Hi Haochen,
Some minor comments are inlined.
on 2022/3/10 2:31 PM, HAO CHEN GUI via Gcc-patches wrote:
> Hi,
> This patch adds V1TI mode into mode iterator used in vector comparison
> expands.With the patch, both built-ins and direct comparison could generate
> P10 new V1TI comparison instructions.
>
> Bootstrapped and tested on ppc64 Linux BE and LE with no regressions. Is
> this okay for trunk? Any recommendations? Thanks a lot.
>
> ChangeLog
> 2022-03-09 Haochen Gui <guih...@linux.ibm.com>
>
> gcc/
> PR target/103316
> * config/rs6000/rs6000-builtin.cc (rs6000_gimple_fold_builtin): Enable
> gimple folding for RS6000_BIF_VCMPEQUT, RS6000_BIF_VCMPNET,
> RS6000_BIF_CMPGE_1TI, RS6000_BIF_CMPGE_U1TI, RS6000_BIF_VCMPGTUT,
> RS6000_BIF_VCMPGTST, RS6000_BIF_CMPLE_1TI, RS6000_BIF_CMPLE_U1TI.
> * config/rs6000/vector.md (VEC_IC): Define. Add support for new Power10
> V1TI instructions.
> (vec_cmp<mode><mode>): Set mode iterator to VEC_IC.
> (vec_cmpu<mode><mode>): Likewise.
>
> gcc/testsuite/
> PR target/103316
> * gcc.target/powerpc/pr103316.c: New.
>
> patch.diff
> diff --git a/gcc/config/rs6000/rs6000-builtin.cc
> b/gcc/config/rs6000/rs6000-builtin.cc
> index 5d34c1bcfc9..143effa89bf 100644
> --- a/gcc/config/rs6000/rs6000-builtin.cc
> +++ b/gcc/config/rs6000/rs6000-builtin.cc
> @@ -1994,6 +1994,7 @@ rs6000_gimple_fold_builtin (gimple_stmt_iterator *gsi)
> case RS6000_BIF_VCMPEQUH:
> case RS6000_BIF_VCMPEQUW:
> case RS6000_BIF_VCMPEQUD:
> + case RS6000_BIF_VCMPEQUT:
> /* We deliberately omit RS6000_BIF_VCMPEQUT for now, because gimple
> folding produces worse code for 128-bit compares. */
The comment above is saying why there is no RS6000_BIF_VCMPEQUT before, IIUC
the point
doesn't hold any more with your patch. So could you remove it to avoid possible
confusions? Also some similar places ...
> fold_compare_helper (gsi, EQ_EXPR, stmt);
> @@ -2002,6 +2003,7 @@ rs6000_gimple_fold_builtin (gimple_stmt_iterator *gsi)
> case RS6000_BIF_VCMPNEB:
> case RS6000_BIF_VCMPNEH:
> case RS6000_BIF_VCMPNEW:
> + case RS6000_BIF_VCMPNET:
> /* We deliberately omit RS6000_BIF_VCMPNET for now, because gimple
> folding produces worse code for 128-bit compares. */
here ...
> fold_compare_helper (gsi, NE_EXPR, stmt);
> @@ -2015,6 +2017,8 @@ rs6000_gimple_fold_builtin (gimple_stmt_iterator *gsi)
> case RS6000_BIF_CMPGE_U4SI:
> case RS6000_BIF_CMPGE_2DI:
> case RS6000_BIF_CMPGE_U2DI:
> + case RS6000_BIF_CMPGE_1TI:
> + case RS6000_BIF_CMPGE_U1TI:
> /* We deliberately omit RS6000_BIF_CMPGE_1TI and RS6000_BIF_CMPGE_U1TI
> for now, because gimple folding produces worse code for 128-bit
> compares. */
here...
> @@ -2029,6 +2033,8 @@ rs6000_gimple_fold_builtin (gimple_stmt_iterator *gsi)
> case RS6000_BIF_VCMPGTUW:
> case RS6000_BIF_VCMPGTUD:
> case RS6000_BIF_VCMPGTSD:
> + case RS6000_BIF_VCMPGTUT:
> + case RS6000_BIF_VCMPGTST:
> /* We deliberately omit RS6000_BIF_VCMPGTUT and RS6000_BIF_VCMPGTST
> for now, because gimple folding produces worse code for 128-bit
> compares. */
here...
> @@ -2043,6 +2049,8 @@ rs6000_gimple_fold_builtin (gimple_stmt_iterator *gsi)
> case RS6000_BIF_CMPLE_U4SI:
> case RS6000_BIF_CMPLE_2DI:
> case RS6000_BIF_CMPLE_U2DI:
> + case RS6000_BIF_CMPLE_1TI:
> + case RS6000_BIF_CMPLE_U1TI:
> /* We deliberately omit RS6000_BIF_CMPLE_1TI and RS6000_BIF_CMPLE_U1TI
> for now, because gimple folding produces worse code for 128-bit
> compares. */
here...
> diff --git a/gcc/config/rs6000/vector.md b/gcc/config/rs6000/vector.md
> index b87a742cca8..1afb8a6d786 100644
> --- a/gcc/config/rs6000/vector.md
> +++ b/gcc/config/rs6000/vector.md
> @@ -26,6 +26,9 @@
> ;; Vector int modes
> (define_mode_iterator VEC_I [V16QI V8HI V4SI V2DI])
>
> +;; Vector int modes for comparison
> +(define_mode_iterator VEC_IC [V16QI V8HI V4SI V2DI V1TI])
> +
Maybe we can make this define be like:
(define_mode_iterator VEC_IC [V16QI V8HI V4SI V2DI (V1TI "TARGET_POWER10")])
...
> ;; 128-bit int modes
> (define_mode_iterator VEC_TI [V1TI TI])
>
> @@ -533,11 +536,12 @@ (define_expand "vcond_mask_<mode><VEC_int>"
>
> ;; For signed integer vectors comparison.
> (define_expand "vec_cmp<mode><mode>"
> - [(set (match_operand:VEC_I 0 "vint_operand")
> + [(set (match_operand:VEC_IC 0 "vint_operand")
> (match_operator 1 "signed_or_equality_comparison_operator"
> - [(match_operand:VEC_I 2 "vint_operand")
> - (match_operand:VEC_I 3 "vint_operand")]))]
> - "VECTOR_UNIT_ALTIVEC_OR_VSX_P (<MODE>mode)"
> + [(match_operand:VEC_IC 2 "vint_operand")
> + (match_operand:VEC_IC 3 "vint_operand")]))]
> + "(VECTOR_UNIT_ALTIVEC_OR_VSX_P (<MODE>mode) && <MODE>mode!= V1TImode)
> + || (<MODE>mode == V1TImode && TARGET_POWER10)"
and this condition can be kept as simple with VECTOR_UNIT_ALTIVEC_OR_VSX_P
(<MODE>mode)?
> {
> enum rtx_code code = GET_CODE (operands[1]);
> rtx tmp = gen_reg_rtx (<MODE>mode);
> @@ -573,11 +577,12 @@ (define_expand "vec_cmp<mode><mode>"
>
> ;; For unsigned integer vectors comparison.
> (define_expand "vec_cmpu<mode><mode>"
> - [(set (match_operand:VEC_I 0 "vint_operand")
> + [(set (match_operand:VEC_IC 0 "vint_operand")
> (match_operator 1 "unsigned_or_equality_comparison_operator"
> - [(match_operand:VEC_I 2 "vint_operand")
> - (match_operand:VEC_I 3 "vint_operand")]))]
> - "VECTOR_UNIT_ALTIVEC_OR_VSX_P (<MODE>mode)"
> + [(match_operand:VEC_IC 2 "vint_operand")
> + (match_operand:VEC_IC 3 "vint_operand")]))]
> + "(VECTOR_UNIT_ALTIVEC_OR_VSX_P (<MODE>mode) && <MODE>mode != V1TImode)
> + || (<MODE>mode == V1TImode && TARGET_POWER10)"
same as above.
> {
> enum rtx_code code = GET_CODE (operands[1]);
> rtx tmp = gen_reg_rtx (<MODE>mode);
> diff --git a/gcc/testsuite/gcc.target/powerpc/pr103316.c
> b/gcc/testsuite/gcc.target/powerpc/pr103316.c
It seems better to have another test case with explicit builtin function calls
like vec_cmp{gt, eq, ge ... }
to cover the changes in rs6000_gimple_fold_builtin?
> new file mode 100644
> index 00000000000..da2121a30de
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/powerpc/pr103316.c
> @@ -0,0 +1,79 @@
> +/* { dg-do compile } */
> +/* { dg-require-effective-target power10_ok } */
> +/* { dg-options "-mdejagnu-cpu=power10 -O2" } */
> +
> +#include <altivec.h>
Nit: I guess this include is useless?
BR,
Kewen
