On Tue, Jun 7, 2022 at 9:41 AM liuhongt <hongtao....@intel.com> wrote:
>
> So alternative v won't be igored in record_reg_classess.
>
> Similar for *r alternatives in some vector patterns.
>
> It helps testcase in the PR, also RA now makes better decisions for
> gcc.target/i386/extract-insert-combining.c
>
> movd %esi, %xmm0
> movd %edi, %xmm1
> - movl %esi, -12(%rsp)
> paddd %xmm0, %xmm1
> pinsrd $0, %esi, %xmm0
> paddd %xmm1, %xmm0
>
> The patch has no big impact on SPEC2017 for both O2 and Ofast
> march=native run.
>
> And I noticed there's some changes in SPEC2017
>
> Before:
> mov mem, %eax
> vmovd %eax, %xmm0
> ..
> mov %eax, 64(%rsp)
>
> After:
> vmovd mem, %xmm0
> ..
> vmovd %xmm0, 64(%rsp)
>
> Which should be exactly what we want?
>
> Bootstrapped and regtested on x86_64-pc-linux-gnu{-m32,}
> Ok for trunk?
>
> gcc/ChangeLog:
>
> * config/i386/i386.md (*movsi_internal): Change alternative
> from *v to ?v.
> (*movdi_internal): Ditto.
> * config/i386/sse.md (vec_set<mode>_0): Change alternative *r
> to ?r.
> (*vec_extractv4sf_mem): Ditto.
> (*vec_extracthf): Ditto.
>
> gcc/testsuite/ChangeLog:
>
> * gcc.target/i386/pr105513-1.c: New test.
> * gcc.target/i386/extract-insert-combining.c: Add new
> scan-assembler-not for spill.
Let's have some experiment with this approach. The above is also
better for TUNE_INTER_UNIT_MOVES_{TO,FROM}_VEC, since moves between
%eax and %xmm will again go through memory (I'm not sure how much we
care for these targets anyway).
OK.
Thanks,
Uros.
> ---
> gcc/config/i386/i386.md | 8 ++++----
> gcc/config/i386/sse.md | 8 ++++----
> .../gcc.target/i386/extract-insert-combining.c | 1 +
> gcc/testsuite/gcc.target/i386/pr105513-1.c | 16 ++++++++++++++++
> 4 files changed, 25 insertions(+), 8 deletions(-)
> create mode 100644 gcc/testsuite/gcc.target/i386/pr105513-1.c
>
> diff --git a/gcc/config/i386/i386.md b/gcc/config/i386/i386.md
> index 48a98e1b68b..5b538413942 100644
> --- a/gcc/config/i386/i386.md
> +++ b/gcc/config/i386/i386.md
> @@ -2251,9 +2251,9 @@ (define_split
>
> (define_insn "*movdi_internal"
> [(set (match_operand:DI 0 "nonimmediate_operand"
> - "=r ,o ,r,r ,r,m ,*y,*y,?*y,?m,?r,?*y,*v,*v,*v,m ,m,?r
> ,?*Yd,?r,?*v,?*y,?*x,*k,*k ,*r,*m,*k")
> + "=r ,o ,r,r ,r,m ,*y,*y,?*y,?m,?r,?*y,?v,?v,?v,m ,m,?r
> ,?*Yd,?r,?v,?*y,?*x,*k,*k ,*r,*m,*k")
> (match_operand:DI 1 "general_operand"
> - "riFo,riF,Z,rem,i,re,C ,*y,Bk ,*y,*y,r ,C ,*v,Bk,*v,v,*Yd,r ,*v,r
> ,*x ,*y ,*r,*kBk,*k,*k,CBC"))]
> + "riFo,riF,Z,rem,i,re,C ,*y,Bk ,*y,*y,r ,C ,?v,Bk,?v,v,*Yd,r ,?v,r
> ,*x ,*y ,*r,*kBk,*k,*k,CBC"))]
> "!(MEM_P (operands[0]) && MEM_P (operands[1]))
> && ix86_hardreg_mov_ok (operands[0], operands[1])"
> {
> @@ -2472,9 +2472,9 @@ (define_peephole2
>
> (define_insn "*movsi_internal"
> [(set (match_operand:SI 0 "nonimmediate_operand"
> - "=r,m ,*y,*y,?*y,?m,?r,?*y,*v,*v,*v,m ,?r,?*v,*k,*k ,*rm,*k")
> + "=r,m ,*y,*y,?*y,?m,?r,?*y,?v,?v,?v,m ,?r,?v,*k,*k ,*rm,*k")
> (match_operand:SI 1 "general_operand"
> - "g ,re,C ,*y,Bk ,*y,*y,r ,C ,*v,Bk,*v,*v,r ,*r,*kBk,*k ,CBC"))]
> + "g ,re,C ,*y,Bk ,*y,*y,r ,C ,?v,Bk,?v,?v,r ,*r,*kBk,*k ,CBC"))]
> "!(MEM_P (operands[0]) && MEM_P (operands[1]))
> && ix86_hardreg_mov_ok (operands[0], operands[1])"
> {
> diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
> index 62688f8e29d..d41ce2e1a9b 100644
> --- a/gcc/config/i386/sse.md
> +++ b/gcc/config/i386/sse.md
> @@ -10590,11 +10590,11 @@ (define_insn "*vec_concatv4sf_0"
> ;; see comment above inline_secondary_memory_needed function in i386.cc
> (define_insn "vec_set<mode>_0"
> [(set (match_operand:VI4F_128 0 "nonimmediate_operand"
> - "=Yr,*x,v,v,v,x,x,v,Yr ,*x ,x ,m ,m ,m")
> + "=Yr,*x,v,v,v,x,x,v,Yr ,?x ,x ,m ,m ,m")
> (vec_merge:VI4F_128
> (vec_duplicate:VI4F_128
> (match_operand:<ssescalarmode> 2 "general_operand"
> - " Yr,*x,v,m,r ,m,x,v,*rm,*rm,*rm,!x,!*re,!*fF"))
> + " Yr,*x,v,m,r ,m,x,v,?rm,?rm,?rm,!x,?re,!*fF"))
> (match_operand:VI4F_128 1 "nonimm_or_0_operand"
> " C , C,C,C,C ,C,0,v,0 ,0 ,x ,0 ,0 ,0")
> (const_int 1)))]
> @@ -11056,7 +11056,7 @@ (define_insn_and_split "*sse4_1_extractps"
> (set_attr "mode" "V4SF,V4SF,V4SF,*,*")])
>
> (define_insn_and_split "*vec_extractv4sf_mem"
> - [(set (match_operand:SF 0 "register_operand" "=v,*r,f")
> + [(set (match_operand:SF 0 "register_operand" "=v,?r,f")
> (vec_select:SF
> (match_operand:V4SF 1 "memory_operand" "o,o,o")
> (parallel [(match_operand 2 "const_0_to_3_operand")])))]
> @@ -11933,7 +11933,7 @@ (define_insn_and_split "*vec_extract<mode>_0"
> "operands[1] = gen_lowpart (HFmode, operands[1]);")
>
> (define_insn "*vec_extracthf"
> - [(set (match_operand:HF 0 "register_sse4nonimm_operand" "=*r,m,x,v")
> + [(set (match_operand:HF 0 "register_sse4nonimm_operand" "=?r,m,x,v")
> (vec_select:HF
> (match_operand:V8HF 1 "register_operand" "v,v,0,v")
> (parallel
> diff --git a/gcc/testsuite/gcc.target/i386/extract-insert-combining.c
> b/gcc/testsuite/gcc.target/i386/extract-insert-combining.c
> index 32d951e6832..5a53d4cbf06 100644
> --- a/gcc/testsuite/gcc.target/i386/extract-insert-combining.c
> +++ b/gcc/testsuite/gcc.target/i386/extract-insert-combining.c
> @@ -4,6 +4,7 @@
> /* { dg-final { scan-assembler-times "(?:vpaddd|paddd)\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]" 2 } } */
> /* { dg-final { scan-assembler-times "(?:vpinsrd|pinsrd)\[
> \\t\]+\[^\{\n\]*%xmm\[0-9\]" 1 } } */
> /* { dg-final { scan-assembler-not "vmovss" } } */
> +/* { dg-final { scan-assembler-not {(?n)mov.*(%rsp)} { target { ! ia32 } } }
> } */
>
> #include <immintrin.h>
>
> diff --git a/gcc/testsuite/gcc.target/i386/pr105513-1.c
> b/gcc/testsuite/gcc.target/i386/pr105513-1.c
> new file mode 100644
> index 00000000000..530f5292252
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/pr105513-1.c
> @@ -0,0 +1,16 @@
> +/* { dg-do compile { target { ! ia32 } } } */
> +/* { dg-options "-O2 -msse2 -mtune=skylake -mfpmath=sse" } */
> +/* { dg-final { scan-assembler-not "\\(%rsp\\)" } } */
> +
> +static int as_int(float x)
> +{
> + return (union{float x; int i;}){x}.i;
> +}
> +
> +float f(double y, float x)
> +{
> + int i = as_int(x);
> + if (__builtin_expect(i > 99, 0)) return 0;
> + if (i*2u < 77) if (i==2) return 0;
> + return y*x;
> +}
> --
> 2.18.1
>
