On 10/13/16 16:02, Alexandra Hájková wrote:
From: Pierre Edouard Lepere <pierre-edouard.lep...@insa-rennes.fr>

Initially written by Pierre Edouard Lepere 
<pierre-edouard.lep...@insa-rennes.fr>,
extended by James Almer <jamr...@gmail.com>.

Signed-off-by: Alexandra Hájková <alexan...@khirnov.net>
---
 libavcodec/x86/Makefile         |   3 +-
 libavcodec/x86/hevc_res_add.asm | 391 ++++++++++++++++++++++++++++++++++++++++

Maybe hevc_add_res.asm for consistency?

+%if HAVE_AVX2_EXTERNAL
+INIT_YMM avx2
+
+cglobal hevc_add_residual_16_10,3,5,6
+    pxor              m4, m4
+    mova              m5, [max_pixels_10]
+    lea               r3, [r2*3]
+
+    mov r4d, 4
+.loop
+    ADD_RESIDUAL_16_AVX2  r0, r2, r3, r1
+    lea               r0, [r0+r2*4]
+    add               r1, 128
+    dec r4d
+    jnz .loop
+    RET
+
+cglobal hevc_add_residual_32_10,3,5,6
+    pxor              m4, m4
+    mova              m5, [max_pixels_10]
+
+    mov r4d, 6

16

+.loop
+    ADD_RESIDUAL_32_AVX2  r0, r2, r1
+    lea               r0, [r0+r2*2]
+    add               r1, 128
+    dec r4d
+    jnz .loop
+    RET
+%endif ;HAVE_AVX2_EXTERNAL

With that change it passes checkasm on avx2 as well.

lu
_______________________________________________
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to