Hello,
In our port, I created a new NOTE to preserve pragma info. The
note is generated as follows in expanding builtins.
rtx note = emit_note(NOTE_INSN_LOOPCOUNT_PRAGMA_BEG);
rtx vector = gen_rtx_PARALLEL (VOIDmode,
gen_rtvec(4, op0, op1, op2, op3));
> What about liveness? No hard reg, pseudo, mem will live avross the
> unspec volatile. Right?
Wrong. A volatile unspec may use/change machine state not directly accessible
by gcc. Any use of or changes to the machine state modelled by gcc should be
explicit in the pattern.
i.e. if your patter
On Thu, 11 Nov 2010, Ian Lance Taylor wrote:
> roy rosen writes:
>
> > If I have two insns:
> > r2 = r3
> > r3 = r4
> > It seems to me that the dependency analysis creates a dependency
> > between the two and prevent parallelization. Although there is a
> > dependency (because of r3) I want GC
"Bingfeng Mei" writes:
> In our port, I created a new NOTE to preserve pragma info. The
> note is generated as follows in expanding builtins.
>
>rtx note = emit_note(NOTE_INSN_LOOPCOUNT_PRAGMA_BEG);
>rtx vector = gen_rtx_PARALLEL (VOIDmode,
> gen_rtvec(
Hi -
2010/11/12 Bingfeng Mei :
> Hello,
> In our port, I created a new NOTE to preserve pragma info. The
> note is generated as follows in expanding builtins.
>
> rtx note = emit_note(NOTE_INSN_LOOPCOUNT_PRAGMA_BEG);
> rtx vector = gen_rtx_PARALLEL (VOIDmode,
>
I am not aware of that. Thank you very much.
Cheers,
Bingfeng
> -Original Message-
> From: Andreas Schwab [mailto:sch...@redhat.com]
> Sent: 12 November 2010 11:24
> To: Bingfeng Mei
> Cc: gcc@gcc.gnu.org
> Subject: Re: How to debug when some rtx are wrongly reclaimed by
> garbage collec
Hi,
Can anybody give me a hint on where (perhaps some branch) I can find
functionality
which allows during scheduling to un-schedule an instruction? I basically need
a function that does the oposite of schedule_insn.
During scheduling I want o schedule_insn(INSN), then check the ready_list and
Alexander is right. Perhaps you can implement the TARGET_SCHED_ADJUST_COST
, then catch in the debugger the two instructions that you expect to be
scheduled together and see what the default latency is or if needed you
may just adjust it to the proper value.
Alex
--- On Fri, 11/12/10, Alexande
Hi,
For a private target that i am porting in GCC 4.5 I have the following
pattern in my md file for call value:
(define_insn "call_value_op"
[(set (match_operand 0 "register_operand" "=da")
(call (mem:QI (match_operand:QI 1 "call_operand" "Wd"))
(match_operand:QI 2 "" ""
Quoting Mohamed Shafi :
So i have the following questions:
1. Why is that constraints are not matched here?
Please read the node "Register Classes" in doc/tm.texi .
On 12 November 2010 18:39, Joern Rennecke wrote:
> Quoting Mohamed Shafi :
>
>> So i have the following questions:
>>
>> 1. Why is that constraints are not matched here?
>
> Please read the node "Register Classes" in doc/tm.texi .
>
I am sorry , could you please highlight the relevant portion for
Quoting Mohamed Shafi :
On 12 November 2010 18:39, Joern Rennecke wrote:
Quoting Mohamed Shafi :
So i have the following questions:
1. Why is that constraints are not matched here?
Please read the node "Register Classes" in doc/tm.texi .
I am sorry , could you please highlight the rele
On Thu, Nov 11, 2010 at 10:56 PM, Ian Lance Taylor wrote:
> Georg Johann Lay writes:
>
>> Suppose an backend implements some unspec_volatile (UV) and has a
>> destinct understanding of what it should be.
>>
>> If other parts of the compiler don't know exactly what to do, it's a
>> potential sourc
Georg Johann Lay writes:
> What about liveness? No hard reg, pseudo, mem will live avross the
> unspec volatile. Right?
As Paul noted, this is incorrect.
> Might debug info cross unspec volatiles?
> Can the back end take the decision?
I don't understand what you mean here. It is not the case
On 12/11/2010 12:51, Mohamed Shafi wrote:
> All the constraints are one letter constraints for my target. Here 'W'
> is for symbol_ref and all others are register constraints. So for a
> particular combination when operand 0 is 'a' and operand 1 is 'W' i
> got the following ICE :
> I get this ICE
Quoting Dave Korn :
By "combine constraints", you mean "Omit all the commas between
alternatives, mash them together in a single string, and expect GCC
to permute
all the possible combinations"?
I didn't know that was possible at all. I thought GCC would
interpret those
as multi-le
Georg Johann Lay writes:
> Implementation as outlined as in
>
> http://gcc.gnu.org/ml/gcc/2010-11/msg00222.html
That message describes an implementation rather than a language change.
What is the corresponding language change?
(My personal answer is almost certainly going to be "no." I don't s
Georg Johann Lay writes:
> Hook in and discuss
>
> http://gcc.gnu.org/ml/gcc/2010-11/msg00232.html
Discuss what? I see at least three different ideas in that message.
Ian
Alex Turjan writes:
> Can anybody give me a hint on where (perhaps some branch) I can find
> functionality
> which allows during scheduling to un-schedule an instruction? I basically
> need
> a function that does the oposite of schedule_insn.
>
> During scheduling I want o schedule_insn(INSN
On 12/11/2010 15:32, Joern Rennecke wrote:
> Quoting Dave Korn :
>
>> By "combine constraints", you mean "Omit all the commas between
>> alternatives, mash them together in a single string, and expect GCC
>> to permute
>> all the possible combinations"?
>>
>> I didn't know that was possible a
Quoting Dave Korn :
Is that documented anywhere? I couldn't find it in the
constraints chapters
of the internals manual.
It's in the reload source code...
And there are subtle register priority implications of having merged or
separate alternatives.
Sorry, is this separate from the
On 12/11/2010 16:06, Joern Rennecke wrote:
> Quoting Dave Korn:
>
>> Is that documented anywhere? I couldn't find it in the constraints
>> chapters of the internals manual.
>
> It's in the reload source code...
Had a good look through and still couldn't find it, would you mind giving me
a poi
Quoting Dave Korn :
Had a good look through and still couldn't find it, would you mind
giving me
a pointer to where should I start reading?
reload.c line 4094 (as of revision 18)
This is the beta release of binutils 2.21.51.0.1 for Linux, which is
based on binutils 2010 1110 in CVS on sourceware.org plus various
changes. It is purely for Linux.
All relevant patches in patches have been applied to the source tree.
You can take a look at patches/README to see what have been
2010/11/13 Витя Истомин :
> Hi, guys. I've got stuck about using dynamic_cast across .so boundaries and
> I've investigated this question. So I'm surprised that when I'm not using
> RTLD_GLOBAL, your new ABI realization f*cked out dynamic_cast, typeid, and
> (!) throw/catch.
>
> Are you in mind,
Using the new CLooG.org support in gcc trunk, I was able to benchmark the
performance
of gcc trunk built against ppl 0.11 and either legacy cloog support (built
against ppl 0.10.2)
or cloog-isl support. The benchmarks were done on x86_64-apple-darwin10 with...
Compile Command : gfortran -ffast
A quick check with...
gfortran -ffast-math -funroll-loops -msse3 -O3 -fgraphite-identity
-ftree-vectorizer-verbose=2 air.f90 -o air >& air_cloog_legacy.txt
grep "LOOP VECTORIZED" air_cloog_legacy.txt | wc -l
5
gfortran -ffast-math -funroll-loops -msse3 -O3 -fgraphite-identity
-ftree-ve
Hello,
* Joseph S. Myers wrote on Sat, Nov 06, 2010 at 11:41:17PM CET:
> On Wed, 3 Nov 2010, Ian Lance Taylor wrote:
>
> > There is also the libgo directory. The contents of libgo/go are a copy
> > of the standard Go library and I don't think a review of that would be
> > useful. But it would b
28 matches
Mail list logo
