Re: [PATCH 2/2] Tune memcpy and memset for Zen cores.

[Martin Liška]

Adding Honza as Uros recommended him for a review.

Martin

On 6/1/20 1:35 PM, Martin Liška wrote:

Based on the collected numbers in PR95435, I suggest the following
tuning changes:

gcc/ChangeLog:

     PR target/95435
     * config/i386/x86-tune-costs.h: Use libcall for large sizes for
     -m32. Start using libcall from 128+ bytes.
---
  gcc/config/i386/x86-tune-costs.h | 12 ++++++------
  1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/gcc/config/i386/x86-tune-costs.h b/gcc/config/i386/x86-tune-costs.h
index 1169178433f..3207404e514 100644
--- a/gcc/config/i386/x86-tune-costs.h
+++ b/gcc/config/i386/x86-tune-costs.h
@@ -1314,20 +1314,20 @@ static stringop_algs znver1_memcpy[2] = {
    /* 32-bit tuning.  */
    {libcall, {{6, loop, false},
           {14, unrolled_loop, false},
-         {-1, rep_prefix_4_byte, false}}},
+         {-1, libcall, false}}},
    /* 64-bit tuning.  */
    {libcall, {{16, loop, false},
-         {8192, rep_prefix_8_byte, false},
+         {128, rep_prefix_8_byte, false},
           {-1, libcall, false}}}};
  static stringop_algs znver1_memset[2] = {
    /* 32-bit tuning.  */
    {libcall, {{8, loop, false},
           {24, unrolled_loop, false},
-         {2048, rep_prefix_4_byte, false},
+         {128, rep_prefix_4_byte, false},
           {-1, libcall, false}}},
    /* 64-bit tuning.  */
    {libcall, {{48, unrolled_loop, false},
-         {8192, rep_prefix_8_byte, false},
+         {128, rep_prefix_8_byte, false},
           {-1, libcall, false}}}};
  struct processor_costs znver1_cost = {
    {
@@ -1460,7 +1460,7 @@ static stringop_algs znver2_memcpy[2] = {
    /* 32-bit tuning.  */
    {libcall, {{6, loop, false},
           {14, unrolled_loop, false},
-         {-1, rep_prefix_4_byte, false}}},
+         {-1, libcall, false}}},
    /* 64-bit tuning.  */
    {libcall, {{16, loop, false},
           {64, rep_prefix_4_byte, false},
@@ -1469,7 +1469,7 @@ static stringop_algs znver2_memset[2] = {
    /* 32-bit tuning.  */
    {libcall, {{8, loop, false},
           {24, unrolled_loop, false},
-         {2048, rep_prefix_4_byte, false}
+         {128, rep_prefix_4_byte, false},
           {-1, libcall, false}}},
    /* 64-bit tuning.  */
    {libcall, {{24, rep_prefix_4_byte, false},