Implementation notes: amd64, rumba5, crypto_encrypt/4hfe

Computer: rumba5
Architecture: amd64
CPU ID: AuthenticAMD-00800f11-178bfbff
SUPERCOP version: 20181216
Operation: crypto_encrypt
Primitive: 4hfe
TimeImplementationCompilerBenchmark dateSUPERCOP version
1259040refg++ -m64 -march=core2 -msse4.1 -O3 -fomit-frame-pointer2018091720180818
1327840refg++ -fno-schedule-insns -O3 -fomit-frame-pointer2018091720180818
1506272refg++ -march=barcelona -O -fomit-frame-pointer2018091720180818
1534464refg++ -fno-schedule-insns -O -fomit-frame-pointer2018091720180818

Test failure

Implementation: crypto_encrypt/4hfe/ref
Compiler: clang++ -O3 -fomit-frame-pointer -Qunused-arguments
error 111
crypto_encrypt_open does not match m

Number of similar (compiler,implementation) pairs: 16, namely:
CompilerImplementations
clang++ -O3 -fomit-frame-pointer -Qunused-arguments ref
g++ -O3 -fomit-frame-pointer ref
g++ -funroll-loops -fno-schedule-insns -O3 -fomit-frame-pointer ref
g++ -funroll-loops -fno-schedule-insns -O -fomit-frame-pointer ref
g++ -m64 -O3 -fomit-frame-pointer ref
g++ -m64 -march=core-avx-i -O3 -fomit-frame-pointer ref
g++ -m64 -march=core-avx-i -O -fomit-frame-pointer ref
g++ -m64 -march=core2 -O3 -fomit-frame-pointer ref
g++ -m64 -march=core2 -msse4.1 -O -fomit-frame-pointer ref
g++ -m64 -march=core2 -msse4 -O3 -fomit-frame-pointer ref
g++ -m64 -march=corei7-avx -O3 -fomit-frame-pointer ref
g++ -m64 -march=corei7 -O3 -fomit-frame-pointer ref
g++ -m64 -march=native -mtune=native -O3 -fomit-frame-pointer ref
g++ -m64 -march=native -mtune=native -O -fomit-frame-pointer ref
g++ -m64 -march=nocona -O -fomit-frame-pointer ref
g++ -march=barcelona -O3 -fomit-frame-pointer ref

Test failure

Implementation: crypto_encrypt/4hfe/ref
Compiler: clang++ -mcpu=cortex-a8 -mfpu=neon -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments
error 111
crypto_encrypt_open is nondeterministic

Number of similar (compiler,implementation) pairs: 9, namely:
CompilerImplementations
clang++ -mcpu=cortex-a8 -mfpu=neon -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments ref
g++ ref
g++ -O -fomit-frame-pointer ref
g++ -m64 -march=core-avx2 -O -fomit-frame-pointer ref
g++ -m64 -march=core2 -msse4 -O -fomit-frame-pointer ref
g++ -m64 -march=corei7 -O -fomit-frame-pointer ref
g++ -m64 -march=k8 -O3 -fomit-frame-pointer ref
g++ -m64 -march=k8 -O -fomit-frame-pointer ref
g++ -march=k8 -O -fomit-frame-pointer ref

Test failure

Implementation: crypto_encrypt/4hfe/ref
Compiler: clang++ -mcpu=cortex-a9 -mfpu=neon -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments
error 111
crypto_encrypt_open returns nonzero

Number of similar (compiler,implementation) pairs: 43, namely:
CompilerImplementations
clang++ -mcpu=cortex-a9 -mfpu=neon -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments ref
clang++ -mcpu=native -mfpu=neon -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments ref
g++ -O2 -fomit-frame-pointer ref
g++ -Os -fomit-frame-pointer ref
g++ -fno-schedule-insns -O2 -fomit-frame-pointer ref
g++ -fno-schedule-insns -Os -fomit-frame-pointer ref
g++ -funroll-loops -fno-schedule-insns -O2 -fomit-frame-pointer ref
g++ -funroll-loops -fno-schedule-insns -Os -fomit-frame-pointer ref
g++ -m64 -O2 -fomit-frame-pointer ref
g++ -m64 -Os -fomit-frame-pointer ref
g++ -m64 -march=core-avx-i -O2 -fomit-frame-pointer ref
g++ -m64 -march=core-avx-i -Os -fomit-frame-pointer ref
g++ -m64 -march=core-avx2 -O2 -fomit-frame-pointer ref
g++ -m64 -march=core-avx2 -Os -fomit-frame-pointer ref
g++ -m64 -march=core2 -O2 -fomit-frame-pointer ref
g++ -m64 -march=core2 -O -fomit-frame-pointer ref
g++ -m64 -march=core2 -Os -fomit-frame-pointer ref
g++ -m64 -march=core2 -msse4.1 -O2 -fomit-frame-pointer ref
g++ -m64 -march=core2 -msse4.1 -Os -fomit-frame-pointer ref
g++ -m64 -march=core2 -msse4 -O2 -fomit-frame-pointer ref
g++ -m64 -march=core2 -msse4 -Os -fomit-frame-pointer ref
g++ -m64 -march=corei7-avx -O2 -fomit-frame-pointer ref
g++ -m64 -march=corei7-avx -O -fomit-frame-pointer ref
g++ -m64 -march=corei7-avx -Os -fomit-frame-pointer ref
g++ -m64 -march=corei7 -O2 -fomit-frame-pointer ref
g++ -m64 -march=corei7 -Os -fomit-frame-pointer ref
g++ -m64 -march=k8 -O2 -fomit-frame-pointer ref
g++ -m64 -march=k8 -Os -fomit-frame-pointer ref
g++ -m64 -march=native -mtune=native -O2 -fomit-frame-pointer ref
g++ -m64 -march=native -mtune=native -Os -fomit-frame-pointer ref
g++ -m64 -march=nocona -O2 -fomit-frame-pointer ref
g++ -m64 -march=nocona -Os -fomit-frame-pointer ref
g++ -march=barcelona -O2 -fomit-frame-pointer ref
g++ -march=barcelona -Os -fomit-frame-pointer ref
g++ -march=k8 -O2 -fomit-frame-pointer ref
g++ -march=k8 -O3 -fomit-frame-pointer ref
g++ -march=k8 -Os -fomit-frame-pointer ref
g++ -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv ref
g++ -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv ref
g++ -march=nocona -O2 -fomit-frame-pointer ref
g++ -march=nocona -O3 -fomit-frame-pointer ref
g++ -march=nocona -O -fomit-frame-pointer ref
g++ -march=nocona -Os -fomit-frame-pointer ref

Test failure

Implementation: crypto_encrypt/4hfe/ref
Compiler: g++ -m64 -O -fomit-frame-pointer
error 111
crypto_encrypt_open does not handle s=t overlap

Number of similar (compiler,implementation) pairs: 3, namely:
CompilerImplementations
g++ -m64 -O -fomit-frame-pointer ref
g++ -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv ref
g++ -march=native -mtune=native -O -fomit-frame-pointer -fwrapv ref

Test failure

Implementation: crypto_encrypt/4hfe/ref
Compiler: g++ -m64 -march=barcelona -O2 -fomit-frame-pointer
error 111
crypto_encrypt_open returns nonzero
error 111
crypto_encrypt_open returns nonzero

Number of similar (compiler,implementation) pairs: 3, namely:
CompilerImplementations
g++ -m64 -march=barcelona -O2 -fomit-frame-pointer ref
g++ -m64 -march=barcelona -O -fomit-frame-pointer ref
g++ -m64 -march=barcelona -Os -fomit-frame-pointer ref

Test failure

Implementation: crypto_encrypt/4hfe/ref
Compiler: g++ -m64 -march=barcelona -O3 -fomit-frame-pointer
error 111
crypto_encrypt_open does not match m
error 111
crypto_encrypt_open does not match m

Number of similar (compiler,implementation) pairs: 1, namely:
CompilerImplementations
g++ -m64 -march=barcelona -O3 -fomit-frame-pointer ref

Test failure

Implementation: crypto_encrypt/4hfe/ref
Compiler: g++ -m64 -march=core-avx2 -O3 -fomit-frame-pointer
error 111
crypto_encrypt_open does not handle c=t overlap

Number of similar (compiler,implementation) pairs: 2, namely:
CompilerImplementations
g++ -m64 -march=core-avx2 -O3 -fomit-frame-pointer ref
g++ -m64 -march=nocona -O3 -fomit-frame-pointer ref

Compiler output

Implementation: crypto_encrypt/4hfe/ref
Compiler: clang++ -O3 -fwrapv -march=native -fomit-frame-pointer -Qunused-arguments
HFE-test.cpp: In file included from HFE-test.cpp:18:
HFE-test.cpp: In file included from ./HFE.h:6:
HFE-test.cpp: ./tower.h:364:9: error: argument to '__builtin_ia32_palignr128' must be a constant integer
HFE-test.cpp: return _mm_alignr_epi8(a1, a0, i*2);
HFE-test.cpp: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
HFE-test.cpp: /usr/lib/llvm-3.8/bin/../lib/clang/3.8.0/include/tmmintrin.h:69:12: note: expanded from macro '_mm_alignr_epi8'
HFE-test.cpp: (__m128i)__builtin_ia32_palignr128((__v16qi)(__m128i)(a), \
HFE-test.cpp: ^
HFE-test.cpp: In file included from HFE-test.cpp:18:
HFE-test.cpp: In file included from ./HFE.h:6:
HFE-test.cpp: ./tower.h:364:9: error: cannot initialize return object of type '__m128i' (vector of 2 'long long' values) with an rvalue of type 'void'
HFE-test.cpp: return _mm_alignr_epi8(a1, a0, i*2);
HFE-test.cpp: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
HFE-test.cpp: /usr/lib/llvm-3.8/bin/../lib/clang/3.8.0/include/tmmintrin.h:68:34: note: expanded from macro '_mm_alignr_epi8'
HFE-test.cpp: #define _mm_alignr_epi8(a, b, n) __extension__ ({ \
HFE-test.cpp: ^~~~~~~~~~~~~~~~~~
HFE-test.cpp: 2 errors generated.

Number of similar (compiler,implementation) pairs: 6, namely:
CompilerImplementations
clang++ -O3 -fwrapv -march=native -fomit-frame-pointer -Qunused-arguments ref
clang++ -O3 -fwrapv -march=x86-64 -mcpu=core-avx2 -mavx2 -maes -mpclmul -fomit-frame-pointer -Qunused-arguments ref
clang++ -O3 -fwrapv -mavx2 -fomit-frame-pointer -Qunused-arguments ref
clang++ -O3 -fwrapv -mavx -fomit-frame-pointer -Qunused-arguments ref
clang++ -O3 -fwrapv -mavx -maes -mpclmul -fomit-frame-pointer -Qunused-arguments ref
clang++ -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments ref