JIT has big improvement in integer related programs but did lack to
improve real world i.e. PMC using apps.
I wanted to test, how much we can gain, by doing vtable calls directly
in JIT and did optimize 4 ops (dec_p, inc_p, if_p_ic, unless_p_ic)
(which happen to be used in mops_p.pasm's MOPS lo
- first chunk of last JIT patch was not really testing for signed range
(reverted)
- mapped registers are now ebx,edi,esi,edx. The first 3 are callee
saved. Prior ecx was used, which was only by chance not failing in
tests, which used shift operations, because, when shifting a register
mapped t
Daniel Grunblatt wrote:
You will see it running as fast as mops.c compiled with -O3 if you change
REDO: sub I4, I4, I3
for
REDO: dec I4
I didn't want to change the test case ;-)
But that's obviously part of a higher level optimizer.
Yes, with constant propagation the (todo) optimizer woul
You will see it running as fast as mops.c compiled with -O3 if you change
REDO: sub I4, I4, I3
for
REDO: dec I4
But that's obviously part of a higher level optimizer.
On Wednesday 13 November 2002 15:10, Leopold Toetsch wrote:
> Watch the mops ;-)
>
> leo