public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r14-3614] RISC-V: Enable VECT_COMPARE_COSTS by default
@ 2023-09-01 11:02 Pan Li
  0 siblings, 0 replies; only message in thread
From: Pan Li @ 2023-09-01 11:02 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:5f2098cce6c75117927fef317c714dd2088b0189

commit r14-3614-g5f2098cce6c75117927fef317c714dd2088b0189
Author: Juzhe-Zhong <juzhe.zhong@rivai.ai>
Date:   Fri Sep 1 07:10:00 2023 +0800

    RISC-V: Enable VECT_COMPARE_COSTS by default
    
    since we have added COST framework, we by default enable VECT_COMPARE_COSTS.
    
    Also, add 16/32/64 to provide more choices for COST comparison.
    
    This patch doesn't change any behavior from the current testsuite since we are using
    default COST model.
    
    gcc/ChangeLog:
    
            * config/riscv/riscv-v.cc (autovectorize_vector_modes): Enable
            VECT_COMPARE_COSTS by default.

Diff:
---
 gcc/config/riscv/riscv-v.cc | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/gcc/config/riscv/riscv-v.cc b/gcc/config/riscv/riscv-v.cc
index 6228ff3d92e9..c8ad96f44d5e 100644
--- a/gcc/config/riscv/riscv-v.cc
+++ b/gcc/config/riscv/riscv-v.cc
@@ -2173,7 +2173,7 @@ autovectorize_vector_modes (vector_modes *modes, bool)
 	   full vectors for wider elements.
 	 - full_size / 8:
 	   Try using 64-bit containers for all element types.  */
-      static const int rvv_factors[] = {1, 2, 4, 8};
+      static const int rvv_factors[] = {1, 2, 4, 8, 16, 32, 64};
       for (unsigned int i = 0; i < sizeof (rvv_factors) / sizeof (int); i++)
 	{
 	  poly_uint64 units;
@@ -2183,12 +2183,8 @@ autovectorize_vector_modes (vector_modes *modes, bool)
 	    modes->safe_push (mode);
 	}
     }
-  unsigned int flag = 0;
   if (TARGET_VECTOR_VLS)
     {
-      /* Enable VECT_COMPARE_COSTS between VLA modes VLS modes for scalable
-	 auto-vectorization.  */
-      flag |= VECT_COMPARE_COSTS;
       /* Push all VLSmodes according to TARGET_MIN_VLEN.  */
       unsigned int i = 0;
       unsigned int base_size = TARGET_MIN_VLEN * riscv_autovec_lmul / 8;
@@ -2201,7 +2197,8 @@ autovectorize_vector_modes (vector_modes *modes, bool)
 	  size = base_size / (1U << i);
 	}
     }
-  return flag;
+  /* Enable LOOP_VINFO comparison in COST model.  */
+  return VECT_COMPARE_COSTS;
 }
 
 /* If the given VECTOR_MODE is an RVV mode,  first get the largest number

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2023-09-01 11:02 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-09-01 11:02 [gcc r14-3614] RISC-V: Enable VECT_COMPARE_COSTS by default Pan Li

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).