forked from enlightenment/efl
slight tune of neon asm - based on stuff from nash. not worth a
changelog entry though. not a major enough improvement as i measure it. SVN revision: 56614
This commit is contained in:
parent
7443e3ecb5
commit
22230fe193
|
@ -113,7 +113,8 @@ _op_blend_p_dp_neon(DATA32 *s, DATA8 *m, DATA32 c, DATA32 *d, int l) {
|
|||
//** Oct loop
|
||||
"vldm %[s]!, {d0,d1,d2,d3) \n\t"
|
||||
"vldm %[d], {d4,d5,d6,d7} \n\t"
|
||||
|
||||
"pld [%[s], #64] \n\t"
|
||||
|
||||
|
||||
// Copy s.a into q2 (>> 24) & subtract from 255
|
||||
"vmvn.u8 q4, q0 \n\t"
|
||||
|
|
Loading…
Reference in New Issue