Im not sure what you did with your patches, Korey, but a literal '\n' is
not the same as an actual newline :-).
I'm confused. What are you referring to Steve?
___
m5-dev mailing list
m5-dev@m5sim.org
http://m5sim.org/mailman/listinfo/m5-dev
Yea, I thought it would do the trick...obviously not!
I was adding the message as : hg qrefresh -m \n
So I guess I have to commit in a real text editor to get the newline to show.
On Sat, Apr 18, 2009 at 12:16 PM, Steve Reinhardt ste...@gmail.com wrote:
Im not sure what you did with
You can actually put a real newline in there. If you're in the middle
of a quoted string, the shell won't try to execute the statement.
Instead, it will allow you to finish the string. So I regularly will
do something like:
hg qref -m foo: this is a foo bar
more comments
even more
On Sat, Apr
Oh, and don't forget that you should be saying hg outgoing to make
sure that you know what you're pushing before you say hg push. You
would have noticed the error if you had done so.
Nate
On Sat, Apr 18, 2009 at 9:51 AM, nathan binkert n...@binkert.org wrote:
You can actually put a real
OK So I finally figured out what's wrong here and admittedly I'm a
bit apprehensive of the universal (all ISA) fix will be but we'll see.
So let me explain the situation from the top:
(1) In the Decode of O3, an unconditional pc-relative branch is
resolved and the prediciton history table is
Thanks for digging into this in more detail, Korey. It sounds to me like
maybe the problem is that MIPS is setting predPC to the branch target, when
(to account for the delay slot) it ought to be leaving predPC as PC+4 and
setting predNPC to be the branch target. That way if the prediction turns
Also, why do you change inst-readNextNPC() to inst-readNextPC()
in iew_impl.hh? Aren't you going to then print the same value twice?
Well, I'm trying to fix error'd DPRINT statements here.
In the DPRINT, it says NPC, but it's printing out the NNPC which was
confusing to me. So maybe the
# HG changeset patch
# User Korey Sewell ksew...@umich.edu
# Date 1239850525 14400
# Node ID ce539a58327b78b049f8af0bf8a00737f2b97640
# Parent 7b685fee8773c2fe0bd09ec7abb0a43873aeed92
Fix Branch Prediction for MIPS. Namely, i dont see the reason for updating the
branch history on anything that
This is the patch that will prevent me from committing.
Basically, the decode stage is finding a mispredicted branch and then
deleting information out the branch history table for squashed instructions
AND itself.
This messes things up for when the actual mispredicted instruction gets to
commit
I'm confused... this patch looks like it pretty much just inserts a bunch of
DPRINTFs (plus the seemingly mistaken inclusion of a whitespace change in
mips/isa/decoder.isa).
I don't know anything specific about the O3 predictor, but it makes sense
that the history is updated speculatively in
I see now, I missed that little change buried at the bottom.
I recall that condition about unconditional PC-relative branches finding out
in decode that their target was mispredicted too.
I'm wondering if you're not just hacking around the real problem here... I
don't see how this particular
Yea, I'm worried about some more deep-seeded problem here too...In the
interim, though, at least we know the main problem is with the branch
prediction (since I fixed the TLB/Syscall stuff in other patches).
Like I mentioned earlier, I did the same *hack:* for Alpha and Sparc and
both of the
My question is why does the branch predictor matter as far as functional
correctness? If it was wrong 100% of the time wouldn't that just make O3
slow? Also, why do you change inst-readNextNPC() to inst-readNextPC()
in iew_impl.hh? Aren't you going to then print the same value twice?
Gabe
Korey
13 matches
Mail list logo
