On Fri, 2024-06-28 at 17:53 -0700, Vineet Gupta wrote:
> + UNSPEC_ISFINITE
> + UNSPEC_ISNORMAL
You don't really need them. The RTL pattern of define_expand has no use
when you expand it via C code and DONE.
i.e. you can just code
(define_expand "isfinite2"
[(match_operand:SI 0 "register_o
The following factors out the code that preserves SSA info of the LHS
of a SSA copy LHS = RHS when LHS is about to be eliminated to RHS.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
PR tree-optimization/115701
* tree-ssanames.h (maybe_duplicate_ssa_info_at_copy): D
The following restricts copying of points-to info from defs that
might be in regions invoking UB and are never executed.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
PR tree-optimization/115701
* tree-ssanames.cc (maybe_duplicate_ssa_info_at_copy):
Only cop
Besides VN and copy-prop also CCP and VRP as well as forwprop
propagate out copies and thus it's worthwhile to try to preserve
range and points-to info there when possible.
Note that this also fixes the testcase from PR115701 but that's
because we do not actually intersect info but only copy info
Hello All:
This patch determines unroll factor based on loop register pressure.
Unroll factor is quotient of max of available registers in loop
by number of liveness.
If available registers increases unroll factor increases.
Wherein unroll factor decreases if number of liveness increases.
Loop
This patch fixes the 4 FAILs of gcc.target/i386/pr192464-vrndscaleph.c
with --target_board='unix{-m32}' on RedHat 7.x. The issue is that this
AVX512 test includes the system math.h, and on older systems this provides
inline versions of floor, ceil and rint (for the 387). The work around
is to de
On 6/29/24 3:07 PM, Vineet Gupta wrote:
On 6/29/24 06:44, Jeff Law wrote:
+;; fclass instruction output bitmap
+;; 0 negative infinity
+;; 1 negative normal number.
+;; 2 negative subnormal number.
+;; 3 -0
+;; 4 +0
+;; 5 positive subnormal number.
+;; 6 positive normal number.
Fixes mismatched predicate and constraint in xmpyu patterns.
The xmpyu instruction doesn't support multiplying a 32-bit const_int
with a floating-point register. So, I just removed these patterns.
Tested on hppa-unknown-linux-gnu. Committed to trunk.
Dave
---
hppa: Fix ICE caused by mismatched
On Sat, Jun 29, 2024 at 6:21 PM Roger Sayle wrote:
>
>
> A common idiom for implementing an integer division that rounds upwards is
> to write (x + y - 1) / y. Conveniently on x86, the two additions to form
> the numerator can be performed by a single lea instruction, and indeed gcc
> currently g
Changes since V1:
Check flag_ext_dce before running the new pass. I'd forgotten that
I had removed that part of the gate to facilitate more testing.
Turn flag_ext_dce on at -O2 and above.
Adjust one of the riscv tests to explicitly avoid vectors
Adjust a few aarch64 tests
In tbz_2
Hi Uros,
> On Sat, Jun 29, 2024 at 6:21 PM Roger Sayle
> wrote:
> > A common idiom for implementing an integer division that rounds
> > upwards is to write (x + y - 1) / y. Conveniently on x86, the two
> > additions to form the numerator can be performed by a single lea
> > instruction, and ind
On Sun, Jun 30, 2024 at 9:09 PM Roger Sayle wrote:
>
>
> Hi Uros,
> > On Sat, Jun 29, 2024 at 6:21 PM Roger Sayle
> > wrote:
> > > A common idiom for implementing an integer division that rounds
> > > upwards is to write (x + y - 1) / y. Conveniently on x86, the two
> > > additions to form the n
Hello-
I noticed this while trying to test another patch on Windows (using the
MSYS2 environment). Tested that it fixes the issue for x86_64-w64-mingw32
and doesn't affect anything for x86_64-pc-linux-gnu. It looks like the same
fix for C was applied back in r11-702. OK? Thanks...
-Lewis
-- >8 -
As promised here's the final ternlog clean-up, that deletes the now
obsolete legacy patterns and mode iterators from sse.md. It also updates
the surviving ternlog patterns to consistently use decimal immediate
operands (instead of hexadecimal), and updates one last test case to
match this change.
On 6/30/24 00:41, Xi Ruoyao wrote:
> On Fri, 2024-06-28 at 17:53 -0700, Vineet Gupta wrote:
>> + UNSPEC_ISFINITE
>> + UNSPEC_ISNORMAL
> You don't really need them. The RTL pattern of define_expand has no use
> when you expand it via C code and DONE.
>
> i.e. you can just code
>
> (define_expa
On 6/30/24 06:59, Jeff Law wrote:
>> Any ideas on how I can keep this and then adjust rest of patterns.
> Yea. Drop the "SImode" references from the RTL template of the
> expander. Then you'll need to verify the modes in the C fragment that
> generates code. You'd want to test the mode of
Changes since v1:
- Removed UNSPEC_{INFINITE,ISNORMAL}
- Don't hardcode SI in patterns, try to keep X to avoid potential
sign extension pitfalls. Implementation wise requires skipping
:MODE specifier in match_operand which is flagged as missing mode
warning.
---
Currently isfinite
From: Pan Li
This patch would like to add test cases for the unsigned scalar
.SAT_ADD IMM form 3. Aka:
Form 3:
#define DEF_SAT_U_ADD_IMM_FMT_3(T) \
T __attribute__((noinline)) \
sat_u_add_imm_##T##_fmt_3 (T x) \
From: Pan Li
This patch would like to add test cases for the unsigned scalar
.SAT_ADD IMM form 2. Aka:
Form 2:
#define DEF_SAT_U_ADD_IMM_FMT_2(T) \
T __attribute__((noinline)) \
sat_u_add_imm_##T##_fmt_1 (T x) \
{ \
retu
On Sun, Jun 30, 2024 at 7:29 PM Roger Sayle wrote:
>
>
> This patch fixes the 4 FAILs of gcc.target/i386/pr192464-vrndscaleph.c
> with --target_board='unix{-m32}' on RedHat 7.x. The issue is that this
> AVX512 test includes the system math.h, and on older systems this provides
> inline versions o
From: Pan Li
This patch would like to add test cases for the unsigned scalar
.SAT_ADD IMM form 1. Aka:
Form 1:
#define DEF_SAT_U_ADD_IMM_FMT_1(T) \
T __attribute__((noinline)) \
sat_u_add_imm_##T##_fmt_1 (T x) \
{\
From: Pan Li
This patch would like to add test cases for the unsigned scalar
.SAT_ADD IMM form 4. Aka:
Form 4:
#define DEF_SAT_U_ADD_IMM_FMT_4(T)\
T __attribute__((noinline)) \
sat_u_add_imm_##T##_fmt_4 (T x)
On Mon, Jul 1, 2024 at 6:14 AM Roger Sayle wrote:
>
>
> As promised here's the final ternlog clean-up, that deletes the now
> obsolete legacy patterns and mode iterators from sse.md. It also updates
> the surviving ternlog patterns to consistently use decimal immediate
> operands (instead of hexa
The following adds a missed check when forwprop attempts to rewrite
a complex store.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed.
PR tree-optimization/115694
* tree-ssa-forwprop.cc (pass_forwprop::execute): Check the
store is complex before rewriting it.
> >
> > gcc/testsuite/ChangeLog
> > * gcc.target/i386/pr100711-6.c: Update to check for decimal
> > immediate operand in ternlog, not hexadecimal.
> I got an ICE when bootstrapped with --enable-checking=yes,rtl,extra
>
The ICE can be walked around with 2 separate define_predicates,
25 matches
Mail list logo
