Re: [PATCH] softfloat: Mark base int-to-float routines QEMU_FLATTEN

2020-10-19 Thread Alex Bennée 


Richard Henderson  writes:

> This merges the int_to_float routine and the round_pack_canonical
> routine into the same function, allowing the FloatParts structure
> to be decomposed by the compiler.
>
> This results in a 60-75% speedup of the flattened function.
>
> Leave the narrower integer inputs to tail-call the int64_t version.
>
> Buglink: https://bugs.launchpad.net/qemu/+bug/1892081
> Signed-off-by: Richard Henderson 

Reviewed-by: Alex Bennée 

-- 
Alex Bennée

Re: [PATCH] softfloat: Mark base int-to-float routines QEMU_FLATTEN

2020-10-19 Thread Philippe Mathieu-Daudé 

On 10/18/20 10:33 PM, Richard Henderson wrote:

This merges the int_to_float routine and the round_pack_canonical
routine into the same function, allowing the FloatParts structure
to be decomposed by the compiler.

This results in a 60-75% speedup of the flattened function.

Leave the narrower integer inputs to tail-call the int64_t version.



Reported-by: Ahmed Karaman 


Buglink: https://bugs.launchpad.net/qemu/+bug/1892081
Signed-off-by: Richard Henderson 


Reviewed-by: Philippe Mathieu-Daudé 


---
  fpu/softfloat.c | 24 
  1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/fpu/softfloat.c b/fpu/softfloat.c
index 7b6aee9323..2cbcf5bf10 100644
--- a/fpu/softfloat.c
+++ b/fpu/softfloat.c
@@ -2794,7 +2794,8 @@ static FloatParts int_to_float(int64_t a, int scale, 
float_status *status)
  return r;
  }
  
-float16 int64_to_float16_scalbn(int64_t a, int scale, float_status *status)

+float16 QEMU_FLATTEN
+int64_to_float16_scalbn(int64_t a, int scale, float_status *status)
  {
  FloatParts pa = int_to_float(a, scale, status);
  return float16_round_pack_canonical(pa, status);
@@ -2830,7 +2831,8 @@ float16 int8_to_float16(int8_t a, float_status *status)
  return int64_to_float16_scalbn(a, 0, status);
  }
  
-float32 int64_to_float32_scalbn(int64_t a, int scale, float_status *status)

+float32 QEMU_FLATTEN
+int64_to_float32_scalbn(int64_t a, int scale, float_status *status)
  {
  FloatParts pa = int_to_float(a, scale, status);
  return float32_round_pack_canonical(pa, status);
@@ -2861,7 +2863,8 @@ float32 int16_to_float32(int16_t a, float_status *status)
  return int64_to_float32_scalbn(a, 0, status);
  }
  
-float64 int64_to_float64_scalbn(int64_t a, int scale, float_status *status)

+float64 QEMU_FLATTEN
+int64_to_float64_scalbn(int64_t a, int scale, float_status *status)
  {
  FloatParts pa = int_to_float(a, scale, status);
  return float64_round_pack_canonical(pa, status);
@@ -2897,7 +2900,8 @@ float64 int16_to_float64(int16_t a, float_status *status)
   * to the bfloat16 format.
   */
  
-bfloat16 int64_to_bfloat16_scalbn(int64_t a, int scale, float_status *status)

+bfloat16 QEMU_FLATTEN
+int64_to_bfloat16_scalbn(int64_t a, int scale, float_status *status)
  {
  FloatParts pa = int_to_float(a, scale, status);
  return bfloat16_round_pack_canonical(pa, status);
@@ -2959,7 +2963,8 @@ static FloatParts uint_to_float(uint64_t a, int scale, 
float_status *status)
  return r;
  }
  
-float16 uint64_to_float16_scalbn(uint64_t a, int scale, float_status *status)

+float16 QEMU_FLATTEN
+uint64_to_float16_scalbn(uint64_t a, int scale, float_status *status)
  {
  FloatParts pa = uint_to_float(a, scale, status);
  return float16_round_pack_canonical(pa, status);
@@ -2995,7 +3000,8 @@ float16 uint8_to_float16(uint8_t a, float_status *status)
  return uint64_to_float16_scalbn(a, 0, status);
  }
  
-float32 uint64_to_float32_scalbn(uint64_t a, int scale, float_status *status)

+float32 QEMU_FLATTEN
+uint64_to_float32_scalbn(uint64_t a, int scale, float_status *status)
  {
  FloatParts pa = uint_to_float(a, scale, status);
  return float32_round_pack_canonical(pa, status);
@@ -3026,7 +3032,8 @@ float32 uint16_to_float32(uint16_t a, float_status 
*status)
  return uint64_to_float32_scalbn(a, 0, status);
  }
  
-float64 uint64_to_float64_scalbn(uint64_t a, int scale, float_status *status)

+float64 QEMU_FLATTEN
+uint64_to_float64_scalbn(uint64_t a, int scale, float_status *status)
  {
  FloatParts pa = uint_to_float(a, scale, status);
  return float64_round_pack_canonical(pa, status);
@@ -3062,7 +3069,8 @@ float64 uint16_to_float64(uint16_t a, float_status 
*status)
   * bfloat16 format.
   */
  
-bfloat16 uint64_to_bfloat16_scalbn(uint64_t a, int scale, float_status *status)

+bfloat16 QEMU_FLATTEN
+uint64_to_bfloat16_scalbn(uint64_t a, int scale, float_status *status)
  {
  FloatParts pa = uint_to_float(a, scale, status);
  return bfloat16_round_pack_canonical(pa, status);

Re: [PATCH] softfloat: Mark base int-to-float routines QEMU_FLATTEN

2020-10-18 Thread no-reply 
Patchew URL: 
https://patchew.org/QEMU/20201018203334.1229243-1-richard.hender...@linaro.org/



Hi,

This series seems to have some coding style problems. See output below for
more information:

Type: series
Message-id: 20201018203334.1229243-1-richard.hender...@linaro.org
Subject: [PATCH] softfloat: Mark base int-to-float routines QEMU_FLATTEN

=== TEST SCRIPT BEGIN ===
#!/bin/bash
git rev-parse base > /dev/null || exit 0
git config --local diff.renamelimit 0
git config --local diff.renames True
git config --local diff.algorithm histogram
./scripts/checkpatch.pl --mailback base..
=== TEST SCRIPT END ===

Updating 3c8cf5a9c21ff8782164d1def7f44bd888713384
Switched to a new branch 'test'
4ffd616 softfloat: Mark base int-to-float routines QEMU_FLATTEN

=== OUTPUT BEGIN ===
ERROR: spaces required around that '*' (ctx:WxV)
#29: FILE: fpu/softfloat.c:2779:
+int64_to_float16_scalbn(int64_t a, int scale, float_status *status)
^

ERROR: spaces required around that '*' (ctx:WxV)
#39: FILE: fpu/softfloat.c:2816:
+int64_to_float32_scalbn(int64_t a, int scale, float_status *status)
^

ERROR: spaces required around that '*' (ctx:WxV)
#49: FILE: fpu/softfloat.c:2848:
+int64_to_float64_scalbn(int64_t a, int scale, float_status *status)
^

ERROR: spaces required around that '*' (ctx:WxV)
#59: FILE: fpu/softfloat.c:2885:
+int64_to_bfloat16_scalbn(int64_t a, int scale, float_status *status)
 ^

ERROR: spaces required around that '*' (ctx:WxV)
#69: FILE: fpu/softfloat.c:2948:
+uint64_to_float16_scalbn(uint64_t a, int scale, float_status *status)
  ^

ERROR: spaces required around that '*' (ctx:WxV)
#79: FILE: fpu/softfloat.c:2985:
+uint64_to_float32_scalbn(uint64_t a, int scale, float_status *status)
  ^

ERROR: spaces required around that '*' (ctx:WxV)
#89: FILE: fpu/softfloat.c:3017:
+uint64_to_float64_scalbn(uint64_t a, int scale, float_status *status)
  ^

ERROR: spaces required around that '*' (ctx:WxV)
#99: FILE: fpu/softfloat.c:3054:
+uint64_to_bfloat16_scalbn(uint64_t a, int scale, float_status *status)
   ^

total: 8 errors, 0 warnings, 72 lines checked

Commit 4ffd616f4c5d (softfloat: Mark base int-to-float routines QEMU_FLATTEN) 
has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
=== OUTPUT END ===

Test command exited with code: 1


The full log is available at
http://patchew.org/logs/20201018203334.1229243-1-richard.hender...@linaro.org/testing.checkpatch/?type=message.
---
Email generated automatically by Patchew [https://patchew.org/].
Please send your feedback to patchew-de...@redhat.com