Implementation notes: x86, thoth, crypto_aead/morus1280128v1

Computer: thoth
Architecture: x86
CPU ID: AuthenticAMD-00000622-0183f9ff
SUPERCOP version: 20160806
Operation: crypto_aead
Primitive: morus1280128v1

Time	Implementation	Compiler	Benchmark date	SUPERCOP version
86227	`ref`	`gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv`	20160727	20160724
86466	`ref64`	`gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv`	20160727	20160724
86579	`ref`	`gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv`	20160727	20160724
88883	`ref`	`gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv`	20160727	20160724
96488	`ref`	`clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments`	20160727	20160724
101085	`ref64`	`gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv`	20160727	20160724
101182	`ref64`	`clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments`	20160727	20160724
103757	`ref64`	`gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv`	20160727	20160724
105044	`ref`	`gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv`	20160727	20160724
105209	`ref64`	`gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv`	20160727	20160724

Compiler output

Implementation: crypto_aead/morus1280128v1/avx2
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

morus1280128v1avx2.c: morus1280128v1avx2.c:52:13: error: always_inline function '_mm256_xor_si256' requires target feature 'sse3', but would be inlined into function 'morus_stateupdate' that is compiled without support for 'sse3'
morus1280128v1avx2.c: state[0] = XOR256(state[0], state[3]);
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:22:27: note: expanded from macro 'XOR256'
morus1280128v1avx2.c: #define XOR256(x,y) _mm256_xor_si256((x),(y)) /*XOR256(x,y) = x ^ y, where x and y are two 256-bit word*/
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:53:13: error: always_inline function '_mm256_xor_si256' requires target feature 'sse3', but would be inlined into function 'morus_stateupdate' that is compiled without support for 'sse3'
morus1280128v1avx2.c: state[0] = XOR256(state[0], AND256(state[1], state[2]));
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:22:27: note: expanded from macro 'XOR256'
morus1280128v1avx2.c: #define XOR256(x,y) _mm256_xor_si256((x),(y)) /*XOR256(x,y) = x ^ y, where x and y are two 256-bit word*/
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:53:30: error: always_inline function '_mm256_and_si256' requires target feature 'sse3', but would be inlined into function 'morus_stateupdate' that is compiled without support for 'sse3'
morus1280128v1avx2.c: state[0] = XOR256(state[0], AND256(state[1], state[2]));
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:23:27: note: expanded from macro 'AND256'
morus1280128v1avx2.c: #define AND256(x,y) _mm256_and_si256((x),(y)) /*AND(x,y) = x & y, where x and y are two 256-bit word*/
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:54:13: error: always_inline function '_mm256_or_si256' requires target feature 'sse3', but would be inlined into function 'morus_stateupdate' that is compiled without support for 'sse3'
morus1280128v1avx2.c: state[0] = ROTL256(state[0], n1);
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:36:27: note: expanded from macro 'ROTL256'
morus1280128v1avx2.c: #define ROTL256(x,n) OR256( _mm256_slli_epi64((x), (n)), _mm256_srli_epi64((x),(64-n)) ) /*Rotate 4 64-bit unsigned integers in x to the left by n-bit positions*/
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:25:27: note: expanded from macro 'OR256'
morus1280128v1avx2.c: ...

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

Compiler	Implementations
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments	avx2

Compiler output

Implementation: crypto_aead/morus1280128v1/sse2
Compiler: clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

morus1280128v1sse2.c: morus1280128v1sse2.c:87:11: error: always_inline function '_mm_setzero_si128' requires target feature 'sse2', but would be inlined into function 'morus_initialization' that is compiled without support for 'sse2'
morus1280128v1sse2.c: tmp[0] = SETZERO();
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:22:24: note: expanded from macro 'SETZERO'
morus1280128v1sse2.c: #define SETZERO() _mm_setzero_si128() /*set the value of 128-bit register to zero*/
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:88:11: error: always_inline function '_mm_setzero_si128' requires target feature 'sse2', but would be inlined into function 'morus_initialization' that is compiled without support for 'sse2'
morus1280128v1sse2.c: tmp[1] = SETZERO();
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:22:24: note: expanded from macro 'SETZERO'
morus1280128v1sse2.c: #define SETZERO() _mm_setzero_si128() /*set the value of 128-bit register to zero*/
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:90:16: error: always_inline function '_mm_load_si128' requires target feature 'sse2', but would be inlined into function 'morus_initialization' that is compiled without support for 'sse2'
morus1280128v1sse2.c: state[0][0] = _mm_load_si128((__m128i*)iv); state[0][1] = SETZERO();
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:90:60: error: always_inline function '_mm_setzero_si128' requires target feature 'sse2', but would be inlined into function 'morus_initialization' that is compiled without support for 'sse2'
morus1280128v1sse2.c: state[0][0] = _mm_load_si128((__m128i*)iv); state[0][1] = SETZERO();
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:22:24: note: expanded from macro 'SETZERO'
morus1280128v1sse2.c: #define SETZERO() _mm_setzero_si128() /*set the value of 128-bit register to zero*/
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:91:16: error: always_inline function '_mm_load_si128' requires target feature 'sse2', but would be inlined into function 'morus_initialization' that is compiled without support for 'sse2'
morus1280128v1sse2.c: state[1][0] = _mm_load_si128((__m128i*)key); state[1][1] = _mm_load_si128((__m128i*)key);
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:91:61: error: always_inline function '_mm_load_si128' requires target feature 'sse2', but would be inlined into function 'morus_initialization' that is compiled without support for 'sse2'
morus1280128v1sse2.c: ...

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

Compiler	Implementations
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments	sse2

Compiler output

Implementation: crypto_aead/morus1280128v1/avx2
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv

morus1280128v1avx2.c: morus1280128v1avx2.c: In function 'morus_stateupdate':
morus1280128v1avx2.c: morus1280128v1avx2.c:52:11: warning: AVX vector return without AVX enabled changes the ABI [-Wpsabi]
morus1280128v1avx2.c: state[0] = XOR256(state[0], state[3]);
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/immintrin.h:43:0,
morus1280128v1avx2.c: from morus1280128v1avx2.c:2:
morus1280128v1avx2.c: morus1280128v1avx2.c: In function 'morus_enc_aut_step':
morus1280128v1avx2.c: /usr/lib/gcc/i686-linux-gnu/5/include/avx2intrin.h:1068:1: error: inlining failed in call to always_inline '_mm256_permute4x64_epi64': target specific option mismatch
morus1280128v1avx2.c: _mm256_permute4x64_epi64 (__m256i __X, const int __M)
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:38:27: error: called from here
morus1280128v1avx2.c: #define ROTL256_64(x) _mm256_permute4x64_epi64((x), _MM_SHUFFLE(2,1,0,3)) /*Rotate x by 64-bit positions to the left*/
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:201:13: note: in expansion of macro 'ROTL256_64'
morus1280128v1avx2.c: state[2] = ROTL256_64(state[2]);
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/immintrin.h:43:0,
morus1280128v1avx2.c: from morus1280128v1avx2.c:2:
morus1280128v1avx2.c: /usr/lib/gcc/i686-linux-gnu/5/include/avx2intrin.h:574:1: error: inlining failed in call to always_inline '_mm256_or_si256': target specific option mismatch
morus1280128v1avx2.c: _mm256_or_si256 (__m256i __A, __m256i __B)
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:25:27: error: called from here
morus1280128v1avx2.c: #define OR256(x,y) _mm256_or_si256((x),(y)) /*OR(x,y) = x | y, where x and y are two 256-bit word*/
morus1280128v1avx2.c: ^
morus1280128v1avx2.c: morus1280128v1avx2.c:36:27: note: in expansion of macro 'OR256'
morus1280128v1avx2.c: ...

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

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

Compiler output

Implementation: crypto_aead/morus1280128v1/sse2
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv

morus1280128v1sse2.c: morus1280128v1sse2.c: In function 'morus_initialization':
morus1280128v1sse2.c: morus1280128v1sse2.c:87:9: warning: SSE vector return without SSE enabled changes the ABI [-Wpsabi]
morus1280128v1sse2.c: tmp[0] = SETZERO();
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/xmmintrin.h:1249:0,
morus1280128v1sse2.c: from /usr/lib/gcc/i686-linux-gnu/5/include/immintrin.h:29,
morus1280128v1sse2.c: from morus1280128v1sse2.c:2:
morus1280128v1sse2.c: morus1280128v1sse2.c: In function 'morus_stateupdate':
morus1280128v1sse2.c: /usr/lib/gcc/i686-linux-gnu/5/include/emmintrin.h:1280:1: error: inlining failed in call to always_inline '_mm_or_si128': target specific option mismatch
morus1280128v1sse2.c: _mm_or_si128 (__m128i __A, __m128i __B)
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:21:24: error: called from here
morus1280128v1sse2.c: #define OR(x,y) _mm_or_si128((x),(y)) /*OR(x,y) = x | y, where x and y are two 128-bit word*/
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:78:19: note: in expansion of macro 'OR'
morus1280128v1sse2.c: state[2][1] = OR(SHIFTR64(t), SHIFTL64(state[2][1]));
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: In file included from /usr/lib/gcc/i686-linux-gnu/5/include/xmmintrin.h:1249:0,
morus1280128v1sse2.c: from /usr/lib/gcc/i686-linux-gnu/5/include/immintrin.h:29,
morus1280128v1sse2.c: from morus1280128v1sse2.c:2:
morus1280128v1sse2.c: /usr/lib/gcc/i686-linux-gnu/5/include/emmintrin.h:1180:1: error: inlining failed in call to always_inline '_mm_srli_si128': target specific option mismatch
morus1280128v1sse2.c: _mm_srli_si128 (__m128i __A, const int __N)
morus1280128v1sse2.c: ^
morus1280128v1sse2.c: morus1280128v1sse2.c:21:24: error: called from here
morus1280128v1sse2.c: #define OR(x,y) _mm_or_si128((x),(y)) /*OR(x,y) = x | y, where x and y are two 128-bit word*/
morus1280128v1sse2.c: ...

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

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv	sse2
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv	sse2
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv	sse2
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv	sse2