On Sun, 2005-03-27 at 20:08 -0500, Kazu Hirata wrote:
Hi Diego,
By merging, do you mean *replacing* CCP with VRP? Yes, it's
doable. No, it's not a good idea.
Understood.
Also, if we are inserting ASSERT_EXPRs, it seems to be a good idea to
run copy-prop before VRP. Otherwise, we
On Wed, 2005-03-30 at 13:22 -0500, Diego Novillo wrote:
On Wed, Mar 30, 2005 at 10:58:39AM -0700, Jeffrey A Law wrote:
Whatever scheme we use to explicitly expose context sensitive
equivalences in the IL needs to be a pure expression.
Well, that's the fundamental mechanism behind
Hi Jeff,
We'd have to go back and find the PR. I don't remember all the
details, but the problem was big enough to make ASSERT_EXPRs a
far inferior solution to the others I'd looked at for recording
context sensitive equivalences.
Yes, inserting a bunch of ASSERT_EXPRs, updating SSA,
On Wed, 2005-03-30 at 14:19 -0500, Kazu Hirata wrote:
Hi Jeff,
We'd have to go back and find the PR. I don't remember all the
details, but the problem was big enough to make ASSERT_EXPRs a
far inferior solution to the others I'd looked at for recording
context sensitive equivalences.
On Mar 28, 2005 03:08 AM, Kazu Hirata [EMAIL PROTECTED] wrote:
Huh, whey I talked to them on IRC they didn't seem to have implemented
this. I'll try to get this issue one of these days.
Ehm. I did in fact implement this. The trouble was that inserting
blocks into the worklist got more
On Mon, 2005-03-28 at 16:08 +0200, Steven Bosscher wrote:
On Mar 28, 2005 03:08 AM, Kazu Hirata [EMAIL PROTECTED] wrote:
Huh, whey I talked to them on IRC they didn't seem to have implemented
this. I'll try to get this issue one of these days.
Ehm. I did in fact implement this. The
Hi Diego,
By merging, do you mean *replacing* CCP with VRP? Yes, it's
doable. No, it's not a good idea.
Understood.
Also, if we are inserting ASSERT_EXPRs, it seems to be a good idea to
run copy-prop before VRP. Otherwise, we would end up with lots of
D.18001_101 = D.18001_198;
Hi Diego,
There is a copy-propagation pass before VRP. Or do you mean
right before? Sure, the ordering of these passes is in eternal
flux anyway.
Before, but doesn't have to be right before. The current ordering
is reasonable.
Currently, we still have these even after copy prop because
Hi Diego,
Have you considered merging CCP and VRP (as suggested by Kenny last
year at the summit)?
Originally I was thinking that ASSERT_EXPRs, or ranges gathered by VRP
rather, were very useful for jump threading, but they seem to be
useful for constant propagation, too. Consider
void bar
On Sat, Mar 26, 2005 at 12:00:43PM -0500, Kazu Hirata wrote:
Have you considered merging CCP and VRP (as suggested by Kenny last
year at the summit)?
By merging, do you mean *replacing* CCP with VRP? Yes, it's
doable. No, it's not a good idea.
Because of its lattice evaluation, VRP
10 matches
Mail list logo
