Implementation notes: aarch64, a53, crypto_aead/scream12v2

Computer: a53
Architecture: aarch64
CPU ID: unknown CPU ID
SUPERCOP version: 20160731
Operation: crypto_aead
Primitive: scream12v2

Time	Implementation	Compiler	Benchmark date	SUPERCOP version
707200	`ref`	`gcc -mcpu=cortex-a53 -O3 -fomit-frame-pointer -fwrapv`	20160802	20160731
1152800	`ref`	`clang -mcpu=cortex-a53 -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments`	20160802	20160731
1532115	`ref`	`gcc -mcpu=cortex-a53 -O2 -fomit-frame-pointer -fwrapv`	20160802	20160731
2027835	`ref`	`gcc -mcpu=cortex-a53 -Os -fomit-frame-pointer -fwrapv`	20160802	20160731
2229525	`ref`	`gcc -mcpu=cortex-a53 -O -fomit-frame-pointer -fwrapv`	20160802	20160731

Compiler output

Implementation: crypto_aead/scream12v2/sse
Compiler: clang -mcpu=cortex-a53 -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments

scream.c: scream.c:182:10: error: use of unknown builtin '__builtin_ia32_psrldi128' [-Wimplicit-function-declaration]
scream.c: t0 = shift_right(in[0]) & V(0xf);
scream.c: ^
scream.c: scream.c:16:32: note: expanded from macro 'shift_right'
scream.c: #define shift_right(x) ((v16qi)__builtin_ia32_psrldi128((v4si)x, 4))
scream.c: ^
scream.c: scream.c:182:10: error: invalid conversion between vector type 'v16qi' (vector of 16 'char' values) and integer type 'int' of different size
scream.c: t0 = shift_right(in[0]) & V(0xf);
scream.c: ^~~~~~~~~~~~~~~~~~
scream.c: scream.c:16:25: note: expanded from macro 'shift_right'
scream.c: #define shift_right(x) ((v16qi)__builtin_ia32_psrldi128((v4si)x, 4))
scream.c: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
scream.c: scream.c:183:10: error: invalid conversion between vector type 'v16qi' (vector of 16 'char' values) and integer type 'int' of different size
scream.c: t1 = shift_right(in[2]) & V(0xf);
scream.c: ^~~~~~~~~~~~~~~~~~
scream.c: scream.c:16:25: note: expanded from macro 'shift_right'
scream.c: #define shift_right(x) ((v16qi)__builtin_ia32_psrldi128((v4si)x, 4))
scream.c: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
scream.c: scream.c:185:10: error: use of unknown builtin '__builtin_ia32_pshufb128' [-Wimplicit-function-declaration]
scream.c: A = __builtin_ia32_pshufb128(table, t0);
scream.c: ^
scream.c: scream.c:185:10: note: did you mean '__builtin_ia32_psrldi128'?
scream.c: scream.c:182:10: note: '__builtin_ia32_psrldi128' declared here
scream.c: t0 = shift_right(in[0]) & V(0xf);
scream.c: ^
scream.c: ...

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

Compiler	Implementations
clang -mcpu=cortex-a53 -O3 -fomit-frame-pointer -fwrapv -Qunused-arguments	sse

Compiler output

Implementation: crypto_aead/scream12v2/sse
Compiler: gcc -mcpu=cortex-a53 -O2 -fomit-frame-pointer -fwrapv

scream.c: scream.c: In function 'LBox16P':
scream.c: scream.c:182:5: error: can't convert between vector values of different size
scream.c: t0 = shift_right(in[0]) & V(0xf);
scream.c: ^
scream.c: scream.c:183:5: error: can't convert between vector values of different size
scream.c: t1 = shift_right(in[2]) & V(0xf);
scream.c: ^
scream.c: scream.c:185:8: error: incompatible types when assigning to type 'v16qi' from type 'int'
scream.c: A = __builtin_ia32_pshufb128(table, t0);
scream.c: ^
scream.c: scream.c:186:8: error: incompatible types when assigning to type 'v16qi' from type 'int'
scream.c: C = __builtin_ia32_pshufb128(table, t1);
scream.c: ^
scream.c: scream.c:190:8: error: incompatible types when assigning to type 'v16qi' from type 'int'
scream.c: B = __builtin_ia32_pshufb128(table, t0);
scream.c: ^
scream.c: scream.c:191:8: error: incompatible types when assigning to type 'v16qi' from type 'int'
scream.c: D = __builtin_ia32_pshufb128(table, t1);
scream.c: ^
scream.c: scream.c:198:7: error: conversion of scalar 'int' to vector 'v16qi' involves truncation
scream.c: A ^= __builtin_ia32_pshufb128(table, in[0]);
scream.c: ^
scream.c: scream.c:199:7: error: conversion of scalar 'int' to vector 'v16qi' involves truncation
scream.c: C ^= __builtin_ia32_pshufb128(table, in[2]);
scream.c: ^
scream.c: ...

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

Compiler	Implementations
gcc -mcpu=cortex-a53 -O2 -fomit-frame-pointer -fwrapv	sse
gcc -mcpu=cortex-a53 -O3 -fomit-frame-pointer -fwrapv	sse
gcc -mcpu=cortex-a53 -O -fomit-frame-pointer -fwrapv	sse
gcc -mcpu=cortex-a53 -Os -fomit-frame-pointer -fwrapv	sse