This is the mail archive of the
mailing list for the GCC project.
Re: PR 30089: Fix ICE in operand allocation
Andrew MacLeod wrote on 12/14/06 09:22:
On Thu, 2006-12-14 at 07:31 -0500, Diego Novillo wrote:
Oh, OK, thanks. I've lowered the number to something that still handles
PR30089 and I'm running tests over all of my sandbox to see if it makes
a measurable difference. If this isn't it, I will leave the limit as
is, and we should be looking at the other patches to find the culprit.
Jan Hubicka wrote on 12/14/06 05:33:
It may be, yes. We shouldn't need the static buffer for long,
hopefully. Andrew is changing this code. I will try to adjust it down
in the meantime.
I would say that this patch is the most likely suspect for memory
increase reported for this night. (memory tester jammed, so there are
quite few patches cumulated together).
It is 12% for insn-attrtrab, but we are still bellow memory usage before
your merge and above memory usage before original Daniel's aliasing
fixes (that was about 100MB, now we are almost 130)
It would be odd if this was responsible for a measurable increase in
memory. Its simply the size of the buffer we create to sub-allocate
operands out of.
The theory is that if you make that buffer 10 times larger, you have to
allocate new buffers 10 times less often, and although your wasted
memory might go up due to a mostly unused buffer, it shouldn't go up by
more than the size of one buffer, which in this case was less than 100K
on a 64 but machine...