kimcinoo
/
efl
forked from enlightenment/efl
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fix slow path in neon - memcpy is actually faster! :) 

SVN revision: 56240
This commit is contained in:
Carsten Haitzler 2011-01-20 09:33:07 +00:00
parent 88a8651acb
commit ecd007c049
1 changed files with 35 additions and 49 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

84
legacy/evas/src/lib/engines/common/evas_op_copy/op_copy_pixel_neon.c
View File

@ -3,61 +3,47 @@
#ifdef BUILD_NEON #ifdef BUILD_NEON
static void static void
_op_copy_p_dp_neon(DATA32 *s, DATA8 *m __UNUSED__, DATA32 c __UNUSED__, DATA32 *d, int l) { _op_copy_p_dp_neon(DATA32 *s, DATA8 *m __UNUSED__, DATA32 c __UNUSED__, DATA32 *d, int l) {
uint32_t *e; DATA32 *e;
e = d + l; // if (((unsigned long)s & 0xf) || ((unsigned long)d & 0xf))
//#ifdef NEON_INSTRINSICS_OK // {
#if 0 memcpy(d, s, l * sizeof(DATA32));
// odd this is faster than the below asm... :( // return;
e -= 15; // }
uint32x4_t col1, col2, col3, col4; /*
// fill a run of 4x4 (16) pixels with the color e = d + l - 23;
for (; d < e; d += 16, s += 16) { if (e > d)
col1 = vld1q_u32(s+0); // OP
col2 = vld1q_u32(s+4); // OP
col3 = vld1q_u32(s+8); // OP
col4 = vld1q_u32(s+12); // OP
vst1q_u32(d+0, col1); // OP
vst1q_u32(d+4, col2); // OP
vst1q_u32(d+8, col3); // OP
vst1q_u32(d+12, col4); // OP
}
e += 15;
#else
if ((e - d) >= 16)
{ {
DATA32 *d2, *d3, *d4; int dl;
DATA32 *s2, *s3, *s4;
e -= 31;
d2 = d + 4;
d3 = d + 8;
d4 = d + 12;
s2 = s + 4;
s3 = s + 8;
s4 = s + 12;
asm volatile ( asm volatile (
".fpu neon \n\t" ".fpu neon \n\t"
"asmloop2:\n\t" "asmloop2: \n\t"
"cmp %[e], %[d]\n\t" "cmp %[e], %[d] \n\t" // compare current and end ptr
"vld1.32 {d16-d17}, [%[s]]!\n\t" "pld [%[s], #64] \n\t" // preload 64 bytes ahead
"vld1.32 {d18-d19}, [%[s2]]!\n\t" "pld [%[s], #256] \n\t" // preload 256 bytes ahead
"vld1.32 {d20-d21}, [%[s3]]!\n\t" "pld [%[s], #320] \n\t" // preload 320 bytes ahead
"vld1.32 {d22-d23}, [%[s4]]!\n\t" "vld1.64 {d0-d3}, [%[s]]! \n\t" // load 256bits (32 bytes 8 pix)
"vst1.32 {d16-d17}, [%[d]]!\n\t" "vld1.64 {d4-d7} , [%[s]]! \n\t" // load 256bits (32 bytes 8 pix)
"vst1.32 {d18-d19}, [%[d2]]!\n\t" "vld1.64 {d8-d11}, [%[s]]! \n\t" // load 256bits (32 bytes 8 pix)
"vst1.32 {d20-d21}, [%[d3]]!\n\t" "vst1.64 {d0-d3}, [%[d]]! \n\t" // store 256bits (32 bytes 8 pix)
"vst1.32 {d22-d23}, [%[d4]]!\n\t" "vst1.64 {d4-d7}, [%[d]]! \n\t" // store 256bits (32 bytes 8 pix)
"bhi asmloop2\n\t" "vst1.64 {d8-d11}, [%[d]]! \n\t" // store 256bits (32 bytes 8 pix)
"bhi asmloop2 \n\t"
: // output regs : // output regs
: [s] "r" (s), [s2] "r" (s2), [s3] "r" (s3), [s4] "r" (s4), [e] "r" (e), [d] "r" (d), [d2] "r" (d2), [d3] "r" (d3), [d4] "r" (d4) // input : [s] "r" (s), [e] "r" (e), [d] "r" (d) // input
: "q8", "q9", "q10", "q11", "d16", "d17", "d18", "d19", "d20", "d21", "d22", "d23", "memory" // clobbered : "q0", "q1", "q2", "q3", "q4", "q5",
); "d0", "d1", "d2", "d3", "d4", "d5",
e += 31; "d6", "d7", "d8", "d9", "d10", "d11",
"memory" // clobbered
);
e = d + l;
dl = l - (l % 24);
s = s + dl;
d = d + dl;
} }
#endif
// fixup any leftover pixels in the run
for (; d < e; d++, s++) { for (; d < e; d++, s++) {
*d = *s; // OP *d = *s;
} }
*/
} }
#define _op_copy_pan_dp_neon _op_copy_p_dp_neon #define _op_copy_pan_dp_neon _op_copy_p_dp_neon