Page 1 of 2

2,5x faster memcpy routine for MIPS devices

PostPosted: Jan 5, 2003 @ 3:47am
by Guest

PostPosted: Jan 6, 2003 @ 6:55pm
by angedelamort

PostPosted: Jan 6, 2003 @ 7:28pm
by Deje

PostPosted: Jan 6, 2003 @ 9:00pm
by Guest

PostPosted: Jan 7, 2003 @ 12:22am
by brendan

PostPosted: Jan 7, 2003 @ 12:52am
by Digby

PostPosted: Jan 7, 2003 @ 1:27am
by Deje

PostPosted: Jan 7, 2003 @ 9:01am
by Sergey Chaban
I don't know about MIPS, but ARM & SH compilers indeed inline the function if count param for memcpy is constant and small enough (and intrinsic option is enabled).
"Small enough" is like this:
For ARM: count <= 8
For SH: count < 16
But generated code copies data byte by byte (ldrb/strb or mov.b), i.e. compiler makes no assumptions about src/dest alignment.

SC

PostPosted: Jan 7, 2003 @ 10:43am
by pazazzo

PostPosted: Jan 7, 2003 @ 12:37pm
by Sergey Chaban

PostPosted: Jan 7, 2003 @ 10:53pm
by Guest

PostPosted: Jan 9, 2003 @ 2:02am
by Dave H
Ah the golden days of computing are back :D

* Smiles from ear to ear *

The perfect antidote to all this Java programming ;)

PostPosted: Jan 9, 2003 @ 5:05am
by Digby

PostPosted: Jan 13, 2003 @ 3:30am
by brendan

PostPosted: Jan 13, 2003 @ 5:28am
by Dan East
You aren't trying to do inline assembly for an ARM build are you?

Dan East