public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc(refs/users/meissner/heads/dmf004)] Add parameter for memcpy inline code moves
@ 2022-11-15 18:54 Michael Meissner
  0 siblings, 0 replies; 4+ messages in thread
From: Michael Meissner @ 2022-11-15 18:54 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:cf83c4f88df6149adbd23dd1e80c5ec328ab7afb

commit cf83c4f88df6149adbd23dd1e80c5ec328ab7afb
Author: Michael Meissner <meissner@linux.ibm.com>
Date:   Tue Nov 15 13:54:16 2022 -0500

    Add parameter for memcpy inline code moves
    
    2022-11-15   Michael Meissner  <meissner@linux.ibm.com>
    
    gcc/
    
            * config/rs6000/rs6000.opt (rs6000-memcpy-inline-bytes): New parameter,
            set to 0.
            * config/rs6000/rs6000-string.cc (expand_block_move): Only do
            optimization if rs6000-memcpy-inline-bytes <= 16.

Diff:
---
 gcc/config/rs6000/rs6000-string.cc | 6 ++++--
 gcc/config/rs6000/rs6000.opt       | 5 +++++
 2 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/gcc/config/rs6000/rs6000-string.cc b/gcc/config/rs6000/rs6000-string.cc
index 2468e375781..243be4f396d 100644
--- a/gcc/config/rs6000/rs6000-string.cc
+++ b/gcc/config/rs6000/rs6000-string.cc
@@ -2766,7 +2766,9 @@ expand_block_move (rtx operands[], bool might_overlap)
   if (! constp)
     {
       if (TARGET_BLOCK_OPS_UNALIGNED_VSX && TARGET_P9_VECTOR && TARGET_64BIT
-	  && !optimize_size)
+	  && rs6000_memcpy_inline_bytes > 0
+	  && rs6000_memcpy_inline_bytes <= GET_MODE_SIZE (V16QImode)
+	  && optimize && !optimize_size)
 	{
 	  rtx join_label = gen_label_rtx ();
 	  rtx inline_label = gen_label_rtx ();
@@ -2779,7 +2781,7 @@ expand_block_move (rtx operands[], bool might_overlap)
 		       : convert_to_mode (Pmode, bytes_rtx, true));
 
 	  rtx cr = gen_reg_rtx (CCUNSmode);
-	  rtx max_size = GEN_INT (16);
+	  rtx max_size = GEN_INT (rs6000_memcpy_inline_bytes);
 	  emit_insn (gen_rtx_SET (cr,
 				  gen_rtx_COMPARE (CCUNSmode, bytes_rtx,
 						   max_size)));
diff --git a/gcc/config/rs6000/rs6000.opt b/gcc/config/rs6000/rs6000.opt
index 6872d359952..90dc91a277f 100644
--- a/gcc/config/rs6000/rs6000.opt
+++ b/gcc/config/rs6000/rs6000.opt
@@ -687,3 +687,8 @@ default value is 4.
 Target Undocumented Joined UInteger Var(rs6000_vect_unroll_reduc_threshold) Init(1) Param
 When reduction factor computed for a loop exceeds the threshold specified by
 this parameter, prefer to unroll this loop.  The default value is 1.
+
+-param=rs6000-memcpy-inline-bytes=
+Target Undocumented Joined UInteger Var(rs6000_memcpy_inline_bytes) Init(0) Param
+Maximum number of bytes to move with inline code before calling the memcpy
+library function.  The default value is 0.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [gcc(refs/users/meissner/heads/dmf004)] Add parameter for memcpy inline code moves
@ 2022-11-17 21:54 Michael Meissner
  0 siblings, 0 replies; 4+ messages in thread
From: Michael Meissner @ 2022-11-17 21:54 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:92f5b212e8b335680dddc344a563f4f973389f01

commit 92f5b212e8b335680dddc344a563f4f973389f01
Author: Michael Meissner <meissner@linux.ibm.com>
Date:   Tue Nov 15 13:54:16 2022 -0500

    Add parameter for memcpy inline code moves
    
    2022-11-15   Michael Meissner  <meissner@linux.ibm.com>
    
    gcc/
    
            * config/rs6000/rs6000.opt (rs6000-memcpy-inline-bytes): New parameter,
            set to 0.
            * config/rs6000/rs6000-string.cc (expand_block_move): Only do
            optimization if rs6000-memcpy-inline-bytes <= 16.

Diff:
---
 gcc/config/rs6000/rs6000-string.cc | 6 ++++--
 gcc/config/rs6000/rs6000.opt       | 5 +++++
 2 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/gcc/config/rs6000/rs6000-string.cc b/gcc/config/rs6000/rs6000-string.cc
index 2468e375781..243be4f396d 100644
--- a/gcc/config/rs6000/rs6000-string.cc
+++ b/gcc/config/rs6000/rs6000-string.cc
@@ -2766,7 +2766,9 @@ expand_block_move (rtx operands[], bool might_overlap)
   if (! constp)
     {
       if (TARGET_BLOCK_OPS_UNALIGNED_VSX && TARGET_P9_VECTOR && TARGET_64BIT
-	  && !optimize_size)
+	  && rs6000_memcpy_inline_bytes > 0
+	  && rs6000_memcpy_inline_bytes <= GET_MODE_SIZE (V16QImode)
+	  && optimize && !optimize_size)
 	{
 	  rtx join_label = gen_label_rtx ();
 	  rtx inline_label = gen_label_rtx ();
@@ -2779,7 +2781,7 @@ expand_block_move (rtx operands[], bool might_overlap)
 		       : convert_to_mode (Pmode, bytes_rtx, true));
 
 	  rtx cr = gen_reg_rtx (CCUNSmode);
-	  rtx max_size = GEN_INT (16);
+	  rtx max_size = GEN_INT (rs6000_memcpy_inline_bytes);
 	  emit_insn (gen_rtx_SET (cr,
 				  gen_rtx_COMPARE (CCUNSmode, bytes_rtx,
 						   max_size)));
diff --git a/gcc/config/rs6000/rs6000.opt b/gcc/config/rs6000/rs6000.opt
index 6872d359952..90dc91a277f 100644
--- a/gcc/config/rs6000/rs6000.opt
+++ b/gcc/config/rs6000/rs6000.opt
@@ -687,3 +687,8 @@ default value is 4.
 Target Undocumented Joined UInteger Var(rs6000_vect_unroll_reduc_threshold) Init(1) Param
 When reduction factor computed for a loop exceeds the threshold specified by
 this parameter, prefer to unroll this loop.  The default value is 1.
+
+-param=rs6000-memcpy-inline-bytes=
+Target Undocumented Joined UInteger Var(rs6000_memcpy_inline_bytes) Init(0) Param
+Maximum number of bytes to move with inline code before calling the memcpy
+library function.  The default value is 0.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [gcc(refs/users/meissner/heads/dmf004)] Add parameter for memcpy inline code moves
@ 2022-11-17 21:54 Michael Meissner
  0 siblings, 0 replies; 4+ messages in thread
From: Michael Meissner @ 2022-11-17 21:54 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:b65e4974ba99db96038cd29363f650ec482bafea

commit b65e4974ba99db96038cd29363f650ec482bafea
Author: Michael Meissner <meissner@linux.ibm.com>
Date:   Tue Nov 15 12:27:09 2022 -0500

    Add parameter for memcpy inline code moves
    
    2022-11-15   Michael Meissner  <meissner@linux.ibm.com>
    
    gcc/
    
            * config/rs6000/rs6000.opt (rs6000-memcpy-inline-bytes): New parameter,
            set to 0.
            * config/rs6000/rs6000-string.cc (expand_block_move): Only do
            optimization if rs6000-memcpy-inline-bytes is 16.

Diff:
---
 gcc/config/rs6000/rs6000-string.cc | 1 +
 gcc/config/rs6000/rs6000.opt       | 5 +++++
 2 files changed, 6 insertions(+)

diff --git a/gcc/config/rs6000/rs6000-string.cc b/gcc/config/rs6000/rs6000-string.cc
index 2468e375781..1fa2d1fe553 100644
--- a/gcc/config/rs6000/rs6000-string.cc
+++ b/gcc/config/rs6000/rs6000-string.cc
@@ -2766,6 +2766,7 @@ expand_block_move (rtx operands[], bool might_overlap)
   if (! constp)
     {
       if (TARGET_BLOCK_OPS_UNALIGNED_VSX && TARGET_P9_VECTOR && TARGET_64BIT
+	  && rs6000_memcpy_inline_bytes == 16
 	  && !optimize_size)
 	{
 	  rtx join_label = gen_label_rtx ();
diff --git a/gcc/config/rs6000/rs6000.opt b/gcc/config/rs6000/rs6000.opt
index 6872d359952..90dc91a277f 100644
--- a/gcc/config/rs6000/rs6000.opt
+++ b/gcc/config/rs6000/rs6000.opt
@@ -687,3 +687,8 @@ default value is 4.
 Target Undocumented Joined UInteger Var(rs6000_vect_unroll_reduc_threshold) Init(1) Param
 When reduction factor computed for a loop exceeds the threshold specified by
 this parameter, prefer to unroll this loop.  The default value is 1.
+
+-param=rs6000-memcpy-inline-bytes=
+Target Undocumented Joined UInteger Var(rs6000_memcpy_inline_bytes) Init(0) Param
+Maximum number of bytes to move with inline code before calling the memcpy
+library function.  The default value is 0.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [gcc(refs/users/meissner/heads/dmf004)] Add parameter for memcpy inline code moves
@ 2022-11-15 17:27 Michael Meissner
  0 siblings, 0 replies; 4+ messages in thread
From: Michael Meissner @ 2022-11-15 17:27 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:e1efca6e71ebf7c6fed3eced75a2cd1cb62e3667

commit e1efca6e71ebf7c6fed3eced75a2cd1cb62e3667
Author: Michael Meissner <meissner@linux.ibm.com>
Date:   Tue Nov 15 12:27:09 2022 -0500

    Add parameter for memcpy inline code moves
    
    2022-11-15   Michael Meissner  <meissner@linux.ibm.com>
    
    gcc/
    
            * config/rs6000/rs6000.opt (rs6000-memcpy-inline-bytes): New parameter,
            set to 0.
            * config/rs6000/rs6000-string.cc (expand_block_move): Only do
            optimization if rs6000-memcpy-inline-bytes is 16.

Diff:
---
 gcc/config/rs6000/rs6000-string.cc | 1 +
 gcc/config/rs6000/rs6000.opt       | 5 +++++
 2 files changed, 6 insertions(+)

diff --git a/gcc/config/rs6000/rs6000-string.cc b/gcc/config/rs6000/rs6000-string.cc
index 2468e375781..1fa2d1fe553 100644
--- a/gcc/config/rs6000/rs6000-string.cc
+++ b/gcc/config/rs6000/rs6000-string.cc
@@ -2766,6 +2766,7 @@ expand_block_move (rtx operands[], bool might_overlap)
   if (! constp)
     {
       if (TARGET_BLOCK_OPS_UNALIGNED_VSX && TARGET_P9_VECTOR && TARGET_64BIT
+	  && rs6000_memcpy_inline_bytes == 16
 	  && !optimize_size)
 	{
 	  rtx join_label = gen_label_rtx ();
diff --git a/gcc/config/rs6000/rs6000.opt b/gcc/config/rs6000/rs6000.opt
index 6872d359952..90dc91a277f 100644
--- a/gcc/config/rs6000/rs6000.opt
+++ b/gcc/config/rs6000/rs6000.opt
@@ -687,3 +687,8 @@ default value is 4.
 Target Undocumented Joined UInteger Var(rs6000_vect_unroll_reduc_threshold) Init(1) Param
 When reduction factor computed for a loop exceeds the threshold specified by
 this parameter, prefer to unroll this loop.  The default value is 1.
+
+-param=rs6000-memcpy-inline-bytes=
+Target Undocumented Joined UInteger Var(rs6000_memcpy_inline_bytes) Init(0) Param
+Maximum number of bytes to move with inline code before calling the memcpy
+library function.  The default value is 0.

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2022-11-17 21:54 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-11-15 18:54 [gcc(refs/users/meissner/heads/dmf004)] Add parameter for memcpy inline code moves Michael Meissner
  -- strict thread matches above, loose matches on Subject: below --
2022-11-17 21:54 Michael Meissner
2022-11-17 21:54 Michael Meissner
2022-11-15 17:27 Michael Meissner

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).