Implementation notes: x86, thoth, crypto_hash/fugue256

Computer: thoth
Architecture: x86
CPU ID: AuthenticAMD-00000622-0183f9ff
SUPERCOP version: 20160806
Operation: crypto_hash
Primitive: fugue256

Time	Implementation	Compiler	Benchmark date	SUPERCOP version
70435	`sphlib`	`gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv`	20160726	20160724
75150	`ANSI_opt32`	`gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv`	20160726	20160724
76253	`ANSI_opt32`	`gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv`	20160726	20160724
80059	`sphlib`	`gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv`	20160726	20160724
80134	`sphlib`	`gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv`	20160726	20160724
82530	`ANSI_opt32`	`gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv`	20160726	20160724
83203	`sphlib`	`gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv`	20160726	20160724
83311	`ANSI_opt32`	`gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv`	20160726	20160724
94694	`sphlib`	`clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments`	20160726	20160724
94747	`ANSI_opt32`	`clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments`	20160726	20160724
115839	`ANSI_opt64`	`gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv`	20160726	20160724
116646	`ANSI_opt64`	`gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv`	20160726	20160724
124533	`ANSI_opt64`	`gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv`	20160726	20160724
128592	`ANSI_opt64`	`gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv`	20160726	20160724
128904	`ANSI_opt64`	`clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments`	20160726	20160724

Compiler output

Implementation: crypto_hash/fugue256/SSSE3
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

hash.c: In file included from hash.c:22:
hash.c: In file included from ./fugue_256_ssse3_code.h:15:
hash.c: ./aestab.h:53:93: warning: backslash and newline separated by space [-Wbackslash-newline-escape]
hash.c: w(r,0x8c), w(r,0xa1), w(r,0x89), w(r,0x0d), w(r,0xbf), w(r,0xe6), w(r,0x42), w(r,0x68),\
hash.c: ^
hash.c: In file included from hash.c:22:
hash.c: ./fugue_256_ssse3_code.h:111:24: warning: backslash and newline separated by space [-Wbackslash-newline-escape]
hash.c: t0 = xmm_out ; \
hash.c: ^
hash.c: hash.c:67:27: warning: signed shift result (0x6671135F000000) requires 56 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xe952bdde),HO2BE_4(0x6671135f),HO2BE_4(0xe0d4f668),HO2BE_4(0xd2b0b594),
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:20:26: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~~
hash.c: hash.c:67:27: warning: signed shift result (0x6671135F00) requires 40 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xe952bdde),HO2BE_4(0x6671135f),HO2BE_4(0xe0d4f668),HO2BE_4(0xd2b0b594),
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:20:36: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~
hash.c: hash.c:68:67: warning: signed shift result (0x34F8C248000000) requires 55 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xf96c621d),HO2BE_4(0xfbf929de),HO2BE_4(0x9149e899),HO2BE_4(0x34f8c248),
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:20:26: note: expanded from macro 'HO2BE_4'
hash.c: ...

Number of similar (compiler,implementation) pairs: 1, namely:

Compiler	Implementations
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments	SSSE3

Compiler output

Implementation: crypto_hash/fugue256/SSE4.1
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

hash.c: In file included from hash.c:24:
hash.c: In file included from ./fugue_256_sse4_code.h:15:
hash.c: ./aestab.h:53:93: warning: backslash and newline separated by space [-Wbackslash-newline-escape]
hash.c: w(r,0x8c), w(r,0xa1), w(r,0x89), w(r,0x0d), w(r,0xbf), w(r,0xe6), w(r,0x42), w(r,0x68),\
hash.c: ^
hash.c: hash.c:69:27: warning: signed shift result (0x6671135F000000) requires 56 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xe952bdde),HO2BE_4(0x6671135f),HO2BE_4(0xe0d4f668),HO2BE_4(0xd2b0b594),
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:20:26: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~~
hash.c: hash.c:69:27: warning: signed shift result (0x6671135F00) requires 40 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xe952bdde),HO2BE_4(0x6671135f),HO2BE_4(0xe0d4f668),HO2BE_4(0xd2b0b594),
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:20:36: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~
hash.c: hash.c:70:67: warning: signed shift result (0x34F8C248000000) requires 55 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xf96c621d),HO2BE_4(0xfbf929de),HO2BE_4(0x9149e899),HO2BE_4(0x34f8c248),
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:20:26: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~~
hash.c: hash.c:70:67: warning: signed shift result (0x34F8C24800) requires 39 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xf96c621d),HO2BE_4(0xfbf929de),HO2BE_4(0x9149e899),HO2BE_4(0x34f8c248),
hash.c: ...

Number of similar (compiler,implementation) pairs: 1, namely:

Compiler	Implementations
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments	SSE4.1

Compiler output

Implementation: crypto_hash/fugue256/ANSI_opt32
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

hash.c: hash.c:32:27: warning: signed shift result (0x6671135F000000) requires 56 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xe952bdde),HO2BE_4(0x6671135f),HO2BE_4(0xe0d4f668),HO2BE_4(0xd2b0b594),
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:19:26: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~~
hash.c: hash.c:32:27: warning: signed shift result (0x6671135F00) requires 40 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xe952bdde),HO2BE_4(0x6671135f),HO2BE_4(0xe0d4f668),HO2BE_4(0xd2b0b594),
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:19:36: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~
hash.c: hash.c:33:67: warning: signed shift result (0x34F8C248000000) requires 55 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xf96c621d),HO2BE_4(0xfbf929de),HO2BE_4(0x9149e899),HO2BE_4(0x34f8c248)};
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:19:26: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~~
hash.c: hash.c:33:67: warning: signed shift result (0x34F8C24800) requires 39 bits to represent, but 'int' only has 32 bits [-Wshift-overflow]
hash.c: HO2BE_4(0xf96c621d),HO2BE_4(0xfbf929de),HO2BE_4(0x9149e899),HO2BE_4(0x34f8c248)};
hash.c: ^~~~~~~~~~~~~~~~~~~
hash.c: ./aestab_t.h:19:36: note: expanded from macro 'HO2BE_4'
hash.c: #define HO2BE_4(_x) ((_xgt;>gt;8)&0xff00)|(_x>gt;>gt;24))
hash.c: ~~^ ~
hash.c: 4 warnings generated.

Number of similar (compiler,implementation) pairs: 2, namely:

Compiler	Implementations
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments	ANSI_opt32 ANSI_opt64

Compiler output

Implementation: crypto_hash/fugue256/ccalik/vperm
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

hash.c: hash.c:447:4: error: always_inline function '_mm_setzero_si128' requires target feature 'sse2', but would be inlined into function 'Compress256' that is compiled without support for 'sse2'
hash.c: TIX256(pmsg, ctx->gt;state[1], ctx->gt;state[0], ctx->gt;state[6], ctx->gt;state[8], _t0, _t1, _t2);
hash.c: ^
hash.c: hash.c:165:7: note: expanded from macro 'TIX256'
hash.c: t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
hash.c: ^
hash.c: /usr/lib/llvm-3.8/bin/../lib/clang/3.8.0/include/emmintrin.h:1337:44: note: expanded from macro '_mm_shuffle_epi32'
hash.c: (__v4si)_mm_setzero_si128(), \
hash.c: ^
hash.c: hash.c:447:4: error: always_inline function '_mm_xor_si128' requires target feature 'sse2', but would be inlined into function 'Compress256' that is compiled without support for 'sse2'
hash.c: hash.c:166:8: note: expanded from macro 'TIX256'
hash.c: s10 = _mm_xor_si128(s10, t1);\
hash.c: ^
hash.c: hash.c:447:4: error: always_inline function '_mm_castps_si128' requires target feature 'sse2', but would be inlined into function 'Compress256' that is compiled without support for 'sse2'
hash.c: hash.c:167:7: note: expanded from macro 'TIX256'
hash.c: t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
hash.c: ^
hash.c: hash.c:447:4: error: always_inline function '_mm_load_ss' requires target feature 'sse', but would be inlined into function 'Compress256' that is compiled without support for 'sse'
hash.c: hash.c:167:24: note: expanded from macro 'TIX256'
hash.c: t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
hash.c: ^
hash.c: hash.c:447:4: error: always_inline function '_mm_andnot_si128' requires target feature 'sse2', but would be inlined into function 'Compress256' that is compiled without support for 'sse2'
hash.c: hash.c:168:2: note: expanded from macro 'TIX256'
hash.c: TRANSFORM(t1, _k_ipt, t2, t3);\
hash.c: ^
hash.c: ...

Number of similar (compiler,implementation) pairs: 1, namely:

Compiler	Implementations
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments	ccalik/vperm

Compiler output

Implementation: crypto_hash/fugue256/ccalik/aesni
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

hash.c: hash.c:448:4: error: always_inline function '_mm_setzero_si128' requires target feature 'sse2', but would be inlined into function 'Compress256' that is compiled without support for 'sse2'
hash.c: TIX256(pmsg, ctx->gt;state[1], ctx->gt;state[0], ctx->gt;state[6], ctx->gt;state[8], _t0, _t1, _t2);
hash.c: ^
hash.c: hash.c:156:7: note: expanded from macro 'TIX256'
hash.c: t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
hash.c: ^
hash.c: /usr/lib/llvm-3.8/bin/../lib/clang/3.8.0/include/emmintrin.h:1337:44: note: expanded from macro '_mm_shuffle_epi32'
hash.c: (__v4si)_mm_setzero_si128(), \
hash.c: ^
hash.c: hash.c:448:4: error: always_inline function '_mm_xor_si128' requires target feature 'sse2', but would be inlined into function 'Compress256' that is compiled without support for 'sse2'
hash.c: hash.c:157:8: note: expanded from macro 'TIX256'
hash.c: s10 = _mm_xor_si128(s10, t1);\
hash.c: ^
hash.c: hash.c:448:4: error: always_inline function '_mm_castps_si128' requires target feature 'sse2', but would be inlined into function 'Compress256' that is compiled without support for 'sse2'
hash.c: hash.c:158:7: note: expanded from macro 'TIX256'
hash.c: t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
hash.c: ^
hash.c: hash.c:448:4: error: always_inline function '_mm_load_ss' requires target feature 'sse', but would be inlined into function 'Compress256' that is compiled without support for 'sse'
hash.c: hash.c:158:24: note: expanded from macro 'TIX256'
hash.c: t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
hash.c: ^
hash.c: hash.c:448:4: error: always_inline function '_mm_castps_si128' requires target feature 'sse2', but would be inlined into function 'Compress256' that is compiled without support for 'sse2'
hash.c: hash.c:159:7: note: expanded from macro 'TIX256'
hash.c: s0 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s0), _mm_castsi128_ps(t1)));\
hash.c: ^
hash.c: ...

Number of similar (compiler,implementation) pairs: 1, namely:

Compiler	Implementations
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments	ccalik/aesni

Compiler output

Implementation: crypto_hash/fugue256/SSE4.1
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv

hash.c: In file included from fugue_256_sse4_code.h:15:0,
hash.c: from hash.c:24:
hash.c: aestab.h:53:92: warning: backslash and newline separated by space
hash.c: w(r,0x8c), w(r,0xa1), w(r,0x89), w(r,0x0d), w(r,0xbf), w(r,0xe6), w(r,0x42), w(r,0x68),\
hash.c: ^
hash.c: hash.c: In function 'crypto_hash_fugue256_SSE4_1':
hash.c: hash.c:102:22: warning: SSE vector return without SSE enabled changes the ABI [-Wpsabi]
hash.c: row_shuffle_mask = _mm_set_epi8(15,11,7,3,
hash.c: ^
hash.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/pmmintrin.h:31:0,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/tmmintrin.h:31,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/smmintrin.h:32,
hash.c: from hash.c:23:
hash.c: /usr/lib/gcc/i686-linux-gnu/5/include/emmintrin.h:753:1: error: inlining failed in call to always_inline '_mm_setzero_si128': target specific option mismatch
hash.c: _mm_setzero_si128 (void)
hash.c: ^
hash.c: hash.c:110:15: error: called from here
hash.c: xmm_out = _mm_setzero_si128();
hash.c: ^
hash.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/pmmintrin.h:31:0,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/tmmintrin.h:31,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/smmintrin.h:32,
hash.c: from hash.c:23:
hash.c: /usr/lib/gcc/i686-linux-gnu/5/include/emmintrin.h:597:1: error: inlining failed in call to always_inline '_mm_set_epi32': target specific option mismatch
hash.c: _mm_set_epi32 (int __q3, int __q2, int __q1, int __q0)
hash.c: ...

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

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv	SSE4.1
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv	SSE4.1
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv	SSE4.1
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv	SSE4.1

Compiler output

Implementation: crypto_hash/fugue256/SSSE3
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv

hash.c: In file included from fugue_256_ssse3_code.h:15:0,
hash.c: from hash.c:22:
hash.c: aestab.h:53:92: warning: backslash and newline separated by space
hash.c: w(r,0x8c), w(r,0xa1), w(r,0x89), w(r,0x0d), w(r,0xbf), w(r,0xe6), w(r,0x42), w(r,0x68),\
hash.c: ^
hash.c: In file included from hash.c:22:0:
hash.c: fugue_256_ssse3_code.h:111:23: warning: backslash and newline separated by space
hash.c: t0 = xmm_out ; \
hash.c: ^
hash.c: hash.c: In function 'crypto_hash_fugue256_SSSE3':
hash.c: hash.c:100:22: warning: SSE vector return without SSE enabled changes the ABI [-Wpsabi]
hash.c: row_shuffle_mask = _mm_set_epi8(15,11,7,3,
hash.c: ^
hash.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/pmmintrin.h:31:0,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/tmmintrin.h:31,
hash.c: from hash.c:21:
hash.c: /usr/lib/gcc/i686-linux-gnu/5/include/emmintrin.h:753:1: error: inlining failed in call to always_inline '_mm_setzero_si128': target specific option mismatch
hash.c: _mm_setzero_si128 (void)
hash.c: ^
hash.c: hash.c:108:15: error: called from here
hash.c: xmm_out = _mm_setzero_si128();
hash.c: ^
hash.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/pmmintrin.h:31:0,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/tmmintrin.h:31,
hash.c: from hash.c:21:
hash.c: ...

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

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv	SSSE3
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv	SSSE3
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv	SSSE3
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv	SSSE3

Compiler output

Implementation: crypto_hash/fugue256/ccalik/aesni
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv

hash.c: hash.c: In function 'Compress256':
hash.c: hash.c:156:5: warning: SSE vector return without SSE enabled changes the ABI [-Wpsabi]
hash.c: t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
hash.c: ^
hash.c: hash.c:448:4: note: in expansion of macro 'TIX256'
hash.c: TIX256(pmsg, ctx->gt;state[1], ctx->gt;state[0], ctx->gt;state[6], ctx->gt;state[8], _t0, _t1, _t2);
hash.c: ^
hash.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/pmmintrin.h:31:0,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/tmmintrin.h:31,
hash.c: from hash_api.h:29,
hash.c: from hash.c:20:
hash.c: /usr/lib/gcc/i686-linux-gnu/5/include/emmintrin.h:1268:1: error: inlining failed in call to always_inline '_mm_and_si128': target specific option mismatch
hash.c: _mm_and_si128 (__m128i __A, __m128i __B)
hash.c: ^
hash.c: hash.c:139:7: error: called from here
hash.c: s0 = _mm_and_si128(s0, M128(_maskd3n))
hash.c: ^
hash.c: hash.c:349:2: note: in expansion of macro 'UNPACK_S0'
hash.c: UNPACK_S0(s0, s1, _t3)
hash.c: ^
hash.c: hash.c:449:4: note: in expansion of macro 'SUBROUND256_2'
hash.c: SUBROUND256_2(ctx->gt;state[8], ctx->gt;state[9], ctx->gt;state[7], ctx->gt;state[2], ctx->gt;state[7], ctx->gt;state[8], ctx->gt;state[6], ctx->gt;state[1]);
hash.c: ^
hash.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/pmmintrin.h:31:0,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/tmmintrin.h:31,
hash.c: ...

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

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv	ccalik/aesni
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv	ccalik/aesni
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv	ccalik/aesni
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv	ccalik/aesni

Compiler output

Implementation: crypto_hash/fugue256/ccalik/vperm
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv

hash.c: hash.c: In function 'Compress256':
hash.c: hash.c:165:5: warning: SSE vector return without SSE enabled changes the ABI [-Wpsabi]
hash.c: t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
hash.c: ^
hash.c: hash.c:447:4: note: in expansion of macro 'TIX256'
hash.c: TIX256(pmsg, ctx->gt;state[1], ctx->gt;state[0], ctx->gt;state[6], ctx->gt;state[8], _t0, _t1, _t2);
hash.c: ^
hash.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/pmmintrin.h:31:0,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/tmmintrin.h:31,
hash.c: from hash_api.h:29,
hash.c: from hash.c:20:
hash.c: /usr/lib/gcc/i686-linux-gnu/5/include/emmintrin.h:1268:1: error: inlining failed in call to always_inline '_mm_and_si128': target specific option mismatch
hash.c: _mm_and_si128 (__m128i __A, __m128i __B)
hash.c: ^
hash.c: hash.c:138:7: error: called from here
hash.c: s0 = _mm_and_si128(s0, M128(_maskd3n))
hash.c: ^
hash.c: hash.c:348:2: note: in expansion of macro 'UNPACK_S0'
hash.c: UNPACK_S0(s0, s1, _t3)
hash.c: ^
hash.c: hash.c:448:4: note: in expansion of macro 'SUBROUND256_2'
hash.c: SUBROUND256_2(ctx->gt;state[8], ctx->gt;state[9], ctx->gt;state[7], ctx->gt;state[2], ctx->gt;state[7], ctx->gt;state[8], ctx->gt;state[6], ctx->gt;state[1]);
hash.c: ^
hash.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/pmmintrin.h:31:0,
hash.c: from /usr/lib/gcc/i686-linux-gnu/5/include/tmmintrin.h:31,
hash.c: ...

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

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv	ccalik/vperm
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv	ccalik/vperm
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv	ccalik/vperm
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv	ccalik/vperm