diff mbox

[PR65915] Fix float conversion split.

Message ID 20150428162049.GA62528@msticlxl7.ims.intel.com
State New
Headers show

Commit Message

Ilya Tocar April 28, 2015, 4:20 p.m. UTC 
> > I've renamed EXT_SSE_REG_P into EXT_REX_SSE_REG_P for consistency.
> > Ok for stage1?
> Patch is OK for stage1.
> 
> --
> Thanks, K
> 
> 
> > On 19 Mar 12:24, Ilya Tocar wrote:
> > > Hi,
> > > 
> > > There were some discussion about "x" constraints being too conservative
> > > for some patterns in i386.md.
> > > Patch below fixes it. This is probably stage1 material.
> > > 
> > > ChangeLog:
> > > 
> > > gcc/
> > > 
> > 2015-03-23  Ilya Tocar  <ilya.tocar@intel.com>
> > 
> > 	* config/i386/i386.h (EXT_REX_SSE_REG_P): New.
> > 	* config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>_mixed): Use "v"
> > 	constraint.
> > 	(*cmpi<FPCMP:unord><MODEF:mode>_sse): Ditto.
> > 	(*movxi_internal_avx512f): Ditto.
> > 	(define_split): Check for xmm16+, when splitting scalar float_extend.
> > 	(*extendsfdf2_mixed): Use "v" constraint.
> > 	(*extendsfdf2_sse): Ditto.
> > 	(define_split): Check for xmm16+, when splitting scalar float_truncate.
> > 	(*truncdfsf_fast_sse): Use "v" constraint.
> > 	(fix_trunc<MODEF:mode><SWI48:mode>_sse): Ditto.
> > 	(*float<SWI48:mode><MODEF:mode>2_sse): Ditto.
> > 	(define_peephole2): Check for xmm16+, when converting scalar
> > 	float_truncate.
> > 	(define_peephole2): Check for xmm16+, when converting scalar
> > 	float_extend.
> > 	(*fop_<mode>_comm_mixed): Use "v" constraint.
> > 	(*fop_<mode>_comm_sse): Ditto.
> > 	(*fop_<mode>_1_mixed): Ditto.
> > 	(*sqrt<mode>2_sse): Ditto.
> > 	(*ieee_s<ieee_maxmin><mode>3): Ditto.
> > 
> >

Hi,

Looks like I missed some splits, which caused PR65915.
Patch below fixes it.
Ok for trunk?

2015-04-28  Ilya Tocar  <ilya.tocar@intel.com>

	* config/i386/i386.md (define_split): Check for xmm16+,
	when splitting scalar float conversion.


---
 gcc/config/i386/i386.md | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

Comments

H.J. Lu April 28, 2015, 4:22 p.m. UTC | #1 
On Tue, Apr 28, 2015 at 9:20 AM, Ilya Tocar <tocarip.intel@gmail.com> wrote:
>> > I've renamed EXT_SSE_REG_P into EXT_REX_SSE_REG_P for consistency.
>> > Ok for stage1?
>> Patch is OK for stage1.
>>
>> --
>> Thanks, K
>>
>>
>> > On 19 Mar 12:24, Ilya Tocar wrote:
>> > > Hi,
>> > >
>> > > There were some discussion about "x" constraints being too conservative
>> > > for some patterns in i386.md.
>> > > Patch below fixes it. This is probably stage1 material.
>> > >
>> > > ChangeLog:
>> > >
>> > > gcc/
>> > >
>> > 2015-03-23  Ilya Tocar  <ilya.tocar@intel.com>
>> >
>> >     * config/i386/i386.h (EXT_REX_SSE_REG_P): New.
>> >     * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>_mixed): Use "v"
>> >     constraint.
>> >     (*cmpi<FPCMP:unord><MODEF:mode>_sse): Ditto.
>> >     (*movxi_internal_avx512f): Ditto.
>> >     (define_split): Check for xmm16+, when splitting scalar float_extend.
>> >     (*extendsfdf2_mixed): Use "v" constraint.
>> >     (*extendsfdf2_sse): Ditto.
>> >     (define_split): Check for xmm16+, when splitting scalar float_truncate.
>> >     (*truncdfsf_fast_sse): Use "v" constraint.
>> >     (fix_trunc<MODEF:mode><SWI48:mode>_sse): Ditto.
>> >     (*float<SWI48:mode><MODEF:mode>2_sse): Ditto.
>> >     (define_peephole2): Check for xmm16+, when converting scalar
>> >     float_truncate.
>> >     (define_peephole2): Check for xmm16+, when converting scalar
>> >     float_extend.
>> >     (*fop_<mode>_comm_mixed): Use "v" constraint.
>> >     (*fop_<mode>_comm_sse): Ditto.
>> >     (*fop_<mode>_1_mixed): Ditto.
>> >     (*sqrt<mode>2_sse): Ditto.
>> >     (*ieee_s<ieee_maxmin><mode>3): Ditto.
>> >
>> >
>
> Hi,
>
> Looks like I missed some splits, which caused PR65915.
> Patch below fixes it.
> Ok for trunk?
>
> 2015-04-28  Ilya Tocar  <ilya.tocar@intel.com>
>
>         * config/i386/i386.md (define_split): Check for xmm16+,
>         when splitting scalar float conversion.
>

Need a testcase and PR# in ChangeLog.
diff mbox

Patch

diff --git a/gcc/config/i386/i386.md b/gcc/config/i386/i386.md
index 937871a..af1cd9b 100644
--- a/gcc/config/i386/i386.md
+++ b/gcc/config/i386/i386.md
@@ -4897,7 +4897,9 @@ 
   "TARGET_SSE2 && TARGET_SSE_MATH
    && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)
    && reload_completed && SSE_REG_P (operands[0])
-   && (MEM_P (operands[1]) || TARGET_INTER_UNIT_MOVES_TO_VEC)"
+   && (MEM_P (operands[1]) || TARGET_INTER_UNIT_MOVES_TO_VEC)
+   && (!EXT_REX_SSE_REG_P (operands[0])
+       || TARGET_AVX512VL)"
   [(const_int 0)]
 {
   operands[3] = simplify_gen_subreg (<ssevecmode>mode, operands[0],
@@ -4921,7 +4923,9 @@ 
   "TARGET_SSE2 && TARGET_SSE_MATH
    && TARGET_SSE_PARTIAL_REG_DEPENDENCY
    && optimize_function_for_speed_p (cfun)
-   && reload_completed && SSE_REG_P (operands[0])"
+   && reload_completed && SSE_REG_P (operands[0])
+   && (!EXT_REX_SSE_REG_P (operands[0])
+       || TARGET_AVX512VL)"
   [(const_int 0)]
 {
   const machine_mode vmode = <MODEF:ssevecmode>mode;