Implementation notes: amd64, rumba3, crypto_stream/speck6496ctr

Computer: rumba3
Microarchitecture: amd64; Zen (800f11)
Architecture: amd64
CPU ID: AuthenticAMD-00800f11-178bfbff
SUPERCOP version: 20221122
Operation: crypto_stream
Primitive: speck6496ctr

Time	Object size	Test size	Implementation	Compiler	Benchmark date	SUPERCOP version
2947	32896 0 0	45392 780 992	`T:avx2`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20230526	20221122
3018	31353 0 0	46851 844 992	`T:avx2`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230526	20221122
3023	33386 0 0	49259 844 992	`T:avx2`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230526	20221122
3024	33347 0 0	47028 796 1024	`T:avx2`	`gcc_-march=native_-mtune=native_-O_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20230526	20221122
3033	31105 0 0	44461 836 1024	`T:avx2`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230526	20221122
3035	31209 0 0	43795 844 960	`T:avx2`	`clang_-march=native_-O_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230526	20221122
3056	27489 0 0	41148 796 1024	`T:sse4`	`gcc_-march=native_-mtune=native_-O_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20230526	20221122
3078	35484 0 0	51125 804 1024	`T:avx2`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20230526	20221122
3123	33319 0 0	47637 804 1024	`T:avx2`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20230526	20221122
3166	26607 0 0	42171 844 992	`T:sse4`	`clang_-march=native_-O2_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230526	20221122
3180	26377 0 0	39797 836 1024	`T:sse4`	`clang_-march=native_-Os_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230526	20221122
3183	28562 0 0	44515 844 992	`T:sse4`	`clang_-march=native_-O3_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230526	20221122
3202	27466 0 0	39944 780 992	`T:sse4`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20230526	20221122
3210	29453 0 0	45069 804 1024	`T:sse4`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20230526	20221122
3213	27437 0 0	41749 804 1024	`T:sse4`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20230526	20221122
3231	26479 0 0	39115 844 960	`T:sse4`	`clang_-march=native_-O_-fomit-frame-pointer_-fwrapv_-Qunused-arguments_-fPIC_-fPIE`	20230526	20221122

Compiler output

Implementation: T:avx2
Security model: timingleaks
Compiler: clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

stream.c: stream.c:328:3: error: always_inline function '_mm256_set_epi32' requires target feature 'avx', but would be inlined into function 'ExpandKey' that is compiled without support for 'avx'
stream.c: EK(A,B,C,rk,key);
stream.c: ^
stream.c: ./Speck6496AVX2.h:53:26: note: expanded from macro 'EK'
stream.c: #define EK(A,B,C,k,key) (RK(B,A,k,key,0), RK(C,A,k,key,1), RK(B,A,k,key,2), RK(C,A,k,key,3), RK(B,A,k,key,4), RK(C,A,k,key,5), RK(B,A,k,key,6), \
stream.c: ^
stream.c: ./Speck6496AVX2.h:51:28: note: expanded from macro 'RK'
stream.c: #define RK(X,Y,k,key,i) (SET1(k[i],Y), key[i]=Y, X=RCS(X,8), X+=Y, X^=i, Y=LCS(Y,3), Y^=X)
stream.c: ^
stream.c: ./Intrinsics_AVX2_64block.h:25:22: note: expanded from macro 'SET1'
stream.c: #define SET1(X,c) (X=SET(c,c,c,c,c,c,c,c))
stream.c: ^
stream.c: ./Intrinsics_AVX2_64block.h:24:13: note: expanded from macro 'SET'
stream.c: #define SET _mm256_set_epi32
stream.c: ^
stream.c: stream.c:328:3: error: AVX vector return of type '__m256i' (vector of 4 'long long' values) without 'avx' enabled changes the ABI
stream.c: ./Speck6496AVX2.h:53:26: note: expanded from macro 'EK'
stream.c: #define EK(A,B,C,k,key) (RK(B,A,k,key,0), RK(C,A,k,key,1), RK(B,A,k,key,2), RK(C,A,k,key,3), RK(B,A,k,key,4), RK(C,A,k,key,5), RK(B,A,k,key,6), \
stream.c: ^
stream.c: ./Speck6496AVX2.h:51:28: note: expanded from macro 'RK'
stream.c: #define RK(X,Y,k,key,i) (SET1(k[i],Y), key[i]=Y, X=RCS(X,8), X+=Y, X^=i, Y=LCS(Y,3), Y^=X)
stream.c: ^
stream.c: ./Intrinsics_AVX2_64block.h:25:22: note: expanded from macro 'SET1'
stream.c: #define SET1(X,c) (X=SET(c,c,c,c,c,c,c,c))
stream.c: ^
stream.c: ...

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

Compiler	Implementations
clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx2

Compiler output

Implementation: T:avx512
Security model: timingleaks
Compiler: clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

stream.c: stream.c:347:3: error: always_inline function '_mm512_set_epi32' requires target feature 'avx512f', but would be inlined into function 'ExpandKey' that is compiled without support for 'avx512f'
stream.c: EK(A,B,C,rk,key);
stream.c: ^
stream.c: ./Speck6496AVX512.h:53:26: note: expanded from macro 'EK'
stream.c: #define EK(A,B,C,k,key) (RK(B,A,k,key,0), RK(C,A,k,key,1), RK(B,A,k,key,2), RK(C,A,k,key,3), RK(B,A,k,key,4), RK(C,A,k,key,5), RK(B,A,k,key,6), \
stream.c: ^
stream.c: ./Speck6496AVX512.h:51:26: note: expanded from macro 'RK'
stream.c: #define RK(X,Y,k,key,i) (SET1(k[i],Y), key[i]=Y, X=RCS(X,8), X+=Y, X^=i, Y=LCS(Y,3), Y^=X)
stream.c: ^
stream.c: ./Intrinsics_AVX512_64block.h:25:22: note: expanded from macro 'SET1'
stream.c: #define SET1(X,c) (X=SET(c,c,c,c,c,c,c,c,c,c,c,c,c,c,c,c))
stream.c: ^
stream.c: ./Intrinsics_AVX512_64block.h:14:13: note: expanded from macro 'SET'
stream.c: #define SET _mm512_set_epi32
stream.c: ^
stream.c: stream.c:347:3: error: AVX vector return of type '__m512i' (vector of 8 'long long' values) without 'avx512f' enabled changes the ABI
stream.c: ./Speck6496AVX512.h:53:26: note: expanded from macro 'EK'
stream.c: #define EK(A,B,C,k,key) (RK(B,A,k,key,0), RK(C,A,k,key,1), RK(B,A,k,key,2), RK(C,A,k,key,3), RK(B,A,k,key,4), RK(C,A,k,key,5), RK(B,A,k,key,6), \
stream.c: ^
stream.c: ./Speck6496AVX512.h:51:26: note: expanded from macro 'RK'
stream.c: #define RK(X,Y,k,key,i) (SET1(k[i],Y), key[i]=Y, X=RCS(X,8), X+=Y, X^=i, Y=LCS(Y,3), Y^=X)
stream.c: ^
stream.c: ./Intrinsics_AVX512_64block.h:25:22: note: expanded from macro 'SET1'
stream.c: #define SET1(X,c) (X=SET(c,c,c,c,c,c,c,c,c,c,c,c,c,c,c,c))
stream.c: ^
stream.c: ...

Number of similar (compiler,implementation) pairs: 5, namely:

Compiler	Implementations
clang -march=native -O2 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512
clang -march=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512
clang -march=native -O -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512
clang -march=native -Os -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512
clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:avx512

Compiler output

Implementation: T:avx512
Security model: timingleaks
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE

stream.c: In file included from Speck6496AVX512.h:17,
stream.c: from stream.c:21:
stream.c: stream.c: In function 'Encrypt':
stream.c: Intrinsics_AVX512_64block.h:25:21: warning: AVX512F vector return without AVX512F enabled changes the ABI [-Wpsabi]
stream.c: 25 | #define SET1(X,c) (X=SET(c,c,c,c,c,c,c,c,c,c,c,c,c,c,c,c))
stream.c: | ~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
stream.c: stream.c:135:5: note: in expansion of macro 'SET1'
stream.c: 135 | SET1(X[0],nonce[1]); SET16(Y[0],nonce[0]);
stream.c: | ^~~~
stream.c: In file included from /usr/lib/gcc/x86_64-linux-gnu/11/include/immintrin.h:49,
stream.c: from Intrinsics_AVX512_64block.h:5,
stream.c: from Speck6496AVX512.h:17,
stream.c: from stream.c:21:
stream.c: /usr/lib/gcc/x86_64-linux-gnu/11/include/avx512fintrin.h:6445:1: error: inlining failed in call to 'always_inline' '_mm512_storeu_si512': target specific option mismatch
stream.c: 6445 | _mm512_storeu_si512 (void *__P, __m512i __A)
stream.c: | ^~~~~~~~~~~~~~~~~~~
stream.c: In file included from Speck6496AVX512.h:17,
stream.c: from stream.c:21:
stream.c: Intrinsics_AVX512_64block.h:31:18: note: called from here
stream.c: 31 | #define ST(ip,X) _mm512_storeu_si512((__m512i *)(ip),X)
stream.c: | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
stream.c: Intrinsics_AVX512_64block.h:32:42: note: in expansion of macro 'ST'
stream.c: 32 | #define STORE(out,X,Y) (ST(out,LOW(Y,X)),ST(out+64,HIGH(Y,X)))
stream.c: | ^~
stream.c: stream.c:138:5: note: in expansion of macro 'STORE'
stream.c: ...

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

Compiler	Implementations
gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512
gcc -march=native -mtune=native -O3 -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512
gcc -march=native -mtune=native -O -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512
gcc -march=native -mtune=native -Os -fomit-frame-pointer -fwrapv -fPIC -fPIE	T:avx512

Compiler output

Implementation: T:sse4
Security model: timingleaks
Compiler: clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE

stream.c: stream.c:133:21: error: always_inline function '_mm_shuffle_epi8' requires target feature 'ssse3', but would be inlined into function 'Encrypt' that is compiled without support for 'ssse3'
stream.c: if (numbytes==32) Enc(X,Y,rk,4);
stream.c: ^
stream.c: ./Speck6496SSE4.h:43:23: note: expanded from macro 'Enc'
stream.c: #define Enc(X,Y,k,n) (Rx##n(X,Y,k[0]), Rx##n(X,Y,k[1]), Rx##n(X,Y,k[2]), Rx##n(X,Y,k[3]), Rx##n(X,Y,k[4]), Rx##n(X,Y,k[5]), Rx##n(X,Y,k[6]), Rx##n(X,Y,k[7]), \
stream.c: ^
stream.c: <scratch space>:125:1: note: expanded from here
stream.c: Rx4
stream.c: ^
stream.c: ./Speck6496SSE4.h:24:22: note: expanded from macro 'Rx4'
stream.c: #define Rx4(X,Y,k) (R(X[0],Y[0],k))
stream.c: ^
stream.c: ./Speck6496SSE4.h:22:29: note: expanded from macro 'R'
stream.c: #define R(X,Y,k) (X=XOR(ADD(ROR8(X),Y),k), Y=XOR(ROL(Y,3),X))
stream.c: ^
stream.c: ./Intrinsics_SSE4_64block.h:40:19: note: expanded from macro 'ROR8'
stream.c: #define ROR8(X) (SHFL(X,R8))
stream.c: ^
stream.c: ./Intrinsics_SSE4_64block.h:35:14: note: expanded from macro 'SHFL'
stream.c: #define SHFL _mm_shuffle_epi8
stream.c: ^
stream.c: stream.c:133:21: error: always_inline function '_mm_shuffle_epi8' requires target feature 'ssse3', but would be inlined into function 'Encrypt' that is compiled without support for 'ssse3'
stream.c: ./Speck6496SSE4.h:43:41: note: expanded from macro 'Enc'
stream.c: #define Enc(X,Y,k,n) (Rx##n(X,Y,k[0]), Rx##n(X,Y,k[1]), Rx##n(X,Y,k[2]), Rx##n(X,Y,k[3]), Rx##n(X,Y,k[4]), Rx##n(X,Y,k[5]), Rx##n(X,Y,k[6]), Rx##n(X,Y,k[7]), \
stream.c: ^
stream.c: ...

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

Compiler	Implementations
clang -mcpu=native -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments -fPIC -fPIE	T:sse4