Implementation notes: aarch64, pmnod145, crypto_aead/iscream14v1

Computer: pmnod145
Architecture: aarch64
CPU ID: 431f0af1
SUPERCOP version: 20191017
Operation: crypto_aead
Primitive: iscream14v1

Time	Object size	Test size	Implementation	Compiler	Benchmark date	SUPERCOP version
609200	6864 1024 0	27416 1888 872	`ref`	`armclang_-march=armv8a+crypto_-mcpu=thunderx2t99_-mtune=thunderx2t99_-O3_-fPIC_-fPIE`	20191018	20191017
613850	6672 1024 0	27296 1888 864	`ref`	`armclang_-march=armv8a+crypto_-mcpu=thunderx2t99_-mtune=thunderx2t99_-O2_-fPIC_-fPIE`	20191018	20191017
628850	8148 1024 0	25451 1904 880	`ref`	`gcc_-march=native_-mtune=native_-O3_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20191018	20191017
1412850	2880 1024 0	17903 1880 856	`ref`	`gcc_-march=native_-mtune=native_-Os_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20191018	20191017
1416775	3576 1024 0	19771 1896 864	`ref`	`gcc_-march=native_-mtune=native_-O2_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20191018	20191017
1458375	3452 1024 0	19603 1896 864	`ref`	`gcc_-march=native_-mtune=native_-O_-fomit-frame-pointer_-fwrapv_-fPIC_-fPIE`	20191018	20191017

Compiler output

Implementation: sse
Security model: unknown
Compiler: armclang -march=armv8a+crypto -mcpu=thunderx2t99 -mtune=thunderx2t99 -O2 -fPIC -fPIE

iscream.c: iscream.c:200:10: error: use of unknown builtin '__builtin_ia32_psrldi128' [-Wimplicit-function-declaration]
iscream.c: t0 = shift_right(in[0]) & V(0xf);
iscream.c: ^
iscream.c: iscream.c:17:32: note: expanded from macro 'shift_right'
iscream.c: #define shift_right(x) ((v16qi)__builtin_ia32_psrldi128((v4si)x, 4))
iscream.c: ^
iscream.c: iscream.c:200:10: error: invalid conversion between vector type 'v16qi' (vector of 16 'char' values) and integer type 'int' of different size
iscream.c: t0 = shift_right(in[0]) & V(0xf);
iscream.c: ^~~~~~~~~~~~~~~~~~
iscream.c: iscream.c:17:25: note: expanded from macro 'shift_right'
iscream.c: #define shift_right(x) ((v16qi)__builtin_ia32_psrldi128((v4si)x, 4))
iscream.c: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
iscream.c: iscream.c:201:10: error: invalid conversion between vector type 'v16qi' (vector of 16 'char' values) and integer type 'int' of different size
iscream.c: t1 = shift_right(in[2]) & V(0xf);
iscream.c: ^~~~~~~~~~~~~~~~~~
iscream.c: iscream.c:17:25: note: expanded from macro 'shift_right'
iscream.c: #define shift_right(x) ((v16qi)__builtin_ia32_psrldi128((v4si)x, 4))
iscream.c: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
iscream.c: iscream.c:203:10: error: use of unknown builtin '__builtin_ia32_pshufb128' [-Wimplicit-function-declaration]
iscream.c: A = __builtin_ia32_pshufb128(table, t0);
iscream.c: ^
iscream.c: iscream.c:203:10: note: did you mean '__builtin_ia32_psrldi128'?
iscream.c: iscream.c:200:10: note: '__builtin_ia32_psrldi128' declared here
iscream.c: t0 = shift_right(in[0]) & V(0xf);
iscream.c: ^
iscream.c: ...

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

Compiler	Implementations
armclang -march=armv8a+crypto -mcpu=thunderx2t99 -mtune=thunderx2t99 -O2 -fPIC -fPIE	sse
armclang -march=armv8a+crypto -mcpu=thunderx2t99 -mtune=thunderx2t99 -O3 -fPIC -fPIE	sse

Compiler output

Implementation: sse
Security model: unknown
Compiler: gcc -march=native -mtune=native -O2 -fomit-frame-pointer -fwrapv -fPIC -fPIE

iscream.c: iscream.c: In function 'LBox16P':
iscream.c: iscream.c:17:32: warning: implicit declaration of function '__builtin_ia32_psrldi128'; did you mean '__builtin_fabsd128'? [-Wimplicit-function-declaration]
iscream.c: 17 | #define shift_right(x) ((v16qi)__builtin_ia32_psrldi128((v4si)x, 4))
iscream.c: | ^~~~~~~~~~~~~~~~~~~~~~~~
iscream.c: iscream.c:200:10: note: in expansion of macro 'shift_right'
iscream.c: 200 | t0 = shift_right(in[0]) & V(0xf);
iscream.c: | ^~~~~~~~~~~
iscream.c: iscream.c:200:5: error: can't convert a value of type 'int' to vector type '__vector(16) char' which has different size
iscream.c: 200 | t0 = shift_right(in[0]) & V(0xf);
iscream.c: | ^~
iscream.c: iscream.c:201:5: error: can't convert a value of type 'int' to vector type '__vector(16) char' which has different size
iscream.c: 201 | t1 = shift_right(in[2]) & V(0xf);
iscream.c: | ^~
iscream.c: iscream.c:203:10: warning: implicit declaration of function '__builtin_ia32_pshufb128'; did you mean '__builtin_fabsf128'? [-Wimplicit-function-declaration]
iscream.c: 203 | A = __builtin_ia32_pshufb128(table, t0);
iscream.c: | ^~~~~~~~~~~~~~~~~~~~~~~~
iscream.c: | __builtin_fabsf128
iscream.c: iscream.c:203:10: error: incompatible types when assigning to type 'v16qi' {aka '__vector(16) char'} from type 'int'
iscream.c: iscream.c:204:10: error: incompatible types when assigning to type 'v16qi' {aka '__vector(16) char'} from type 'int'
iscream.c: 204 | C = __builtin_ia32_pshufb128(table, t1);
iscream.c: | ^~~~~~~~~~~~~~~~~~~~~~~~
iscream.c: iscream.c:208:10: error: incompatible types when assigning to type 'v16qi' {aka '__vector(16) char'} from type 'int'
iscream.c: 208 | B = __builtin_ia32_pshufb128(table, t0);
iscream.c: | ^~~~~~~~~~~~~~~~~~~~~~~~
iscream.c: iscream.c:209:10: error: incompatible types when assigning to type 'v16qi' {aka '__vector(16) char'} from type 'int'
iscream.c: ...

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

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