[1/3] arm: vp9itxfm: Only reload the idct coeffs for the iadst_idct combination

Message ID 1480325162-7688-1-git-send-email-martin@martin.st
State Committed
Commit 3c87039a404c5659ae9bf7454a04e186532eb40b
Headers show

Commit Message

Martin Storsjö Nov. 28, 2016, 9:26 a.m.
This avoids reloading them if they haven't been clobbered, if the
first pass also was idct.

This is similar to what was done in the aarch64 version.
---
 libavcodec/arm/vp9itxfm_neon.S | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Janne Grunau Nov. 30, 2016, 7:29 p.m. | #1
On 2016-11-28 11:26:00 +0200, Martin Storsjö wrote:
> This avoids reloading them if they haven't been clobbered, if the
> first pass also was idct.
> 
> This is similar to what was done in the aarch64 version.
> ---
>  libavcodec/arm/vp9itxfm_neon.S | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/libavcodec/arm/vp9itxfm_neon.S b/libavcodec/arm/vp9itxfm_neon.S
> index 01944bd..2049241 100644
> --- a/libavcodec/arm/vp9itxfm_neon.S
> +++ b/libavcodec/arm/vp9itxfm_neon.S
> @@ -814,7 +814,7 @@ A       and             r7,  sp,  #15
>          mov             r3,  #\i
>          bl              \txfm1\()16_1d_4x16_pass1_neon
>  .endr
> -.ifc \txfm2,idct
> +.ifc \txfm1\()_\txfm2,iadst_idct
>          movrel          r12, idct_coeffs
>          vld1.16         {q0-q1}, [r12,:128]
>  .endif

ok

Janne

Patch

diff --git a/libavcodec/arm/vp9itxfm_neon.S b/libavcodec/arm/vp9itxfm_neon.S
index 01944bd..2049241 100644
--- a/libavcodec/arm/vp9itxfm_neon.S
+++ b/libavcodec/arm/vp9itxfm_neon.S
@@ -814,7 +814,7 @@  A       and             r7,  sp,  #15
         mov             r3,  #\i
         bl              \txfm1\()16_1d_4x16_pass1_neon
 .endr
-.ifc \txfm2,idct
+.ifc \txfm1\()_\txfm2,iadst_idct
         movrel          r12, idct_coeffs
         vld1.16         {q0-q1}, [r12,:128]
 .endif