arm: vp9itxfm: Only reload the idct coeffs for the iadst_idct combination
authorMartin Storsjö <martin@martin.st>
Mon, 28 Nov 2016 09:05:18 +0000 (11:05 +0200)
committerMartin Storsjö <martin@martin.st>
Wed, 30 Nov 2016 21:53:52 +0000 (23:53 +0200)
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.

Signed-off-by: Martin Storsjö <martin@martin.st>
libavcodec/arm/vp9itxfm_neon.S

index 01944bd..2049241 100644 (file)
@@ -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