summaryrefslogtreecommitdiff
path: root/libavcodec/aarch64/vp8dsp_neon.S
diff options
context:
space:
mode:
Diffstat (limited to 'libavcodec/aarch64/vp8dsp_neon.S')
-rw-r--r--libavcodec/aarch64/vp8dsp_neon.S292
1 files changed, 292 insertions, 0 deletions
diff --git a/libavcodec/aarch64/vp8dsp_neon.S b/libavcodec/aarch64/vp8dsp_neon.S
index 2c86eef1e5..aefe8fdcde 100644
--- a/libavcodec/aarch64/vp8dsp_neon.S
+++ b/libavcodec/aarch64/vp8dsp_neon.S
@@ -1509,3 +1509,295 @@ function ff_put_vp8_epel4_h4v4_neon, export=1
add sp, sp, #44
ret
endfunc
+
+/* Bilinear MC */
+
+function ff_put_vp8_bilin16_h_neon, export=1
+ mov w7, #8
+ dup v0.8b, w5
+ sub w5, w7, w5
+ dup v1.8b, w5
+1:
+ subs w4, w4, #2
+ ld1 {v2.8b,v3.8b,v4.8b}, [x2], x3
+ ext v5.8b, v3.8b, v4.8b, #1
+ ext v4.8b, v2.8b, v3.8b, #1
+ umull v16.8h, v2.8b, v1.8b
+ umlal v16.8h, v4.8b, v0.8b
+ ld1 {v18.8b,v19.8b,v20.8b}, [x2], x3
+ umull v6.8h, v3.8b, v1.8b
+ umlal v6.8h, v5.8b, v0.8b
+ ext v21.8b, v19.8b, v20.8b, #1
+ ext v20.8b, v18.8b, v19.8b, #1
+ umull v22.8h, v18.8b, v1.8b
+ umlal v22.8h, v20.8b, v0.8b
+ umull v24.8h, v19.8b, v1.8b
+ umlal v24.8h, v21.8b, v0.8b
+ rshrn v4.8b, v16.8h, #3
+ rshrn2 v4.16b, v6.8h, #3
+ rshrn v6.8b, v22.8h, #3
+ rshrn2 v6.16b, v24.8h, #3
+ st1 {v4.16b}, [x0], x1
+ st1 {v6.16b}, [x0], x1
+ b.gt 1b
+
+ ret
+endfunc
+
+function ff_put_vp8_bilin16_v_neon, export=1
+ mov w7, #8
+ dup v0.16b, w6
+ sub w6, w7, w6
+ dup v1.16b, w6
+
+ ld1 {v2.16b}, [x2], x3
+1:
+ subs w4, w4, #2
+ ld1 {v4.16b}, [x2], x3
+ umull v6.8h, v2.8b, v1.8b
+ umlal v6.8h, v4.8b, v0.8b
+ umull2 v16.8h, v2.16b, v1.16b
+ umlal2 v16.8h, v4.16b, v0.16b
+ ld1 {v2.16b}, [x2], x3
+ umull v18.8h, v4.8b, v1.8b
+ umlal v18.8h, v2.8b, v0.8b
+ umull2 v20.8h, v4.16b, v1.16b
+ umlal2 v20.8h, v2.16b, v0.16b
+ rshrn v4.8b, v6.8h, #3
+ rshrn2 v4.16b, v16.8h, #3
+ rshrn v6.8b, v18.8h, #3
+ rshrn2 v6.16b, v20.8h, #3
+ st1 {v4.16b}, [x0], x1
+ st1 {v6.16b}, [x0], x1
+ b.gt 1b
+
+ ret
+endfunc
+
+function ff_put_vp8_bilin16_hv_neon, export=1
+ mov w7, #8
+ dup v0.8b, w5 // mx
+ sub w5, w7, w5
+ dup v1.8b, w5
+ dup v2.16b, w6 // my
+ sub w6, w7, w6
+ dup v3.16b, w6
+
+ ld1 {v4.8b,v5.8b,v6.8b}, [x2], x3
+
+ ext v7.8b, v5.8b, v6.8b, #1
+ ext v6.8b, v4.8b, v5.8b, #1
+ umull v16.8h, v4.8b, v1.8b
+ umlal v16.8h, v6.8b, v0.8b
+ umull v18.8h, v5.8b, v1.8b
+ umlal v18.8h, v7.8b, v0.8b
+ rshrn v4.8b, v16.8h, #3
+ rshrn2 v4.16b, v18.8h, #3
+1:
+ subs w4, w4, #2
+ ld1 {v18.8b,v19.8b,v20.8b}, [x2], x3
+ ext v21.8b, v19.8b, v20.8b, #1
+ ext v20.8b, v18.8b, v19.8b, #1
+ umull v22.8h, v18.8b, v1.8b
+ umlal v22.8h, v20.8b, v0.8b
+ ld1 {v26.8b,v27.8b,v28.8b}, [x2], x3
+ umull v24.8h, v19.8b, v1.8b
+ umlal v24.8h, v21.8b, v0.8b
+ ext v29.8b, v27.8b, v28.8b, #1
+ ext v28.8b, v26.8b, v27.8b, #1
+ umull v16.8h, v26.8b, v1.8b
+ umlal v16.8h, v28.8b, v0.8b
+ umull v18.8h, v27.8b, v1.8b
+ umlal v18.8h, v29.8b, v0.8b
+ rshrn v6.8b, v22.8h, #3
+ rshrn2 v6.16b, v24.8h, #3
+ umull v24.8h, v4.8b, v3.8b
+ umlal v24.8h, v6.8b, v2.8b
+ umull2 v30.8h, v4.16b, v3.16b
+ umlal2 v30.8h, v6.16b, v2.16b
+ rshrn v4.8b, v16.8h, #3
+ rshrn2 v4.16b, v18.8h, #3
+ umull v20.8h, v6.8b, v3.8b
+ umlal v20.8h, v4.8b, v2.8b
+ umull2 v22.8h, v6.16b, v3.16b
+ umlal2 v22.8h, v4.16b, v2.16b
+ rshrn v24.8b, v24.8h, #3
+ rshrn2 v24.16b, v30.8h, #3
+ st1 {v24.16b}, [x0], x1
+ rshrn v20.8b, v20.8h, #3
+ rshrn2 v20.16b, v22.8h, #3
+ st1 {v20.16b}, [x0], x1
+ b.gt 1b
+
+ ret
+endfunc
+
+function ff_put_vp8_bilin8_h_neon, export=1
+ mov w7, #8
+ dup v0.8b, w5
+ sub w5, w7, w5
+ dup v1.8b, w5
+1:
+ subs w4, w4, #2
+ ld1 {v2.8b,v3.8b}, [x2], x3
+ ext v3.8b, v2.8b, v3.8b, #1
+ umull v4.8h, v2.8b, v1.8b
+ umlal v4.8h, v3.8b, v0.8b
+ ld1 {v6.8b,v7.8b}, [x2], x3
+ ext v7.8b, v6.8b, v7.8b, #1
+ umull v16.8h, v6.8b, v1.8b
+ umlal v16.8h, v7.8b, v0.8b
+ rshrn v4.8b, v4.8h, #3
+ rshrn v16.8b, v16.8h, #3
+ st1 {v4.8b}, [x0], x1
+ st1 {v16.8b}, [x0], x1
+ b.gt 1b
+
+ ret
+endfunc
+
+function ff_put_vp8_bilin8_v_neon, export=1
+ mov w7, #8
+ dup v0.8b, w6
+ sub w6, w7, w6
+ dup v1.8b, w6
+
+ ld1 {v2.8b}, [x2], x3
+1:
+ subs w4, w4, #2
+ ld1 {v3.8b}, [x2], x3
+ umull v4.8h, v2.8b, v1.8b
+ umlal v4.8h, v3.8b, v0.8b
+ ld1 {v2.8b}, [x2], x3
+ umull v6.8h, v3.8b, v1.8b
+ umlal v6.8h, v2.8b, v0.8b
+ rshrn v4.8b, v4.8h, #3
+ rshrn v6.8b, v6.8h, #3
+ st1 {v4.8b}, [x0], x1
+ st1 {v6.8b}, [x0], x1
+ b.gt 1b
+
+ ret
+endfunc
+
+function ff_put_vp8_bilin8_hv_neon, export=1
+ mov w7, #8
+ dup v0.8b, w5 // mx
+ sub w5, w7, w5
+ dup v1.8b, w5
+ dup v2.8b, w6 // my
+ sub w6, w7, w6
+ dup v3.8b, w6
+
+ ld1 {v4.8b,v5.8b}, [x2], x3
+ ext v5.8b, v4.8b, v5.8b, #1
+ umull v18.8h, v4.8b, v1.8b
+ umlal v18.8h, v5.8b, v0.8b
+ rshrn v22.8b, v18.8h, #3
+1:
+ subs w4, w4, #2
+ ld1 {v6.8b,v7.8b}, [x2], x3
+ ext v7.8b, v6.8b, v7.8b, #1
+ umull v16.8h, v6.8b, v1.8b
+ umlal v16.8h, v7.8b, v0.8b
+ ld1 {v4.8b,v5.8b}, [x2], x3
+ ext v5.8b, v4.8b, v5.8b, #1
+ umull v18.8h, v4.8b, v1.8b
+ umlal v18.8h, v5.8b, v0.8b
+ rshrn v16.8b, v16.8h, #3
+ umull v20.8h, v22.8b, v3.8b
+ umlal v20.8h, v16.8b, v2.8b
+ rshrn v22.8b, v18.8h, #3
+ umull v24.8h, v16.8b, v3.8b
+ umlal v24.8h, v22.8b, v2.8b
+ rshrn v20.8b, v20.8h, #3
+ st1 {v20.8b}, [x0], x1
+ rshrn v23.8b, v24.8h, #3
+ st1 {v23.8b}, [x0], x1
+ b.gt 1b
+
+ ret
+endfunc
+
+function ff_put_vp8_bilin4_h_neon, export=1
+ mov w7, #8
+ dup v0.8b, w5
+ sub w5, w7, w5
+ dup v1.8b, w5
+1:
+ subs w4, w4, #2
+ ld1 {v2.8b}, [x2], x3
+ ext v3.8b, v2.8b, v3.8b, #1
+ ld1 {v6.8b}, [x2], x3
+ ext v7.8b, v6.8b, v7.8b, #1
+ trn1 v2.2s, v2.2s, v6.2s
+ trn1 v3.2s, v3.2s, v7.2s
+ umull v4.8h, v2.8b, v1.8b
+ umlal v4.8h, v3.8b, v0.8b
+ rshrn v4.8b, v4.8h, #3
+ st1 {v4.s}[0], [x0], x1
+ st1 {v4.s}[1], [x0], x1
+ b.gt 1b
+
+ ret
+endfunc
+
+function ff_put_vp8_bilin4_v_neon, export=1
+ mov w7, #8
+ dup v0.8b, w6
+ sub w6, w7, w6
+ dup v1.8b, w6
+
+ ld1r {v2.2s}, [x2], x3
+1:
+ ld1r {v3.2s}, [x2]
+ ld1 {v2.s}[1], [x2], x3
+ ld1 {v3.s}[1], [x2], x3
+ umull v4.8h, v2.8b, v1.8b
+ umlal v4.8h, v3.8b, v0.8b
+ trn2 v2.2s, v3.2s, v2.2s
+ rshrn v4.8b, v4.8h, #3
+ st1 {v4.s}[0], [x0], x1
+ st1 {v4.s}[1], [x0], x1
+ subs w4, w4, #2
+ b.gt 1b
+
+ ret
+endfunc
+
+function ff_put_vp8_bilin4_hv_neon, export=1
+ mov w7, #8
+ dup v0.8b, w5 // mx
+ sub w5, w7, w5
+ dup v1.8b, w5
+ dup v2.8b, w6 // my
+ sub w6, w7, w6
+ dup v3.8b, w6
+
+ ld1 {v4.8b}, [x2], x3
+ ext v5.8b, v4.8b, v4.8b, #1
+ umull v18.8h, v4.8b, v1.8b
+ umlal v18.8h, v5.8b, v0.8b
+ rshrn v22.8b, v18.8h, #3
+1:
+ subs w4, w4, #2
+ ld1 {v6.8b}, [x2], x3
+ ext v7.8b, v6.8b, v6.8b, #1
+ ld1 {v4.8b}, [x2], x3
+ ext v5.8b, v4.8b, v4.8b, #1
+ trn1 v6.2s, v6.2s, v4.2s
+ trn1 v7.2s, v7.2s, v5.2s
+ umull v16.8h, v6.8b, v1.8b
+ umlal v16.8h, v7.8b, v0.8b
+ rshrn v16.8b, v16.8h, #3
+ umull v20.8h, v16.8b, v2.8b
+ trn1 v22.2s, v22.2s, v16.2s
+ umlal v20.8h, v22.8b, v3.8b
+ rev64 v22.2s, v16.2s
+ rshrn v20.8b, v20.8h, #3
+ st1 {v20.s}[0], [x0], x1
+ st1 {v20.s}[1], [x0], x1
+ b.gt 1b
+
+ ret
+endfunc
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-16 22:08:06 +0000