On Fri, Aug 30, 2024 at 2:24 PM Richard Biener wrote:
>
> On Fri, 30 Aug 2024, Manolis Tsamis wrote:
>
> > For more context please see
> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116352#c11.
> >
> > I've sent this as an RFC because it's hard to be s
On Thu, Aug 29, 2024 at 9:16 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > On Fri, Aug 16, 2024 at 5:33 PM Richard Sandiford
> > wrote:
> >> Manolis Tsamis writes:
> >> > +}
> >> > +
> >> > + virtual
h the code,
this looked like a general issue to me and hence I'm proposing this
fix.
Thanks,
Manolis
On Fri, Aug 30, 2024 at 1:46 PM Manolis Tsamis wrote:
>
> Currently we do SLP CSE after permute optimization using a single map across
> all SLP instances. These SLP instances c
map per SLP instance.
gcc/testsuite/ChangeLog:
* gcc.dg/pr116352.c: New test.
Signed-off-by: Manolis Tsamis
---
gcc/testsuite/gcc.dg/pr116352.c | 18 ++
gcc/tree-vect-slp.cc| 12 +++-
2 files changed, 25 insertions(+), 5 deletions(-)
create
Hi Richard,
Thanks a lot for the extensive review and useful suggestions (and
sorry for the late reply). I have implemented most of these for a next
version, so let me address your individual comments below:
On Fri, Aug 16, 2024 at 5:33 PM Richard Sandiford
wrote:
>
> Manolis Tsamis
vt.cc (noce_convert_multiple_sets): Disallow call insns.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/pr116358.c: New test.
Signed-off-by: Manolis Tsamis
---
gcc/ifcvt.cc| 2 +-
gcc/testsuite/gcc.target/aarch64/pr116358.c | 15 +++
2 files changed, 16 insertions(
On Tue, Aug 13, 2024 at 10:48 PM Jeff Law wrote:
>
>
>
> On 8/13/24 5:57 AM, Manolis Tsamis wrote:
> > Now that more operations are allowed for noce_convert_multiple_sets, we
> > need to
> > check noce_can_force_operand on the sequence before calling
> >
372.c: New test.
* gcc.dg/pr116405.c: New test.
Signed-off-by: Manolis Tsamis
---
gcc/ifcvt.cc| 22 ++
gcc/testsuite/gcc.dg/pr116372.c | 13 +
gcc/testsuite/gcc.dg/pr116405.c | 17 +
3 files changed, 48 insertions(+), 4
: Manolis Tsamis
---
gcc/ifcvt.cc | 6 ++-
gcc/testsuite/gcc.target/i386/pr116353.c | 55
2 files changed, 59 insertions(+), 2 deletions(-)
create mode 100644 gcc/testsuite/gcc.target/i386/pr116353.c
diff --git a/gcc/ifcvt.cc b/gcc/ifcvt.cc
.
* gcc.target/aarch64/avoid-store-forwarding-3.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-4.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-5.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v5:
- Fix bug with BIG_ENDIAN targets
s no change needed for that comment.
> I can't comment on ifcvt changes to say if they're OK or not otherwise,
> sorry.
>
> >
> > Thanks,
> > Philipp.
> >
> >
> > On Fri, 26 Jul 2024 at 12:50, Sam James wrote:
> >>
> >> Mano
Pinging this for a review and/or further feedback.
Thanks,
Manolis
On Wed, Jun 26, 2024 at 3:06 PM Manolis Tsamis wrote:
>
> This change checks when a two_operators SLP node has multiple occurrences of
> the same statement (e.g. {A, B, A, B, ...}) and tries to rearrange the
> o
(struct noce_multiple_sets_info): Introduce new struct
noce_multiple_sets_info to store info for noce_convert_multiple_sets.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/ifcvt_multiple_sets_rewire.c: New test.
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc
et/aarch64/ifcvt_multiple_sets_arithm.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v5:
- Loop over SEQ and check modified_in_p for all instructions.
- Fix x86-related bug when SEQ modifies COND.
gcc/ifcvt.cc | 34 +++-
.../aarch64/ifcvt_multiple_set
f the
if_then_else emitted more robust.
gcc/ChangeLog:
* ifcvt.cc (check_for_cc_cmp_clobbers): Use modified_in_p instead.
(noce_convert_multiple_sets_1): Don't use seq2 if it clobbers cc_cmp.
Punt if seq clobbers cond. Refactor the code that sets read_comparison.
Signed-off-b
SEQ and check modified_in_p for all instructions.
- Fix x86-related bug when SEQ modifies COND.
Manolis Tsamis (3):
ifcvt: handle sequences that clobber flags in
noce_convert_multiple_sets
ifcvt: Allow more operations in multiple set if conversion
ifcvt: Handle multiple rewired
On Wed, Jun 5, 2024 at 2:00 PM Richard Sandiford
wrote:
>
> Sorry for the slow review.
>
> Manolis Tsamis writes:
> > This is an extension of what was done in PR106590.
> >
> > Currently if a sequence generated in noce_convert_multiple_sets clobbers the
> >
On Thu, Jul 11, 2024 at 1:03 AM Jeff Law wrote:
>
>
>
> On 6/3/24 5:34 AM, Manolis Tsamis wrote:
> > This is an extension of what was done in PR106590.
> >
> > Currently if a sequence generated in noce_convert_multiple_sets clobbers the
> > condition rtx (cc_
f the
if_then_else emitted more robust.
gcc/ChangeLog:
* ifcvt.cc (check_for_cc_cmp_clobbers): Use modified_in_p instead.
(noce_convert_multiple_sets_1): Don't use seq2 if it clobbers cc_cmp.
Punt if seq clobbers cond. Refactor the code that sets read_comparison.
Signed-off-b
(struct noce_multiple_sets_info): Introduce new struct
noce_multiple_sets_info to store info for noce_convert_multiple_sets.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/ifcvt_multiple_sets_rewire.c: New test.
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc
et/aarch64/ifcvt_multiple_sets_arithm.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v5:
- Loop over SEQ and check modified_in_p for all instructions.
- Fix x86-related bug when SEQ modifies COND.
gcc/ifcvt.cc | 34 +++-
.../aarch64/ifcvt_multiple_set
-related bug when SEQ modifies COND.
Manolis Tsamis (3):
[RFC] ifcvt: handle sequences that clobber flags in
noce_convert_multiple_sets
[RFC] ifcvt: Allow more operations in multiple set if conversion
[RFC] ifcvt: Handle multiple rewired regs and refactor
noce_convert_multiple_sets
gcc
On Mon, Jul 8, 2024 at 6:41 PM Andi Kleen wrote:
>
> > I have added a target hook for this in v4 of this patch. The hook
> > receives all the information about the stores, the load, the estimated
> > sequence cost and whether we expect to eliminate the load. With this
> > information the target sh
On Thu, Jun 13, 2024 at 7:18 PM Andi Kleen wrote:
>
> Manolis Tsamis writes:
> >
> > Assembly like this can appear with bitfields or type punning / unions.
> > On stress-ng when running the cpu-union microbenchmark the following
> > speedups
> > have
.
* gcc.target/aarch64/avoid-store-forwarding-3.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-4.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-5.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v4:
- Change pass scheduling to run after sched1
On Wed, Jun 5, 2024 at 11:07 AM Richard Biener wrote:
>
> On Tue, 4 Jun 2024, Manolis Tsamis wrote:
>
> > This change adds a function that checks for SLP nodes with multiple
> > occurrences
> > of the same statement (e.g. {A, B, A, B, ...}) and tries to rearrange the
n 26, 2024 at 3:06 PM Manolis Tsamis wrote:
>
> This change checks when a two_operators SLP node has multiple occurrences of
> the same statement (e.g. {A, B, A, B, ...}) and tries to rearrange the
> operands
> so that there are no duplicates. Two vec_perm expressions are then introdu
=98138
gcc/ChangeLog:
* tree-vect-slp.cc: Avoid duplicates in two_operators nodes.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/vect-slp-two-operator.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v2:
- Do not use predefined patterns; support rearrangement of
On Sun, Jun 9, 2024 at 5:29 PM Jeff Law wrote:
>
>
>
> On 6/7/24 4:31 PM, Jeff Law wrote:
>
> >
> > I've actually added it to my tester just to see if there's any fallout.
> > It'll take a week to churn through the long running targets that
> > bootstrap in QEMU, but the crosses should have data M
-2.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-3.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-4.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-5.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v3:
- Only emit SUBREG
On Mon, Jun 10, 2024 at 9:27 PM Philipp Tomsich
wrote:
>
> On Mon, 10 Jun 2024 at 20:03, Jeff Law wrote:
> >
> >
> >
> > On 6/10/24 1:55 AM, Manolis Tsamis wrote:
> >
> > >>
> > > There was an older submission of a load-pair specific pass
On Wed, Jun 5, 2024 at 11:07 AM Richard Biener wrote:
>
> On Tue, 4 Jun 2024, Manolis Tsamis wrote:
>
> > This change adds a function that checks for SLP nodes with multiple
> > occurrences
> > of the same statement (e.g. {A, B, A, B, ...}) and tries to rearrange the
On Sun, Jun 9, 2024 at 5:29 PM Jeff Law wrote:
>
>
>
> On 6/7/24 4:31 PM, Jeff Law wrote:
>
> >
> > I've actually added it to my tester just to see if there's any fallout.
> > It'll take a week to churn through the long running targets that
> > bootstrap in QEMU, but the crosses should have data M
On Sat, Jun 8, 2024 at 1:31 AM Jeff Law wrote:
>
>
>
> On 6/6/24 4:10 AM, Manolis Tsamis wrote:
> > This pass detects cases of expensive store forwarding and tries to avoid
> > them
> > by reordering the stores and using suitable bit insertion sequences.
> &g
On Fri, May 24, 2024 at 9:27 AM Richard Biener wrote:
>
> On Thu, 23 May 2024, Manolis Tsamis wrote:
>
> > This pass detects cases of expensive store forwarding and tries to avoid
> > them
> > by reordering the stores and using suitable bit insertion sequences.
>
-2.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-3.c: New test.
* gcc.target/aarch64/avoid-store-forwarding-4.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v2:
- Allow modes that are not scalar_int_mode.
- Introduce simple costing to avoid
On Wed, Jun 5, 2024 at 2:00 PM Richard Sandiford
wrote:
>
> Sorry for the slow review.
>
> Manolis Tsamis writes:
> > This is an extension of what was done in PR106590.
> >
> > Currently if a sequence generated in noce_convert_multiple_sets clobbers the
> >
-off-by: Manolis Tsamis
---
.../aarch64/vect-slp-two-operator.c | 42
gcc/tree-vect-slp.cc | 234 ++
2 files changed, 276 insertions(+)
create mode 100644 gcc/testsuite/gcc.target/aarch64/vect-slp-two-operator.c
diff --git a/gcc/testsuite
(struct noce_multiple_sets_info): Introduce new struct
noce_multiple_sets_info to store info for noce_convert_multiple_sets.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/ifcvt_multiple_sets_rewire.c: New test.
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc
bbers): Use modified_in_p instead.
(noce_convert_multiple_sets_1): Don't use seq2 if it clobbers cc_cmp.
Refactor the code that sets read_comparison.
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc | 106 -
unnecessary hardcoded list of allowed ops in
bb_ok_for_noce_convert_multiple_sets.
- Set need_cmov based on BB live_out instead of REG_DEAD notes.
- Fix preexisting issues and improve the code that sets read_comparison.
Manolis Tsamis (3):
[RFC] ifcvt: handle sequences that
et/aarch64/ifcvt_multiple_sets_arithm.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v4:
- Remove unnecessary hardcoded list of allowed ops in
bb_ok_for_noce_convert_multiple_sets.
- Set need_cmov based on BB live_out instead of REG_DEAD notes.
- Fix preexisting issues and i
On Fri, May 24, 2024 at 9:27 AM Richard Biener wrote:
>
> On Thu, 23 May 2024, Manolis Tsamis wrote:
>
> > This pass detects cases of expensive store forwarding and tries to avoid
> > them
> > by reordering the stores and using suitable bit insertion sequences.
>
:
* gcc.dg/fold-perm-2.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v2:
Use TYPE_SIZE (TREE_TYPE (TREE_TYPE (@))) instead of element_precision (@).
gcc/match.pd | 14 --
gcc/testsuite/gcc.dg/fold-perm-2.c | 16
2 files changed, 24
vector (r3, veci);
+}
+
+/* { dg-final { scan-tree-dump "VEC_PERM_EXPR.*{ 2, 7, 5, 0 }" "fre1" } } */
+/* { dg-final { scan-tree-dump-times "VEC_PERM_EXPR" 1 "fre1" } } */
--
2.44.0
On Fri, May 24, 2024 at 11:30 AM Richard Biener wrote:
>
> On Fri,
On Fri, May 24, 2024 at 10:46 AM Richard Biener wrote:
>
> On Fri, 24 May 2024, Manolis Tsamis wrote:
>
> > On Fri, May 24, 2024 at 9:31 AM Richard Biener wrote:
> > >
> > > On Wed, 22 May 2024, Manolis Tsamis wrote:
> > >
> > > > The match.
On Fri, May 24, 2024 at 9:31 AM Richard Biener wrote:
>
> On Wed, 22 May 2024, Manolis Tsamis wrote:
>
> > The match.pd patterns to merge two vector permutes into one fail when a
> > potentially no-op view convert expressions is between the two permutes.
> > This ch
-off-by: Manolis Tsamis
---
gcc/Makefile.in | 1 +
gcc/avoid-store-forwarding.cc | 554 ++
gcc/common.opt| 4 +
gcc/params.opt| 4 +
gcc/passes.def
:
* gcc.dg/fold-perm-2.c: New test.
Signed-off-by: Manolis Tsamis
---
gcc/match.pd | 14 --
gcc/testsuite/gcc.dg/fold-perm-2.c | 16
2 files changed, 24 insertions(+), 6 deletions(-)
create mode 100644 gcc/testsuite/gcc.dg/fold-perm-2.c
diff --git
On Fri, May 17, 2024 at 12:22 PM Richard Biener wrote:
>
> On Fri, 17 May 2024, Manolis Tsamis wrote:
>
> > Hi Richard,
> >
> > While I was re-testing the latest version of this patch I noticed that
> > it FAILs an AArch64 test, gcc.target/aarch64/subsp.c. W
lly find a way to solve the regression.
Any thoughts on this?
Thanks,
Manolis
On Thu, May 16, 2024 at 11:15 AM Richard Biener
wrote:
>
> On Tue, May 14, 2024 at 10:58 AM Manolis Tsamis
> wrote:
> >
> > New patch with the requested changes can be found below.
> >
&
Pinging this for GCC15.
Thanks
On Tue, Apr 23, 2024 at 1:47 PM Manolis Tsamis wrote:
>
>
> noce_convert_multiple_sets has been introduced and extended over time to
> handle
> if conversion for blocks with multiple sets. Currently this is focused on
> register moves and r
} */
+
+int foo(int *a, int j)
+{
+ int k = j - 1;
+ return a[j - 1] == a[k];
+}
+
+int bar(int *a, int j)
+{
+ int k = j - 1;
+ return (&a[j + 1] - 2) == &a[k];
+}
--
2.44.0
On Tue, Apr 23, 2024 at 1:33 PM Manolis Tsamis wrote:
>
> The original motivation for this pattern
On Thu, May 2, 2024 at 4:00 PM Richard Biener wrote:
>
> On Tue, 23 Apr 2024, Manolis Tsamis wrote:
>
> > The original motivation for this pattern was that the following function
> > does
> > not fold to 'return 1':
> >
> > int foo(int *a, int j
On Thu, Apr 25, 2024 at 2:40 AM Hans-Peter Nilsson wrote:
>
> On Tue, 23 Apr 2024, Manolis Tsamis wrote:
> > diff --git a/gcc/testsuite/gcc.target/aarch64/ifcvt_multiple_sets_arithm.c
> > b/gcc/testsuite/gcc.target/aarch64/ifcvt_multiple_sets_arithm.c
> ...
> > +/*
On Thu, Nov 23, 2023 at 11:01 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > The existing implementation of need_cmov_or_rewire and
> > noce_convert_multiple_sets_1 assumes that sets are either REG or SUBREG.
> > This commit enchances them so they can han
On Thu, Oct 19, 2023 at 10:46 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > Currently the operations allowed for if conversion of a basic block with
> > multiple sets are few, namely REG, SUBREG and CONST_INT (as controlled by
> > bb_ok_for_noce_convert_mult
On Thu, Oct 19, 2023 at 10:41 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > This is an extension of what was done in PR106590.
> >
> > Currently if a sequence generated in noce_convert_multiple_sets clobbers the
> > condition rtx (cc_cmp or rev_
et/aarch64/ifcvt_multiple_sets_arithm.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v4:
- Remove unnecessary hardcoded list of allowed ops in
bb_ok_for_noce_convert_multiple_sets.
- Set need_cmov based on BB live_out instead of REG_DEAD notes.
- Fix preexisting issues and i
(struct noce_multiple_sets_info): Introduce new struct
noce_multiple_sets_info to store info for noce_convert_multiple_sets.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/ifcvt_multiple_sets_rewire.c: New test.
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc
bbers): Use modified_in_p instead.
(noce_convert_multiple_sets_1): Don't use seq2 if it clobbers cc_cmp.
Refactor the code that sets read_comparison.
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc | 106 -
unnecessary hardcoded list of allowed ops in
bb_ok_for_noce_convert_multiple_sets.
- Set need_cmov based on BB live_out instead of REG_DEAD notes.
- Fix preexisting issues and improve the code that sets read_comparison.
Manolis Tsamis (3):
[RFC] ifcvt: handle sequences that
ned-off-by: Manolis Tsamis
---
gcc/match.pd| 30 ++
gcc/testsuite/gcc.dg/pr109393.c | 16
2 files changed, 46 insertions(+)
create mode 100644 gcc/testsuite/gcc.dg/pr109393.c
diff --git a/gcc/match.pd b/gcc/match.pd
index d401e7503e6..1
On Fri, Apr 5, 2024 at 3:43 PM Richard Biener
wrote:
>
> On Fri, Apr 5, 2024 at 2:28 PM Manolis Tsamis wrote:
> >
> > If we consider code like:
> >
> > if (bar1 == x)
> > return foo();
> > if (bar2 != y)
> > return foo();
>
: Change flag to -fno-tree-ifcombine.
* gcc.dg/uninit-pred-6_c.c: Remove inconsistent check.
* gcc.target/aarch64/pr102793.c: New test.
Signed-off-by: Manolis Tsamis
---
gcc/common.opt | 4 +++
gcc/doc/invoke.texi | 5
gcc
On Thu, Nov 23, 2023 at 11:01 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > The existing implementation of need_cmov_or_rewire and
> > noce_convert_multiple_sets_1 assumes that sets are either REG or SUBREG.
> > This commit enchances them so they can han
On Tue, Nov 28, 2023 at 12:53 PM Jakub Jelinek wrote:
>
> On Tue, Nov 28, 2023 at 11:45:58AM +0200, Manolis Tsamis wrote:
> > > But, while get_single_def_in_bb checks for
> > > if (DF_INSN_LUID (def) > DF_INSN_LUID (insn))
> > > return NULL;
On Tue, Nov 28, 2023 at 12:12 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > On Thu, Nov 23, 2023 at 11:01 PM Richard Sandiford
> > wrote:
> >>
> >> Manolis Tsamis writes:
> >> > The existing implementation of need_cmov_or_rewire and
Hi Jakub,
Thanks a lot for tracking this down and providing both testcases and a
fix. Some thoughts below.
On Mon, Nov 27, 2023 at 10:52 PM Jakub Jelinek wrote:
>
> On Mon, Oct 16, 2023 at 01:11:01PM -0600, Jeff Law wrote:
> > > gcc/ChangeLog:
> > >
> > > * Makefile.in: Add fold-mem-offsets.
On Thu, Nov 23, 2023 at 11:01 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > The existing implementation of need_cmov_or_rewire and
> > noce_convert_multiple_sets_1 assumes that sets are either REG or SUBREG.
> > This commit enchances them so they can han
I have bootstrapped and checked for no testsuite regressions on x86 and aarch64.
Thanks,
Manolis
On Tue, Nov 21, 2023 at 8:00 PM Manolis Tsamis wrote:
>
> The existing implementation of need_cmov_or_rewire and
> noce_convert_multiple_sets_1 assumes that sets are either REG or SUBRE
I have made this independent from the rest of the series, cleaned up
some comments and moved some code to its original position.
Re-submitted as
https://gcc.gnu.org/pipermail/gcc-patches/2023-November/637660.html.
Manolis
On Mon, Nov 13, 2023 at 2:40 PM Manolis Tsamis wrote:
>
> H
Retested, made independent of the rest of the series and submitted as
https://gcc.gnu.org/pipermail/gcc-patches/2023-November/637662.html.
Manolis
On Mon, Nov 13, 2023 at 2:43 PM Manolis Tsamis wrote:
>
> Yes, my finding back then was that this is leftover code from the
> initial imple
Bootstrapped and tested on x86 and aarch64. This only assumes that the
mode of what simplify_replace_rtx returns is the same with its first
argument.
Thanks,
Manolis
On Tue, Nov 21, 2023 at 8:04 PM Manolis Tsamis wrote:
>
> This code used to handle SUBREG for register replacement when ifc
(noce_convert_multiple_sets_1): Remove old code.
Signed-off-by: Manolis Tsamis
---
Changes in v2:
- Made standalone patch.
gcc/ifcvt.cc | 38 --
1 file changed, 38 deletions(-)
diff --git a/gcc/ifcvt.cc b/gcc/ifcvt.cc
index a0af553b9ff..be42609a848 100644
--- a/gcc
(struct noce_multiple_sets_info): Introduce new struct
noce_multiple_sets_info to store info for noce_convert_multiple_sets.
Signed-off-by: Manolis Tsamis
---
Changes in v2:
- Made standalone patch.
- Better comments, some more checks.
gcc/ifcvt.cc | 252
On Thu, Oct 19, 2023 at 10:46 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > Currently the operations allowed for if conversion of a basic block with
> > multiple sets are few, namely REG, SUBREG and CONST_INT (as controlled by
> > bb_ok_for_noce_convert_mult
at 12:03 AM Jeff Law wrote:
>
>
>
> On 8/30/23 04:14, Manolis Tsamis wrote:
> > This code used to handle register replacement issues with SUBREG before
> > simplify_replace_rtx was introduced. This should not be needed anymore as
> > new_val has the correct m
Hi Jeff,
Indeed, that sounds like a good idea. I will make this separate and
send it after the required testing.
I'll see what can be done about a testcase.
Best,
Manolis
On Sat, Nov 11, 2023 at 1:20 AM Jeff Law wrote:
>
>
>
> On 8/30/23 04:13, Manolis Tsamis wrote:
Hi all,
Pinging once more (
https://gcc.gnu.org/pipermail/gcc-patches/2023-August/628820.html).
Manolis
On Mon, Sep 18, 2023 at 11:18 AM Manolis Tsamis wrote:
>
> Kind ping for V3 of these ifcvt changes
> https://gcc.gnu.org/pipermail/gcc-patches/2023-August/628820.html
>
> Th
This version has been bootstrapped and tested to cause no regressions
in both x86 and Aarch64.
Manolis
On Mon, Oct 16, 2023 at 9:01 PM Manolis Tsamis wrote:
>
> This is a new RTL pass that tries to optimize memory offset calculations
> by moving them from add immediate instructio
On Tue, Oct 10, 2023 at 5:15 PM Jeff Law wrote:
>
>
>
> On 10/10/23 05:59, Manolis Tsamis wrote:
>
> >> It's a code quality issue as long as we don't transform the code into
> >> movl $0, -18874240, at which point it would become a correctness issue.
>
.
* fold-mem-offsets.cc: New file.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/fold-mem-offsets-1.c: New test.
* gcc.target/riscv/fold-mem-offsets-2.c: New test.
* gcc.target/riscv/fold-mem-offsets-3.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v7
On Fri, Oct 6, 2023 at 8:57 PM Jeff Law wrote:
>
>
>
> On 10/6/23 08:17, Manolis Tsamis wrote:
> SNIP
> >> So I was ready to ACK, but realized there weren't any testresults for a
> >> primary platform mentioned. So I ran this on x86.
> >>
&g
On Thu, Oct 5, 2023 at 1:05 AM Jeff Law wrote:
>
>
>
> On 10/3/23 05:45, Manolis Tsamis wrote:
> > This is a new RTL pass that tries to optimize memory offset calculations
> > by moving them from add immediate instructions to the memory loads/stores.
> > For
On Thu, Oct 5, 2023 at 5:54 PM Jeff Law wrote:
>
>
>
> On 10/3/23 05:45, Manolis Tsamis wrote:
> > This is a new RTL pass that tries to optimize memory offset calculations
>
> > +
> > +/* If INSN is a root memory instruction then compute a potentially new
>
On Fri, Sep 29, 2023 at 10:22 PM Jeff Law wrote:
>
>
>
> On 9/12/23 04:13, Manolis Tsamis wrote:
>
> >>> +
> >>> +/* Get the single reaching definition of an instruction inside a BB.
> >>> + The definition is desired for REG used in INSN.
>
.
* fold-mem-offsets.cc: New file.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/fold-mem-offsets-1.c: New test.
* gcc.target/riscv/fold-mem-offsets-2.c: New test.
* gcc.target/riscv/fold-mem-offsets-3.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v6
Kind ping for V3 of these ifcvt changes
https://gcc.gnu.org/pipermail/gcc-patches/2023-August/628820.html
Thanks,
Manolis
On Wed, Aug 30, 2023 at 1:14 PM Manolis Tsamis wrote:
>
>
> noce_convert_multiple_sets has been introduced and extended over time to
> handle
> if conver
On Tue, Sep 12, 2023 at 3:47 AM Jeff Law wrote:
>
>
>
> On 9/9/23 02:46, Manolis Tsamis wrote:
> > This is a new RTL pass that tries to optimize memory offset calculations
> > by moving them from add immediate instructions to the memory loads/stores.
> > For
541: 97
548: 101
557: 9
AArch64:
500: 71
502: 318
505: 0
520: 23
523: 205
525: 73
531: 7
541: 56
548: 0
557: 2
x86-64:
500: 8
502: 16
505: 0
520: 4
523: 5
525: 2
531: 0
541: 0
548: 0
557: 0
Thanks,
Manolis
On Sat, Sep 9, 2023 at 11:47 AM Manolis T
.
* fold-mem-offsets.cc: New file.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/fold-mem-offsets-1.c: New test.
* gcc.target/riscv/fold-mem-offsets-2.c: New test.
* gcc.target/riscv/fold-mem-offsets-3.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v5
nvenience.
> Best regards,
> Roger
> --
>
> > -Original Message-
> > From: Manolis Tsamis
> > Sent: 01 September 2023 11:45
> > To: Uros Bizjak
> > Cc: Roger Sayle ; gcc-patches@gcc.gnu.org
> > Subject: Re: [x86_64 PATCH] Improve __int128 arg
Hi Roger,
I've (accidentally) found a codegen regression that I bisected down to
this patch.
For these two functions:
typedef struct {
float minx, miny;
float maxx, maxy;
} AABB;
int TestOverlap(AABB a, AABB b) {
return a.minx <= b.maxx
&& a.miny <= b.maxy
&& a.maxx >= b.minx
On Tue, Jul 18, 2023 at 9:38 PM Richard Sandiford
wrote:
>
> Manolis Tsamis writes:
> > On Tue, Jul 18, 2023 at 1:12 AM Richard Sandiford
> > wrote:
> >>
> >> Manolis Tsamis writes:
> >> > noce_convert_multiple_sets has been introduced an
(struct noce_multiple_sets_info): Introduce new struct
noce_multiple_sets_info to store info for noce_convert_multiple_sets.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/ifcvt_multiple_sets_rewire.c: New test.
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc
): Remove old code.
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc | 38 --
1 file changed, 38 deletions(-)
diff --git a/gcc/ifcvt.cc b/gcc/ifcvt.cc
index ecc0cbabef9..3b4b873612c 100644
--- a/gcc/ifcvt.cc
+++ b/gcc/ifcvt.cc
@@ -3449,44
et/aarch64/ifcvt_multiple_sets_arithm.c: New test.
Signed-off-by: Manolis Tsamis
---
Changes in v3:
- Add SCALAR_INT_MODE_P check in bb_ok_for_noce_convert_multiple_sets.
- Allow rewiring of multiple regs.
- Refactor code with noce_multiple_sets_info.
- Remove old code for subregs.
gc
Signed-off-by: Manolis Tsamis
---
(no changes since v1)
gcc/ifcvt.cc | 49 +++--
1 file changed, 19 insertions(+), 30 deletions(-)
diff --git a/gcc/ifcvt.cc b/gcc/ifcvt.cc
index a0af553b9ff..3273aeca125 100644
--- a/gcc/ifcvt.cc
+++ b/gcc/ifcvt.cc
@
SCALAR_INT_MODE_P check in bb_ok_for_noce_convert_multiple_sets.
- Allow rewiring of multiple regs.
- Refactor code with noce_multiple_sets_info.
- Remove old code for subregs.
Manolis Tsamis (4):
ifcvt: handle sequences that clobber flags in
noce_convert_multiple_sets
1 - 100 of 161 matches
Mail list logo
