avx -Os -fomit-frame-pointer	avx2

Compiler output

Implementation: crypto_kem/kyber1024/avx2
Compiler: gcc -m32 -march=core-avx2 -O2 -fomit-frame-pointer

basemul.S: basemul.S: Assembler messages:
basemul.S: basemul.S:79: Error: bad register name `%rip)'
basemul.S: basemul.S:80: Error: bad register name `%rip)'
basemul.S: basemul.S:81: Error: bad register name `%rcx)'
basemul.S: basemul.S:84: Error: bad register name `%rsi)'
basemul.S: basemul.S:84: Error: bad register name `%rdx)'
basemul.S: basemul.S:84: Error: bad register name `%rsi)'
basemul.S: basemul.S:84: Error: bad register name `%rdx)'
basemul.S: basemul.S:84: Error: bad register name `%ymm8'
basemul.S: basemul.S:84: Error: bad register name `%ymm8'
basemul.S: basemul.S:84: Error: bad register name `%ymm10'
basemul.S: basemul.S:84: Error: bad register name `%ymm10'
basemul.S: basemul.S:84: Error: bad register name `%ymm9'
basemul.S: basemul.S:84: Error: bad register name `%ymm9'
basemul.S: basemul.S:84: Error: bad register name `%ymm9'
basemul.S: basemul.S:84: Error: bad register name `%ymm9'
basemul.S: basemul.S:84: Error: bad register name `%ymm11'
basemul.S: basemul.S:84: Error: bad register name `%ymm11'
basemul.S: basemul.S:84: Error: bad register name `%ymm11'
basemul.S: basemul.S:84: Error: bad register name `%ymm11'
basemul.S: basemul.S:84: Error: bad register name `%ymm11'
basemul.S: basemul.S:84: Error: bad register name `%ymm13'
basemul.S: basemul.S:84: Error: bad register name `%ymm13'
basemul.S: basemul.S:84: Error: bad register name `%ymm8'
basemul.S: basemul.S:84: Error: bad register name `%ymm8'
basemul.S: ...

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

Compiler	Implementations
gcc -m32 -march=core-avx2 -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=core-avx2 -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=core-avx2 -O -fomit-frame-pointer	avx2
gcc -m32 -march=core-avx2 -Os -fomit-frame-pointer	avx2
gcc -m32 -march=native -mtune=native -O2 -fomit-frame-pointer	avx2
gcc -m32 -march=native -mtune=native -O3 -fomit-frame-pointer	avx2
gcc -m32 -march=native -mtune=native -O -fomit-frame-pointer	avx2
gcc -m32 -march=native -mtune=native -Os -fomit-frame-pointer	avx2