[x265] [PATCH 292 of 307] x86:AVX512 intra_pred_ang32 mode 5 and 31 high bit depth
mythreyi at multicorewareinc.com
mythreyi at multicorewareinc.com
Sat Apr 7 04:34:50 CEST 2018
# HG changeset patch
# User Jayashree
# Date 1515990226 -19800
# Mon Jan 15 09:53:46 2018 +0530
# Node ID c1daa99a8c14edbe5e9e5a59a74a6b0936c27a82
# Parent e4983d90f403d968d6760ae044f86a7a2e1865a2
x86:AVX512 intra_pred_ang32 mode 5 and 31 high bit depth
Mode | AVX2 performance | AVX512 performance
---------------------------------------------------
5 | 9.5x | 17.11x
31 | 11.5x | 24.1x
diff -r e4983d90f403 -r c1daa99a8c14 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp Fri Jan 12 15:17:56 2018 +0530
+++ b/source/common/x86/asm-primitives.cpp Mon Jan 15 09:53:46 2018 +0530
@@ -3105,11 +3105,13 @@
p.cu[BLOCK_32x32].intra_pred[25] = PFX(intra_pred_ang32_25_avx512);
p.cu[BLOCK_32x32].intra_pred[26] = PFX(intra_pred_ang32_26_avx512);
p.cu[BLOCK_32x32].intra_pred[27] = PFX(intra_pred_ang32_27_avx512);
-
+ p.cu[BLOCK_32x32].intra_pred[5] = PFX(intra_pred_ang32_5_avx512);
+ p.cu[BLOCK_32x32].intra_pred[31] = PFX(intra_pred_ang32_31_avx512);
p.cu[BLOCK_16x16].intra_pred[9] = PFX(intra_pred_ang16_9_avx512);
p.cu[BLOCK_16x16].intra_pred[11] = PFX(intra_pred_ang16_11_avx512);
p.cu[BLOCK_16x16].intra_pred[25] = PFX(intra_pred_ang16_25_avx512);
p.cu[BLOCK_16x16].intra_pred[27] = PFX(intra_pred_ang16_27_avx512);
+
p.pu[LUMA_64x64].luma_hvpp = interp_8tap_hv_pp_cpu<LUMA_64x64>;
p.pu[LUMA_64x48].luma_hvpp = interp_8tap_hv_pp_cpu<LUMA_64x48>;
p.pu[LUMA_64x32].luma_hvpp = interp_8tap_hv_pp_cpu<LUMA_64x32>;
diff -r e4983d90f403 -r c1daa99a8c14 source/common/x86/intrapred16.asm
--- a/source/common/x86/intrapred16.asm Fri Jan 12 15:17:56 2018 +0530
+++ b/source/common/x86/intrapred16.asm Mon Jan 15 09:53:46 2018 +0530
@@ -18977,6 +18977,316 @@
lea r4, [r1 * 3]
call ang16_mode_11_25
RET
+cglobal ang16_mode_5_31
+ test r6d, r6d
+
+ vbroadcasti32x8 m0, [r2 + 2] ; [16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1]
+ vbroadcasti32x8 m1, [r2 + 4] ; [17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2]
+
+ punpcklwd m3, m0, m1 ; [13 12 12 11 11 10 10 9 5 4 4 3 3 2 2 1]
+ punpckhwd m0, m1 ; [17 16 16 15 15 14 14 13 9 8 8 7 7 6 6 5]
+
+ vbroadcasti32x8 m1, [r2 + 18] ; [24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9]
+ vbroadcasti32x8 m4, [r2 + 20] ; [25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10]
+ punpcklwd m2, m1, m4 ; [21 20 20 19 19 18 18 17 13 12 12 11 11 10 10 9]
+ punpckhwd m1, m4 ; [25 24 24 23 23 22 22 21 17 16 16 15 15 14 14 13]
+
+ pmaddwd m4, m3, [r3 + 1 * 32] ; [17]
+ paddd m4, m15
+ psrld m4, 5
+ pmaddwd m5, m0, [r3 + 1 * 32]
+ paddd m5, m15
+ psrld m5, 5
+ packusdw m4, m5
+
+ movu ym16, [r3 - 14 * 32] ; [2]
+ vinserti32x8 m16, [r3 + 3 * 32] ,1 ; [19]
+ palignr m6, m0, m3, 4
+ pmaddwd m5, m6, m16
+ paddd m5, m15
+ psrld m5, 5
+ palignr m7, m2, m0, 4
+ pmaddwd m8, m7, m16
+ paddd m8, m15
+ psrld m8, 5
+ packusdw m5, m8
+ vextracti32x8 ym6, m5, 1
+
+ palignr m8, m0, m3, 8
+ palignr m9, m2, m0, 8
+ movu ym16, [r3 - 12 * 32] ; [4]
+ vinserti32x8 m16, [r3 + 5 * 32],1 ; [21]
+ pmaddwd m7, m8, m16
+ paddd m7, m15
+ psrld m7, 5
+ pmaddwd m10, m9,m16
+ paddd m10, m15
+ psrld m10, 5
+ packusdw m7, m10
+ vextracti32x8 ym8, m7, 1
+
+ palignr m10, m0, m3, 12
+ palignr m11, m2, m0, 12
+ movu ym16,[r3 - 10 * 32] ; [6]
+ vinserti32x8 m16, [r3 + 7 * 32] ,1 ; [23]
+ pmaddwd m9, m10, m16
+ paddd m9, m15
+ psrld m9, 5
+ pmaddwd m3, m11, m16
+ paddd m3, m15
+ psrld m3, 5
+ packusdw m9, m3
+ vextracti32x8 ym10, m9, 1
+
+ pmaddwd m11, m0, [r3 - 8 * 32] ; [8]
+ paddd m11, m15
+ psrld m11, 5
+ pmaddwd m3, m2, [r3 - 8 * 32]
+ paddd m3, m15
+ psrld m3, 5
+ packusdw m11, m3
+
+ TRANSPOSE_STORE_AVX2 4, 5, 6, 7, 8, 9, 10, 11, 12, 3, 0
+
+ pmaddwd m4, m0, [r3 + 9 * 32] ; [25]
+ paddd m4, m15
+ psrld m4, 5
+ pmaddwd m5, m2, [r3 + 9 * 32]
+ paddd m5, m15
+ psrld m5, 5
+ packusdw m4, m5
+
+ palignr m6, m2, m0, 4
+ movu ym16, [r3 - 6 * 32] ; [10]
+ vinserti32x8 m16, [r3 + 11 * 32] ,1 ; [27]
+ pmaddwd m5, m6,m16
+ paddd m5, m15
+ psrld m5, 5
+ palignr m7, m1, m2, 4
+ pmaddwd m3, m7,m16
+ paddd m3, m15
+ psrld m3, 5
+ packusdw m5, m3
+ vextracti32x8 ym6, m5, 1
+
+ palignr m8, m2, m0, 8
+ palignr m9, m1, m2, 8
+ movu ym16, [r3 - 4 * 32] ; [12]
+ vinserti32x8 m16, [r3 + 13 * 32] ,1 ; [29]
+ pmaddwd m7, m8, m16
+ paddd m7, m15
+ psrld m7, 5
+ pmaddwd m3, m9, m16
+ paddd m3, m15
+ psrld m3, 5
+ packusdw m7, m3
+ vextracti32x8 ym8, m7, 1
+
+
+ palignr m10, m2, m0, 12
+ palignr m11, m1, m2, 12
+ movu ym16, [r3 - 2 * 32] ; [14]
+ vinserti32x8 m16, [r3 + 15 * 32],1 ; [31]
+ pmaddwd m9, m10, m16
+ paddd m9, m15
+ psrld m9, 5
+ pmaddwd m3, m11, m16
+ paddd m3, m15
+ psrld m3, 5
+ packusdw m9, m3
+ vextracti32x8 ym10, m9, 1
+
+ pmaddwd m2, [r3] ; [16]
+ paddd m2, m15
+ psrld m2, 5
+ pmaddwd m1, [r3]
+ paddd m1, m15
+ psrld m1, 5
+ packusdw m2, m1
+ TRANSPOSE_STORE_AVX2 4, 5, 6, 7, 8, 9, 10, 2, 0, 1, 16
+ ret
+;; angle 32, modes 5 and 31
+cglobal ang32_mode_5_31
+ test r6d, r6d
+
+ vbroadcasti32x8 m0, [r2 + 2] ; [16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1]
+ vbroadcasti32x8 m1, [r2 + 4] ; [17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2]
+
+ punpcklwd m3, m0, m1 ; [13 12 12 11 11 10 10 9 5 4 4 3 3 2 2 1]
+ punpckhwd m0, m1 ; [17 16 16 15 15 14 14 13 9 8 8 7 7 6 6 5]
+
+ vbroadcasti32x8 m1, [r2 + 18] ; [24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9]
+ vbroadcasti32x8 m4, [r2 + 20] ; [25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10]
+ punpcklwd m2, m1, m4 ; [21 20 20 19 19 18 18 17 13 12 12 11 11 10 10 9]
+ punpckhwd m1, m4 ; [25 24 24 23 23 22 22 21 17 16 16 15 15 14 14 13]
+
+ movu ym16, [r3 - 15 * 32] ; [1]
+ vinserti32x8 m16, [r3 + 2 * 32],1 ; [18]
+ pmaddwd m4, m3, m16
+ paddd m4, m15
+ psrld m4, 5
+ pmaddwd m5, m0, m16
+ paddd m5, m15
+ psrld m5, 5
+ packusdw m4, m5
+ vextracti32x8 ym5, m4, 1
+
+
+ palignr m7, m0, m3, 4
+ movu ym16, [r3 - 13 * 32] ; [3]
+ vinserti32x8 m16, [r3 + 4 * 32] ,1 ; [20]
+ pmaddwd m6, m7, m16
+ paddd m6, m15
+ psrld m6, 5
+ palignr m8, m2, m0, 4
+ pmaddwd m9, m8,m16
+ paddd m9, m15
+ psrld m9, 5
+ packusdw m6, m9
+ vextracti32x8 ym7, m6, 1
+
+
+ palignr m9, m0, m3, 8
+ movu ym16, [r3 - 11 * 32] ; [5]
+ vinserti32x8 m16, [r3 + 6 * 32] ,1 ; [22]
+ pmaddwd m8, m9,m16
+ paddd m8, m15
+ psrld m8, 5
+ palignr m10, m2, m0, 8
+ pmaddwd m11, m10,m16
+ paddd m11, m15
+ psrld m11, 5
+ packusdw m8, m11
+ vextracti32x8 ym9, m8, 1
+
+
+ palignr m11, m0, m3, 12
+ movu ym16, [r3 - 9 * 32] ; [7]
+ vinserti32x8 m16, [r3 + 8 * 32] ,1 ; [24]
+ pmaddwd m10, m11,m16
+ paddd m10, m15
+ psrld m10, 5
+ palignr m12, m2, m0, 12
+ pmaddwd m3, m12, m16
+ paddd m3, m15
+ psrld m3, 5
+ packusdw m10, m3
+ vextracti32x8 ym11, m10, 1
+
+
+
+ TRANSPOSE_STORE_AVX2 4, 5, 6, 7, 8, 9, 10, 11, 12, 3, 0
+
+ movu ym16, [r3 - 7 * 32] ; [9]
+ vinserti32x8 m16, [r3 + 10 * 32] ,1 ; [26]
+ pmaddwd m4, m0, m16
+ paddd m4, m15
+ psrld m4, 5
+ pmaddwd m5, m2, m16
+ paddd m5, m15
+ psrld m5, 5
+ packusdw m4, m5
+ vextracti32x8 ym5, m4, 1
+
+
+ palignr m7, m2, m0, 4
+ movu ym16, [r3 - 5 * 32] ; [11]
+ vinserti32x8 m16, [r3 + 12 * 32],1 ; [28]
+ pmaddwd m6, m7, m16
+ paddd m6, m15
+ psrld m6, 5
+ palignr m8, m1, m2, 4
+ pmaddwd m9, m8,m16
+ paddd m9, m15
+ psrld m9, 5
+ packusdw m6, m9
+ vextracti32x8 ym7, m6, 1
+
+ palignr m9, m2, m0, 8
+ movu ym16, [r3 - 3 * 32] ; [13]
+ vinserti32x8 m16, [r3 + 14 * 32] ,1 ; [30]
+ pmaddwd m8, m9, m16
+ paddd m8, m15
+ psrld m8, 5
+ palignr m3, m1, m2, 8
+ pmaddwd m10, m3, m16
+ paddd m10, m15
+ psrld m10, 5
+ packusdw m8, m10
+ vextracti32x8 ym9, m8, 1
+
+
+
+ palignr m10, m2, m0, 12
+ pmaddwd m10, [r3 - 1 * 32] ; [15]
+ paddd m10, m15
+ psrld m10, 5
+ palignr m11, m1, m2, 12
+ pmaddwd m11, [r3 - 1 * 32]
+ paddd m11, m15
+ psrld m11, 5
+ packusdw m10, m11
+
+ pmaddwd m2, [r3 - 16 * 32] ; [0]
+ paddd m2, m15
+ psrld m2, 5
+ pmaddwd m1, [r3 - 16 * 32]
+ paddd m1, m15
+ psrld m1, 5
+ packusdw m2, m1
+ TRANSPOSE_STORE_AVX2 4, 5, 6, 7, 8, 9, 10, 2, 0, 1, 16
+ ret
+
+cglobal intra_pred_ang32_5, 3,8,13
+ add r2, 128
+ xor r6d, r6d
+ lea r3, [ang_table_avx2 + 16 * 32]
+ add r1d, r1d
+ lea r4, [r1 * 3]
+ lea r7, [r0 + 8 * r1]
+ vbroadcasti32x8 m15, [pd_16]
+ call ang16_mode_5_31
+
+ add r2, 18
+ lea r0, [r0 + 32]
+
+ call ang32_mode_5_31
+
+ add r2, 14
+ lea r0, [r7 + 8 * r1]
+
+ call ang16_mode_5_31
+ vbroadcasti32x8 m15, [pd_16]
+ add r2, 18
+ lea r0, [r0 + 32]
+
+ call ang32_mode_5_31
+ RET
+
+cglobal intra_pred_ang32_31, 3,7,13
+ xor r6d, r6d
+ inc r6d
+ lea r3, [ang_table_avx2 + 16 * 32]
+ add r1d, r1d
+ lea r4, [r1 * 3]
+ lea r5, [r0 + 32]
+ vbroadcasti32x8 m15, [pd_16]
+ call ang16_mode_5_31
+
+ add r2, 18
+
+ call ang32_mode_5_31
+
+ add r2, 14
+ mov r0, r5
+
+ call ang16_mode_5_31
+
+ add r2, 18
+
+ call ang32_mode_5_31
+ RET
+
;-------------------------------------------------------------------------------------------------------
; avx512 code for intra_pred_ang32 mode 2 to 34 end
;-------------------------------------------------------------------------------------------------------
More information about the x265-devel
mailing list