--- Comment #7 from dje at gcc dot gnu dot org 2006-04-13 15:37 ---
Paolo, I'm not sure why you think that two splats and two adds is too
expensive. Memory is moving farther and farther away from the processor.
--
dje at gcc dot gnu dot org changed:
What|Re
--- Comment #3 from dje at gcc dot gnu dot org 2006-04-14 17:39 ---
I don't know how you got to this point at all. GNU Java currently is not
supported on AIX.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26275
--- Comment #4 from dje at gcc dot gnu dot org 2006-04-14 17:52 ---
The testcase works for me. I am a little confused about what you actually have
installed if you have "gcc" and "g++" installed with different names (with and
without -4.0.2). Do you have a consist
--- Comment #4 from dje at gcc dot gnu dot org 2006-04-25 04:24 ---
This optimization occurred in 4.1 between life2 and lreg:
life2:
(insn 24 23 25 2 (set (reg:SI 126)
(const_int -503316480 [0xe200])) 279 {*movsi_internal1}
(nil)
(nil))
(insn 30 29 31 2
--- Comment #9 from dje at gcc dot gnu dot org 2006-04-25 16:09 ---
By the way, while Andrew and my patch does produce correct code, it reverts to
the original behavior of the constant propagating into the AND between life2
and lreg, not between life1 and combine. Somehow combine is
--- Comment #12 from dje at gcc dot gnu dot org 2006-04-26 14:33 ---
Subject: Bug 27282
Author: dje
Date: Wed Apr 26 14:33:49 2006
New Revision: 113275
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113275
Log:
2006-04-26 David Edelsohn <[EMAIL PROTECTED]>
--- Comment #14 from dje at gcc dot gnu dot org 2006-04-26 17:57 ---
Subject: Bug 27282
Author: dje
Date: Wed Apr 26 17:57:03 2006
New Revision: 113278
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113278
Log:
PR middle-end/27282
* gcc.c-torture/compile
--- Comment #12 from dje at gcc dot gnu dot org 2006-05-04 13:52 ---
Subject: Bug 26481
Author: dje
Date: Thu May 4 13:52:45 2006
New Revision: 113525
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113525
Log:
PR target/26481
* config/rs6000/r
--- Comment #13 from dje at gcc dot gnu dot org 2006-05-04 13:55 ---
I was waiting for feedback from the original reporter, which never was
supplied. I have committed the patch on mainline to the 4.1 branch. I do not
have access to an AIX 5.1 system and without more details, it is
--- Comment #15 from dje at gcc dot gnu dot org 2006-05-04 20:41 ---
Subject: Bug 26481
Author: dje
Date: Thu May 4 20:40:52 2006
New Revision: 113539
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113539
Log:
PR target/26481
* config/rs6000/r
--- Comment #16 from dje at gcc dot gnu dot org 2006-05-04 20:44 ---
Subject: Bug 26481
Author: dje
Date: Thu May 4 20:43:57 2006
New Revision: 113540
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113540
Log:
PR target/26481
* config/rs6000/r
--- Comment #4 from dje at gcc dot gnu dot org 2006-05-09 17:09 ---
Subject: Bug 26545
Author: dje
Date: Tue May 9 17:09:06 2006
New Revision: 113659
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113659
Log:
PR target/26545
* config/rs6000
--- Comment #4 from dje at gcc dot gnu dot org 2006-05-09 17:12 ---
AIX 4.2 does not have enough support for weak symbols to compile C++ shared
library correctly. You must use --disable-shared on AIX 4.2.
--
dje at gcc dot gnu dot org changed:
What|Removed
--- Comment #5 from dje at gcc dot gnu dot org 2006-05-09 17:15 ---
Subject: Bug 26545
Author: dje
Date: Tue May 9 17:14:56 2006
New Revision: 113660
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113660
Log:
PR target/26545
* config/rs6000
--- Comment #6 from dje at gcc dot gnu dot org 2006-05-09 17:15 ---
patch applied.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
Status|NEW
--- Comment #4 from dje at gcc dot gnu dot org 2006-05-09 17:20 ---
You might be able to combine the address generation with the load in
legitimize_reload_address. See the existing code for SYMBOL_REF.
--
dje at gcc dot gnu dot org changed:
What|Removed
--- Comment #6 from dje at gcc dot gnu dot org 2006-05-10 14:56 ---
The section anchors feature does not like "__FUNCTION__" or "__func__" as an
inlined asm argument.
Also, "some rs6000 work" is not very informative or useful.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27528
--- Comment #1 from dje at gcc dot gnu dot org 2006-05-10 21:33 ---
I can disable section anchors for Ada, similar to Objective C and Objective
C++, but this failure likely means that there is a bug in the Trees generated
by GNU Ada.
--
dje at gcc dot gnu dot org changed
--- Comment #4 from dje at gcc dot gnu dot org 2006-05-11 15:13 ---
It looks like something in exception handling is creating a new object very
late, after reload, and the new address is not validized before being handed to
replace_equiv_address().
Should explow.c:use_anchored_address
--- Comment #8 from dje at gcc dot gnu dot org 2006-05-15 15:30 ---
Andreas or AJ,:
Can you test with gcc/except.c:output_ttype() changed from expand_normal() to
expand_expr (..., EXPAND_INITIALIZER, ...) ?
--
dje at gcc dot gnu dot org changed:
What|Removed
--- Comment #12 from dje at gcc dot gnu dot org 2006-05-18 15:56 ---
Users are suppose to control the architecture and code generation through
-mcpu=XXX, which sets the architecture options in self-consistent ways for the
target. On Darwin, -mcpu=G5 sets -mpowerpc64; on Linux
--- Comment #3 from dje at gcc dot gnu dot org 2006-05-25 17:27 ---
If the failures are associated with enabling section anchors by default, then
this has uncovered another problem with the section anchor support. Do you
have any more information about what is failing? Could this be
--- Comment #11 from dje at gcc dot gnu dot org 2006-05-25 17:34 ---
What is happening with this patch? Are these test results considered good?
Will this patch be submitted?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27536
--- Comment #20 from dje at gcc dot gnu dot org 2006-05-26 14:07 ---
The patch you reference enables section anchors by default. Neither AIX nor
PPC Linux show new Gfortran testsuite failures from the use of section anchors.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27683
--- Comment #3 from dje at gcc dot gnu dot org 2006-06-01 14:20 ---
dup
*** This bug has been marked as a duplicate of 27536 ***
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #12 from dje at gcc dot gnu dot org 2006-06-01 14:20 ---
*** Bug 27862 has been marked as a duplicate of this bug. ***
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #6 from dje at gcc dot gnu dot org 2006-06-05 21:25 ---
I still think this looks like an alignment problem. Without section anchors
GCC generates:
.lcomm domain_array,13916,16
.type domain_array, @object
.lcomm N_domains,4,4
.type
--- Comment #10 from dje at gcc dot gnu dot org 2006-06-06 15:10 ---
The auto-vectorizer is a Tree-SSA pass. The section anchors are an RTL pass.
I do not understand why the alignment of the vectorized variables is not known
at section anchor creation time.
--
http://gcc.gnu.org
--- Comment #13 from dje at gcc dot gnu dot org 2006-06-06 15:22 ---
We're performing the auto-vectorization in unit-at-a-time-mode, so maybe we
need to recompile the other functions. It seems that we're going to encounter
more problems along these lines with LTO.
--- Comment #2 from dje at gcc dot gnu dot org 2006-06-07 02:31 ---
Can you provide any information about what initially produces the
(subreg:DF (reg:DI)) ?
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #16 from dje at gcc dot gnu dot org 2006-06-07 21:39 ---
I checked with the IBM XLC team and they speculatively increase the alignment
of variables that could be auto-vectorized, so that gives another vote for that
method. They did mention that whole-program IPA allows them
--- Comment #9 from dje at gcc dot gnu dot org 2006-06-17 01:37 ---
The bug is in flow.c and fixed by the new df.c rewrite of dataflow. Ken and I
tripped over the same problem.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #4 from dje at gcc dot gnu dot org 2006-06-26 18:51 ---
This appears to be a problem in the insvdi_internal2 and insvdi_internal3
combiner patterns.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #6 from dje at gcc dot gnu dot org 2006-06-27 01:34 ---
rs6000.c:insvdi_rshift_rlwimi_p() is wrong, specifically the lines checking the
shift count.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28170
--- Comment #7 from dje at gcc dot gnu dot org 2006-06-27 01:40 ---
I think the correct test is something like
Index: rs6000.c
===
--- rs6000.c(revision 115003)
+++ rs6000.c(working copy)
@@ -9794,8 +9794,8
--- Comment #6 from dje at gcc dot gnu dot org 2006-06-30 13:39 ---
This may be due to spe_build_register_parallel() called from
rs6000_function_value().
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #7 from dje at gcc dot gnu dot org 2006-07-05 20:03 ---
I think this simply is a case of not allowing TFmode in PRE_INC addresses:
Index: rs6000.c
===
--- rs6000.c(revision 115196)
+++ rs6000.c(working
--- Comment #11 from dje at gcc dot gnu dot org 2006-07-06 14:37 ---
The original code could not be "totally broken" unless Alan's code also is
totally broken because he computes many of the same tests. Also, Alan does not
include an important test from the original co
--- Comment #13 from dje at gcc dot gnu dot org 2006-07-06 19:00 ---
Subject: Bug 28170
Author: dje
Date: Thu Jul 6 18:59:58 2006
New Revision: 115229
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=115229
Log:
2006-07-06 David Edelsohn <[EMAIL PROTECTED]>
--- Comment #8 from dje at gcc dot gnu dot org 2006-07-06 19:00 ---
Subject: Bug 28150
Author: dje
Date: Thu Jul 6 18:59:58 2006
New Revision: 115229
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=115229
Log:
2006-07-06 David Edelsohn <[EMAIL PROTECTED]>
--- Comment #8 from dje at gcc dot gnu dot org 2006-07-08 03:13 ---
The patch looks consistent with the rest of the e500 design, although I am not
familiar enough with the e500 ISA to determine if the code will be correct
under all circumstances. I'm willing to approve the pat
--- Comment #10 from dje at gcc dot gnu dot org 2006-07-11 13:28 ---
This now causes bootstrap failure on primary platforms.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #8 from dje at gcc dot gnu dot org 2006-07-11 16:59 ---
Switching actually causes problems on AIX due to AIX math.h header enabling
additional macros with types that conflict with GCC.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19115
--- Comment #17 from dje at gcc dot gnu dot org 2006-07-14 17:44 ---
Subject: Bug 27287
Author: dje
Date: Fri Jul 14 17:44:27 2006
New Revision: 115451
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=115451
Log:
2006-07-14 Eliot Dresselhaus <[EMAIL PROTECTED]&g
--- Comment #18 from dje at gcc dot gnu dot org 2006-07-14 17:47 ---
Remove 4.2 regression designation.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #3 from dje at gcc dot gnu dot org 2006-07-14 19:11 ---
This is because the E500 DOUBLE support wasn't prepared for complex doubles.
Index: rs6000.c
===
--- rs6000.c(revision 115431)
+++ rs6
--- Comment #4 from dje at gcc dot gnu dot org 2006-07-14 19:21 ---
rs6000_legitimate_offset_address_p() accepts constant offsets:
case V4HImode:
case V2SImode:
case V1DImode:
case V2SFmode:
/* SPE vector modes. */
return SPE_CONST_OFFSET_OK (offset);
It
--- Comment #4 from dje at gcc dot gnu dot org 2006-07-17 01:43 ---
Double Complex on e500 with double GPRs fundamentally does not interact well
with GCC's current design. GCC does not want to have a double placed in the
same register that can contain an int if the width o
--- Comment #5 from dje at gcc dot gnu dot org 2006-07-17 02:05 ---
Created an attachment (id=11902)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=11902&action=view)
consistently calculate nregs for FLOAT_MODES
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24036
--- Comment #1 from dje at gcc dot gnu dot org 2008-11-24 21:57 ---
Yep. confirmed.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
Status
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: dje at gcc dot gnu dot org
GCC build triplet: powerpc*-*-*
GCC host triplet: powerpc*-*-*
GCC target triplet: powerpc*-*-*
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38731
--- Comment #2 from dje at gcc dot gnu dot org 2009-01-05 19:54 ---
For example, the following, when declared locally:
typedef char Str_30 [31];
Str_30 Str_1_Par_Ref;
Str_30 Str_2_Par_Ref;
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38731
--- Comment #1 from dje at gcc dot gnu dot org 2009-05-21 12:40 ---
Confirmed
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
CC
--- Comment #4 from dje at gcc dot gnu dot org 2009-06-01 21:49 ---
Local variable debugging is broken for AIX stabs as well.
On AIX, the DECL and the insn contain different offsets from virtual-stack-vars
for the same variable:
(mem/c/i:SI (plus:SI (reg/f:SI 115 virtual-stack-vars
--- Comment #5 from dje at gcc dot gnu dot org 2009-06-02 12:51 ---
cfgexpand.c:expand_used_vars() expands and instantiates virtual-stack-vars for
INSNs when it walks the partion map:
for (i = 0; i < SA.map->num_partitions; i++)
{
tree var = partition_to_var (SA.
--- Comment #2 from dje at gcc dot gnu dot org 2009-06-04 16:18 ---
AIX started miscomparing in libgcc as well.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #5 from dje at gcc dot gnu dot org 2009-06-04 17:11 ---
Jakub mentioned that Alexandre patch added comparison of libgcc, which may not
have been compared before. But I successfully bootstrapped with Alexandre's
patch the previous day.
--
http://gcc.gnu.org/bug
--- Comment #6 from dje at gcc dot gnu dot org 2009-06-04 17:17 ---
Alexandre committed the change in revision 148080 and I successfully
bootstrapped AIX yesterday with revision 148105.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40338
--- Comment #7 from dje at gcc dot gnu dot org 2009-06-04 19:25 ---
Okay, I think these are separate bugs. The HP-PA error may be due to
Alexandre's change.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40338
ormal
Priority: P3
Component: bootstrap
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: dje at gcc dot gnu dot org
GCC build triplet: powerpc-ibm-aix5.3.0.0
GCC host triplet: powerpc-ibm-aix5.3.0.0
GCC target triplet: powerpc-ibm-aix5.3.0.0
http://gcc.gn
--- Comment #1 from dje at gcc dot gnu dot org 2009-06-04 19:27 ---
confirmed.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
CC
--- Comment #14 from dje at gcc dot gnu dot org 2009-06-15 23:06 ---
This patch also fixes the gcc.c-torture/compile/pr34688.c failures. RTL-PRE
finds RTL with deep LABEL_REFs. When it creates a move, the emit_use and the
REG_NOTE on the move itself share RTL.
I suspect we need to
--- Comment #8 from dje at gcc dot gnu dot org 2009-06-30 14:50 ---
Now I see failures when building libgcc:
../../.././gcc/dp-bit.c: In function '__pack_d':
../../.././gcc/dp-bit.c:268:39: internal compiler error: in simplify_subreg, at
simplify-rtx.c:4966
--
http://g
--- Comment #9 from dje at gcc dot gnu dot org 2009-06-30 15:07 ---
The traceback leads to the store flag change.
(gdb) print op
$3 = (rtx) 0x30153558
(gdb) pr
(reg:SI 234)
(gdb) print innermode
$4 = DImode
(gdb) print outermode
$5 = SImode
#0 fancy_abort (file=0x10e8a5b8 "/far
--- Comment #12 from dje at gcc dot gnu dot org 2009-06-30 17:01 ---
I tried bootstrapping with that patch and it did not solve the bootstrap
failure in simplify_subreg.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40597
--- Comment #13 from dje at gcc dot gnu dot org 2009-06-30 18:16 ---
The failure appears to need a compiler built in 32 bit mode.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40597
--- Comment #5 from dje at gcc dot gnu dot org 2009-07-15 12:38 ---
The patch seems reasonable, but it exposes a latent bug in Nathan's mmultiple
patch that was not tested because of this bug.
This patch should have been tested and posted to the GCC mailinglist
referencing th
--- Comment #6 from dje at gcc dot gnu dot org 2009-07-15 12:50 ---
The fix for this actually is more involved. Nathan had a patch last October
which exposed similar bootstrap problems. He posted a revised patch
http://gcc.gnu.org/ml/gcc-patches/2008-11/msg01103.html
--
dje at
Priority: P3
Component: tree-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: dje at gcc dot gnu dot org
GCC build triplet: power-ibm-aix
GCC host triplet: powerpc-ibm-aix
GCC target triplet: power-ibm-aix
http://gcc.gnu.org/bugzilla
--- Comment #1 from dje at gcc dot gnu dot org 2009-08-12 17:35 ---
This started failing after I started compiling with MPC library. Is this an
MPC bug or interaction bug?
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #26 from dje at gcc dot gnu dot org 2005-11-11 17:59 ---
Subject: Bug 24644
Author: dje
Date: Fri Nov 11 17:59:54 2005
New Revision: 106791
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=106791
Log:
PR 24644
* common.opt (Wvolatile-register-v
--- Comment #9 from dje at gcc dot gnu dot org 2005-11-14 20:28 ---
I do not have access to an AIX 4.3.2 system, but I compiled the reconstructed C
testcase with:
gcc-3.3.3 on AIX 5.1
gcc-3.4.4 on AIX 5.1
gcc-4.0.2 on AIX 5.1
gcc-4.1 experimental on AIX 5.2
All four compilers produced
--- Comment #4 from dje at gcc dot gnu dot org 2005-11-15 22:27 ---
Patch
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc
--- Comment #5 from dje at gcc dot gnu dot org 2005-11-16 15:59 ---
Subject: Bug 24772
Author: dje
Date: Wed Nov 16 15:59:27 2005
New Revision: 107085
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=107085
Log:
PR target/24772
* config/rs6000/predi
--- Comment #6 from dje at gcc dot gnu dot org 2005-11-16 19:49 ---
Patch committed. Should be fixed.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #7 from dje at gcc dot gnu dot org 2005-11-17 00:59 ---
That failure is fixed by the zero_reg_mem predicate changes from March:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/rs6000/predicates.md.diff?cvsroot=gcc&r1=1.6&r2=1.7
This is why I didn't want t
--- Comment #2 from dje at gcc dot gnu dot org 2005-11-21 02:08 ---
This one is nonsense. We do not have a language type for Treelang. So what?
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #1 from dje at gcc dot gnu dot org 2005-11-21 02:09 ---
We just need to check that we have enough elts in the VEC.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #2 from dje at gcc dot gnu dot org 2005-11-21 02:54 ---
Index: predicates.md
===
I think the following patch fixes the checking failure. I will bootstrap
overnight.
--- predicates.md (revision 107244
--- Comment #3 from dje at gcc dot gnu dot org 2005-11-21 16:12 ---
Subject: Bug 24953
Author: dje
Date: Mon Nov 21 16:12:32 2005
New Revision: 107311
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=107311
Log:
PR target/24953
* config/rs6000/predi
--- Comment #4 from dje at gcc dot gnu dot org 2005-11-21 18:04 ---
Subject: Bug 24953
Author: dje
Date: Mon Nov 21 18:04:14 2005
New Revision: 107314
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=107314
Log:
PR target/24953
* config/rs6000/predi
--- Comment #5 from dje at gcc dot gnu dot org 2005-11-21 18:52 ---
Fixed.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #4 from dje at gcc dot gnu dot org 2005-11-21 18:57 ---
Treelang is a sample language. There is no language type defined for it in any
ABI and no reason to create a type. This actually is a good test that the
assert works for undefined languages.
--
http://gcc.gnu.org
--- Comment #4 from dje at gcc dot gnu dot org 2005-11-24 03:56 ---
Many people are reporting that they cannot bootstrap GCC with IBM's VisualAge
compiler. This appears to be a problem with VAC/XLC, not with GCC. This bug
should be reported to IBM as a XLC bug. In the meantime
--- Comment #1 from dje at gcc dot gnu dot org 2005-11-27 22:06 ---
Build GCC on AIX as 64-bit application is not a tested configuration. A better
way to set the flag for AR probably is in new config/mh-aix file or with
OBJECT_MODE=64 environment variable.
You probably need to do this
--- Comment #3 from dje at gcc dot gnu dot org 2005-12-01 23:07 ---
Yes, I think this is the same as PR 24997.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #7 from dje at gcc dot gnu dot org 2005-12-07 13:19 ---
Changing the behavior of GCC -print-search-dirs is a bad idea and will break
other applications relying on the current behavior. Introducing a major
incompatibility into GCC is not a solution and I strongly oppose the
--- Comment #19 from dje at gcc dot gnu dot org 2006-01-03 14:54 ---
Is this still a regression in 4.2? The code generated for PowerPC now looks
better, possibly due to the new reassociation pass.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23181
--- Comment #22 from dje at gcc dot gnu dot org 2006-01-03 15:22 ---
Is it possible for backends to inquire how many reciprocals were discovered?
PowerPC and IA-64 have reciprocal instructions and could benefit by choosing to
generate reciprocal earlier with additional information
--- Comment #24 from dje at gcc dot gnu dot org 2006-01-03 15:50 ---
Comment #21 mentions three divisions for the optimization to be more
profitable, but that information is lost by the time GCC reaches the expanders.
Minimally, it would be helpful to know that the reciprocal
--- Comment #2 from dje at gcc dot gnu dot org 2006-01-03 16:25 ---
Not powerpc-specific.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
GCC build triplet
--- Comment #2 from dje at gcc dot gnu dot org 2006-01-03 16:27 ---
Not powerpc-specific.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
GCC build triplet
--- Comment #2 from dje at gcc dot gnu dot org 2006-01-03 16:29 ---
Not powerpc-specific.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
GCC build triplet
--- Comment #5 from dje at gcc dot gnu dot org 2006-01-05 18:26 ---
Andrew, you are confusing the issue by quoting the AIX documentation. Darwin
and Linux implement different semantics from AIX. On Darwin and Linux, the
most significant doubleword always is correctly rounded, so it
--- Comment #5 from dje at gcc dot gnu dot org 2006-01-06 20:18 ---
The patched compiler does build and does fix the testcase.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #7 from dje at gcc dot gnu dot org 2006-01-07 22:23 ---
Subject: Bug 25662
Author: dje
Date: Sat Jan 7 22:23:27 2006
New Revision: 109456
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109456
Log:
2005-01-07 Ian Lance Taylor
David Edelsohn
--- Comment #8 from dje at gcc dot gnu dot org 2006-01-08 20:54 ---
Subject: Bug 25662
Author: dje
Date: Sun Jan 8 20:54:28 2006
New Revision: 109476
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109476
Log:
2006-01-08 Ian Lance Taylor
David Edelsohn
--- Comment #9 from dje at gcc dot gnu dot org 2006-01-08 20:55 ---
Subject: Bug 25662
Author: dje
Date: Sun Jan 8 20:55:39 2006
New Revision: 109477
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=109477
Log:
2006-01-08 Ian Lance Taylor
David Edelsohn
--- Comment #10 from dje at gcc dot gnu dot org 2006-01-09 02:19 ---
Fix committed to all mainline, 4.1, 4.0.
--
dje at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #5 from dje at gcc dot gnu dot org 2006-01-11 02:20 ---
What is the specific testcase compiled by XLC? What version of XLC? And what
options were used?
I cannot reproduce strength reduction of a floating point multiply to floating
point adds with a testcase that uses a
1 - 100 of 559 matches
Mail list logo
