Implementation notes: x86, thoth, crypto_aead/tiaoxinv2

Computer: thoth
Architecture: x86
CPU ID: AuthenticAMD-00000622-0183f9ff
SUPERCOP version: 20160806
Operation: crypto_aead
Primitive: tiaoxinv2

Time	Implementation	Compiler	Benchmark date	SUPERCOP version
98964	`ref`	`gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv`	20160727	20160724
99002	`ref`	`gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv`	20160727	20160724
101871	`ref`	`gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv`	20160727	20160724
107439	`ref`	`gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv`	20160727	20160724
232212	`ref`	`clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments`	20160727	20160724

Compiler output

Implementation: crypto_aead/tiaoxinv2/nim
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

tiaoxin-optimized.c: tiaoxin-optimized.c:108:16: error: always_inline function '_mm_set_epi8' requires target feature 'sse2', but would be inlined into function 'tiaoxin_optimized_encrypt' that is compiled without support for 'sse2'
tiaoxin-optimized.c: __m128i Z0 = constant8( 0x42,0x8a,0x2f,0x98,0xd7,0x28,0xae,0x22,0x71,0x37,0x44,0x91,0x23,0xef,0x65,0xcd );
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:33:76: note: expanded from macro 'constant8'
tiaoxin-optimized.c: #define constant8(b15,b14,b13,b12,b11,b10,b9,b8,b7,b6,b5,b4,b3,b2,b1,b0) _mm_set_epi8 ((b0),(b1),(b2),(b3),(b4),(b5),(b6),(b7),(b8),(b9),(b10),(b11),(b12),(b13),(b14),(b15))
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:109:16: error: always_inline function '_mm_set_epi8' requires target feature 'sse2', but would be inlined into function 'tiaoxin_optimized_encrypt' that is compiled without support for 'sse2'
tiaoxin-optimized.c: __m128i Z1 = constant8( 0xb5,0xc0,0xfb,0xcf,0xec,0x4d,0x3b,0x2f,0xe9,0xb5,0xdb,0xa5,0x81,0x89,0xdb,0xbc );
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:33:76: note: expanded from macro 'constant8'
tiaoxin-optimized.c: #define constant8(b15,b14,b13,b12,b11,b10,b9,b8,b7,b6,b5,b4,b3,b2,b1,b0) _mm_set_epi8 ((b0),(b1),(b2),(b3),(b4),(b5),(b6),(b7),(b8),(b9),(b10),(b11),(b12),(b13),(b14),(b15))
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:110:20: error: always_inline function '_mm_set_epi8' requires target feature 'sse2', but would be inlined into function 'tiaoxin_optimized_encrypt' that is compiled without support for 'sse2'
tiaoxin-optimized.c: __m128i perm = constant8( 7,6,5,4,3,2,1,0,15,14,13,12,11,10,9,8 );
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:33:76: note: expanded from macro 'constant8'
tiaoxin-optimized.c: #define constant8(b15,b14,b13,b12,b11,b10,b9,b8,b7,b6,b5,b4,b3,b2,b1,b0) _mm_set_epi8 ((b0),(b1),(b2),(b3),(b4),(b5),(b6),(b7),(b8),(b9),(b10),(b11),(b12),(b13),(b14),(b15))
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:120:7: error: always_inline function '_mm_load_si128' requires target feature 'sse2', but would be inlined into function 'tiaoxin_optimized_encrypt' that is compiled without support for 'sse2'
tiaoxin-optimized.c: K = load( (__m128i *) ( k ) );
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:31:24: note: expanded from macro 'load'
tiaoxin-optimized.c: #define load(p) _mm_load_si128((__m128i *)(p))
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:121:7: error: always_inline function '_mm_load_si128' requires target feature 'sse2', but would be inlined into function 'tiaoxin_optimized_encrypt' that is compiled without support for 'sse2'
tiaoxin-optimized.c: ...

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

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

Compiler output

Implementation: crypto_aead/tiaoxinv2/ref
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

tiaoxin-tablebased.c: tiaoxin-tablebased.c:257:12: warning: comparison of unsigned expression tiaoxin-tablebased.c: if( clen tiaoxin-tablebased.c: ~~~~ ^ ~
tiaoxin-tablebased.c: 1 warning generated.

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

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

Compiler output

Implementation: crypto_aead/tiaoxinv2/nim
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv

tiaoxin-optimized.c: tiaoxin-optimized.c: In function 'tiaoxin_optimized_encrypt':
tiaoxin-optimized.c: tiaoxin-optimized.c:108:10: warning: SSE vector return without SSE enabled changes the ABI [-Wpsabi]
tiaoxin-optimized.c: __m128i Z0 = constant8( 0x42,0x8a,0x2f,0x98,0xd7,0x28,0xae,0x22,0x71,0x37,0x44,0x91,0x23,0xef,0x65,0xcd );
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/xmmintrin.h:1249:0,
tiaoxin-optimized.c: from tiaoxin-optimized.c:20:
tiaoxin-optimized.c: /usr/lib/gcc/i686-linux-gnu/5/include/emmintrin.h:1286:1: error: inlining failed in call to always_inline '_mm_xor_si128': target specific option mismatch
tiaoxin-optimized.c: _mm_xor_si128 (__m128i __A, __m128i __B)
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:28:19: error: called from here
tiaoxin-optimized.c: #define xor(a,b) _mm_xor_si128(a,b)
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:42:127: note: in expansion of macro 'xor'
tiaoxin-optimized.c: #define RT6( T , M ) tmp = T[0];T[0] = enc(T[5],M);T[5] = T[4];T[4] = T[3];T[3] = T[2];T[2] = T[1];T[1] = enc(tmp,Z0);T[0] = xor(T[0], tmp);
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:50:1: note: in expansion of macro 'RT6'
tiaoxin-optimized.c: RT6( T6, M2 );
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: tiaoxin-optimized.c:151:2: note: in expansion of macro 'UPDATE'
tiaoxin-optimized.c: UPDATE( T3 , T4 , T6, Z0 , Z1 , Z0 );
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: In file included from tiaoxin-optimized.c:22:0:
tiaoxin-optimized.c: /usr/lib/gcc/i686-linux-gnu/5/include/wmmintrin.h:61:1: error: inlining failed in call to always_inline '_mm_aesenc_si128': target specific option mismatch
tiaoxin-optimized.c: _mm_aesenc_si128 (__m128i __X, __m128i __Y)
tiaoxin-optimized.c: ^
tiaoxin-optimized.c: ...

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

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