Implementation notes: armeabi, cubox, crypto_stream/salsa2012

Computer: cubox
Architecture: armeabi
CPU ID: unknown CPU ID
SUPERCOP version: 20161026
Operation: crypto_stream
Primitive: salsa2012

Time	Implementation	Compiler	Benchmark date	SUPERCOP version
18418	`e/merged`	`gcc -funroll-loops -mcpu=marvell-pj4 -O2`	20161210	20161026
18529	`e/merged`	`gcc -funroll-loops -mcpu=marvell-pj4 -O3`	20161210	20161026
18584	`e/merged`	`gcc -mcpu=marvell-pj4 -O3`	20161210	20161026
18624	`e/merged`	`gcc -mcpu=marvell-pj4 -O2`	20161210	20161026
21175	`ref`	`gcc -funroll-loops -mcpu=marvell-pj4 -O3`	20161210	20161026
21176	`ref`	`gcc -mcpu=marvell-pj4 -O3`	20161210	20161026
21267	`e/ref`	`gcc -funroll-loops -mcpu=marvell-pj4 -O3`	20161210	20161026
21352	`e/regs`	`gcc -funroll-loops -mcpu=marvell-pj4 -O3`	20161210	20161026
23684	`e/ref`	`gcc -mcpu=marvell-pj4 -O3`	20161210	20161026
23696	`e/regs`	`gcc -mcpu=marvell-pj4 -O3`	20161210	20161026
23729	`e/ref`	`gcc -funroll-loops -mcpu=marvell-pj4 -O2`	20161210	20161026
23948	`e/merged`	`gcc -funroll-loops -mcpu=marvell-pj4 -Os`	20161210	20161026
24751	`e/regs`	`gcc -funroll-loops -mcpu=marvell-pj4 -O2`	20161210	20161026
25484	`e/merged`	`gcc -mcpu=marvell-pj4 -Os`	20161210	20161026
27105	`ref`	`gcc -funroll-loops -mcpu=marvell-pj4 -O2`	20161210	20161026
29360	`e/regs`	`gcc -mcpu=marvell-pj4 -O2`	20161210	20161026
30480	`ref`	`gcc -mcpu=marvell-pj4 -O2`	20161210	20161026
31339	`e/ref`	`gcc -mcpu=marvell-pj4 -O2`	20161210	20161026
32380	`ref`	`gcc -funroll-loops -mcpu=marvell-pj4 -Os`	20161210	20161026
32456	`ref`	`gcc -mcpu=marvell-pj4 -Os`	20161210	20161026
35272	`e/regs`	`gcc -mcpu=marvell-pj4 -Os`	20161210	20161026
35376	`e/regs`	`gcc -funroll-loops -mcpu=marvell-pj4 -Os`	20161210	20161026
38124	`e/ref`	`gcc -funroll-loops -mcpu=marvell-pj4 -Os`	20161210	20161026
38144	`e/ref`	`gcc -mcpu=marvell-pj4 -Os`	20161210	20161026

Test failure

Implementation: crypto_stream/salsa2012/armneon3
Compiler: gcc -funroll-loops -mcpu=marvell-pj4 -O2

error 111

Number of similar (compiler,implementation) pairs: 6, namely:

Compiler	Implementations
gcc -funroll-loops -mcpu=marvell-pj4 -O2	armneon3
gcc -funroll-loops -mcpu=marvell-pj4 -O3	armneon3
gcc -funroll-loops -mcpu=marvell-pj4 -Os	armneon3
gcc -mcpu=marvell-pj4 -O2	armneon3
gcc -mcpu=marvell-pj4 -O3	armneon3
gcc -mcpu=marvell-pj4 -Os	armneon3

Compiler output

Implementation: crypto_stream/salsa2012/armneon2
Compiler: gcc -funroll-loops -mcpu=marvell-pj4 -O2

xor.c: In file included from xor.c:9:0:
xor.c: xor.c: In function 'crypto_stream_salsa2012_armneon2_xor':
xor.c: /usr/lib/gcc/armv7l-unknown-linux-gnueabihf/6.2.1/include/arm_neon.h:6187:1: error: inlining failed in call to always_inline 'vcombine_u32': target specific option mismatch
xor.c: vcombine_u32 (uint32x2_t __a, uint32x2_t __b)
xor.c: ^~~~~~~~~~~~
xor.c: xor.c:40:14: note: called from here
xor.c: uint32x4_t start1 = vcombine_u32(k5k0,n0k4);
xor.c: ^~~~~~
xor.c: In file included from xor.c:9:0:
xor.c: /usr/lib/gcc/armv7l-unknown-linux-gnueabihf/6.2.1/include/arm_neon.h:7553:1: error: inlining failed in call to always_inline 'vext_u32': target specific option mismatch
xor.c: ...
xor.c: xor.c:354:3: note: called from here
xor.c: vst1q_u8((uint8_t *) c,(uint8x16_t) x0x1x2x3);
xor.c: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
xor.c: In file included from xor.c:9:0:
xor.c: /usr/lib/gcc/armv7l-unknown-linux-gnueabihf/6.2.1/include/arm_neon.h:565:1: error: inlining failed in call to always_inline 'vadd_u64': target specific option mismatch
xor.c: vadd_u64 (uint64x1_t __a, uint64x1_t __b)
xor.c: ^~~~~~~~
xor.c: xor.c:364:23: note: called from here
xor.c: n2n3 = (uint32x2_t) vadd_u64(nextblock,(uint64x1_t) n2n3);
xor.c: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Number of similar (compiler,implementation) pairs: 6, namely:

Compiler	Implementations
gcc -funroll-loops -mcpu=marvell-pj4 -O2	armneon2
gcc -funroll-loops -mcpu=marvell-pj4 -O3	armneon2
gcc -funroll-loops -mcpu=marvell-pj4 -Os	armneon2
gcc -mcpu=marvell-pj4 -O2	armneon2
gcc -mcpu=marvell-pj4 -O3	armneon2
gcc -mcpu=marvell-pj4 -Os	armneon2

Compiler output

Implementation: crypto_stream/salsa2012/armneon
Compiler: gcc -funroll-loops -mcpu=marvell-pj4 -O2

xor.c: In file included from xor.c:9:0:
xor.c: xor.c: In function 'crypto_stream_salsa2012_armneon_xor':
xor.c: /usr/lib/gcc/armv7l-unknown-linux-gnueabihf/6.2.1/include/arm_neon.h:6187:1: error: inlining failed in call to always_inline 'vcombine_u32': target specific option mismatch
xor.c: vcombine_u32 (uint32x2_t __a, uint32x2_t __b)
xor.c: ^~~~~~~~~~~~
xor.c: xor.c:40:14: note: called from here
xor.c: uint32x4_t start1 = vcombine_u32(k5k0,n0k4);
xor.c: ^~~~~~
xor.c: In file included from xor.c:9:0:
xor.c: /usr/lib/gcc/armv7l-unknown-linux-gnueabihf/6.2.1/include/arm_neon.h:7553:1: error: inlining failed in call to always_inline 'vext_u32': target specific option mismatch
xor.c: ...
xor.c: xor.c:166:3: note: called from here
xor.c: vst1q_u8((uint8_t *) c,(uint8x16_t) x0x1x2x3);
xor.c: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
xor.c: In file included from xor.c:9:0:
xor.c: /usr/lib/gcc/armv7l-unknown-linux-gnueabihf/6.2.1/include/arm_neon.h:565:1: error: inlining failed in call to always_inline 'vadd_u64': target specific option mismatch
xor.c: vadd_u64 (uint64x1_t __a, uint64x1_t __b)
xor.c: ^~~~~~~~
xor.c: xor.c:176:23: note: called from here
xor.c: n2n3 = (uint32x2_t) vadd_u64(nextblock,(uint64x1_t) n2n3);
xor.c: ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Number of similar (compiler,implementation) pairs: 6, namely:

Compiler	Implementations
gcc -funroll-loops -mcpu=marvell-pj4 -O2	armneon
gcc -funroll-loops -mcpu=marvell-pj4 -O3	armneon
gcc -funroll-loops -mcpu=marvell-pj4 -Os	armneon
gcc -mcpu=marvell-pj4 -O2	armneon
gcc -mcpu=marvell-pj4 -O3	armneon
gcc -mcpu=marvell-pj4 -Os	armneon