Quoting Paolo Bonzini <bonz...@gnu.org>:
On 12/23/2009 01:01 PM, Steven Bosscher wrote:
On Wed, Dec 23, 2009 at 12:49 PM, Bingfeng Mei<b...@broadcom.com> wrote:
Hello,
I encounter an issue with PRE optimization, which created worse
Is this at -O2 or -O3?
I think this could be fixed if fwprop propagated addresses into loops;
it doesn't because it made performance worse on x86. The real reason
is "address_cost on x86 sucks and nobody knows how to fix it exactly",
And nobody has bothered to put in a bug report into bugzilla to describe
what they see malfunctioning... or if they did, it doesn't mention
address_cost.
but the performance hit was bad enough that we (Steven Bosscher and I)
decided to put that hack into fwprop.
So if this is only useful for a limited set of targets, why isn't it
controlled by an option or a target hook so that it is only turned on
on the targets where it is deemed to make sense overall?
