From: "Logan.Lyu" <Logan.Lyu@myais.com.cn>
To: "Martin Storsjö" <martin@martin.st>,
"FFmpeg development discussions and patches"
<ffmpeg-devel@ffmpeg.org>
Subject: Re: [FFmpeg-devel] [PATCH 2/5] lavc/aarch64: new optimization for 8-bit hevc_epel_uni_w_h
Date: Sun, 18 Jun 2023 16:21:21 +0800
Message-ID: <d588cdb0-6883-17cc-a832-d99a7e9ad59c@myais.com.cn> (raw)
In-Reply-To: <4dd1f09f-188b-d9ff-c8ad-4950bab5b661@martin.st>
[-- Attachment #1: Type: text/plain, Size: 4504 bytes --]
Hi, Martin,
I modified it according to your comments. Please review again.
And here are the checkasm benchmark results of the related functions:
put_hevc_epel_uni_w_h4_8_c: 126.1
put_hevc_epel_uni_w_h4_8_i8mm: 41.6
put_hevc_epel_uni_w_h6_8_c: 222.9
put_hevc_epel_uni_w_h6_8_i8mm: 91.4
put_hevc_epel_uni_w_h8_8_c: 374.4
put_hevc_epel_uni_w_h8_8_i8mm: 102.1
put_hevc_epel_uni_w_h12_8_c: 806.1
put_hevc_epel_uni_w_h12_8_i8mm: 225.6
put_hevc_epel_uni_w_h16_8_c: 1414.4
put_hevc_epel_uni_w_h16_8_i8mm: 333.4
put_hevc_epel_uni_w_h24_8_c: 3128.6
put_hevc_epel_uni_w_h24_8_i8mm: 713.1
put_hevc_epel_uni_w_h32_8_c: 5519.1
put_hevc_epel_uni_w_h32_8_i8mm: 1118.1
put_hevc_epel_uni_w_h48_8_c: 12364.4
put_hevc_epel_uni_w_h48_8_i8mm: 2541.1
put_hevc_epel_uni_w_h64_8_c: 21925.9
put_hevc_epel_uni_w_h64_8_i8mm: 4383.6
在 2023/6/12 15:59, Martin Storsjö 写道:
> On Sun, 4 Jun 2023, Logan.Lyu@myais.com.cn wrote:
>
>> From: Logan Lyu <Logan.Lyu@myais.com.cn>
>>
>> Signed-off-by: Logan Lyu <Logan.Lyu@myais.com.cn>
>> ---
>> libavcodec/aarch64/Makefile | 1 +
>> libavcodec/aarch64/hevcdsp_epel_neon.S | 378 ++++++++++++++++++++++
>> libavcodec/aarch64/hevcdsp_init_aarch64.c | 7 +-
>> 3 files changed, 385 insertions(+), 1 deletion(-)
>> create mode 100644 libavcodec/aarch64/hevcdsp_epel_neon.S
>>
>> diff --git a/libavcodec/aarch64/Makefile b/libavcodec/aarch64/Makefile
>> index 216191640c..cb428b49e0 100644
>> --- a/libavcodec/aarch64/Makefile
>> +++ b/libavcodec/aarch64/Makefile
>> @@ -69,4 +69,5 @@ NEON-OBJS-$(CONFIG_HEVC_DECODER) +=
>> aarch64/hevcdsp_deblock_neon.o \
>> aarch64/hevcdsp_idct_neon.o \
>> aarch64/hevcdsp_init_aarch64.o \
>> aarch64/hevcdsp_qpel_neon.o \
>> + aarch64/hevcdsp_epel_neon.o \
>> aarch64/hevcdsp_sao_neon.o
>> diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S
>> b/libavcodec/aarch64/hevcdsp_epel_neon.S
>> new file mode 100644
>> index 0000000000..fe494dd843
>> --- /dev/null
>> +++ b/libavcodec/aarch64/hevcdsp_epel_neon.S
>> @@ -0,0 +1,378 @@
>> +/* -*-arm64-*-
>> + * vim: syntax=arm64asm
>> + *
>> + * This file is part of FFmpeg.
>> + *
>> + * FFmpeg is free software; you can redistribute it and/or
>> + * modify it under the terms of the GNU Lesser General Public
>> + * License as published by the Free Software Foundation; either
>> + * version 2.1 of the License, or (at your option) any later version.
>> + *
>> + * FFmpeg is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
>> + * Lesser General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU Lesser General Public
>> + * License along with FFmpeg; if not, write to the Free Software
>> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
>> 02110-1301 USA
>> + */
>> +
>> +#include "libavutil/aarch64/asm.S"
>> +#define MAX_PB_SIZE 64
>> +
>> +const epel_filters, align=4
>> + .byte 0, 0, 0, 0
>> + .byte -2, 58, 10, -2
>> + .byte -4, 54, 16, -2
>> + .byte -6, 46, 28, -4
>> + .byte -4, 36, 36, -4
>> + .byte -4, 28, 46, -6
>> + .byte -2, 16, 54, -4
>> + .byte -2, 10, 58, -2
>> +endconst
>> +
>> +#if HAVE_I8MM
>> +.macro EPEL_UNI_W_H_HEADER
>> + ldr x12, [sp]
>> + sub x2, x2, #1
>> + movrel x9, epel_filters
>> + add x9, x9, x12, lsl #2
>> + ldr w11, [x9]
>> + dup v28.4s, w11
>
> Why not just do "ld1r {v28.4s}, [x9]" here instead, avoiding the
> indirection via GPRs?
>
> Other than that, I think this mostly looks reasonable.
>
> Btw, for any assembly patches like these, it would be appreciated if
> you can provide benchmarks from checkasm, e.g. "checkasm
> --test=hevc_pel --bench=put_hevc" (or maybe just "--bench") and
> extract the relevant lines for the functions that you've
> added/modified, and mention what system you've benchmarked it on. You
> get the most useful benchmarks for micro-tuning if you can enable
> userspace access to the timing registers and configure with
> --disable-linux-perf.
>
> // Martin
>
[-- Attachment #2: 0002-lavc-aarch64-new-optimization-for-8-bit-hevc_epel_uni_w_h.patch --]
[-- Type: text/plain, Size: 17409 bytes --]
From 9985cbcc0aa402d9920dd690b6f6a71392d62f79 Mon Sep 17 00:00:00 2001
From: Logan Lyu <Logan.Lyu@myais.com.cn>
Date: Sun, 28 May 2023 10:07:28 +0800
Subject: [PATCH 2/5] lavc/aarch64: new optimization for 8-bit
hevc_epel_uni_w_h
Signed-off-by: Logan Lyu <Logan.Lyu@myais.com.cn>
---
libavcodec/aarch64/Makefile | 1 +
libavcodec/aarch64/hevcdsp_epel_neon.S | 377 ++++++++++++++++++++++
libavcodec/aarch64/hevcdsp_init_aarch64.c | 7 +-
3 files changed, 384 insertions(+), 1 deletion(-)
create mode 100644 libavcodec/aarch64/hevcdsp_epel_neon.S
diff --git a/libavcodec/aarch64/Makefile b/libavcodec/aarch64/Makefile
index 216191640c..cb428b49e0 100644
--- a/libavcodec/aarch64/Makefile
+++ b/libavcodec/aarch64/Makefile
@@ -69,4 +69,5 @@ NEON-OBJS-$(CONFIG_HEVC_DECODER) += aarch64/hevcdsp_deblock_neon.o \
aarch64/hevcdsp_idct_neon.o \
aarch64/hevcdsp_init_aarch64.o \
aarch64/hevcdsp_qpel_neon.o \
+ aarch64/hevcdsp_epel_neon.o \
aarch64/hevcdsp_sao_neon.o
diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S b/libavcodec/aarch64/hevcdsp_epel_neon.S
new file mode 100644
index 0000000000..0411de9864
--- /dev/null
+++ b/libavcodec/aarch64/hevcdsp_epel_neon.S
@@ -0,0 +1,377 @@
+/* -*-arm64-*-
+ * vim: syntax=arm64asm
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/aarch64/asm.S"
+#define MAX_PB_SIZE 64
+
+const epel_filters, align=4
+ .byte 0, 0, 0, 0
+ .byte -2, 58, 10, -2
+ .byte -4, 54, 16, -2
+ .byte -6, 46, 28, -4
+ .byte -4, 36, 36, -4
+ .byte -4, 28, 46, -6
+ .byte -2, 16, 54, -4
+ .byte -2, 10, 58, -2
+endconst
+
+#if HAVE_I8MM
+.macro EPEL_UNI_W_H_HEADER
+ ldr x12, [sp]
+ sub x2, x2, #1
+ movrel x9, epel_filters
+ add x9, x9, x12, lsl #2
+ ld1r {v28.4s}, [x9]
+ mov w10, #-6
+ sub w10, w10, w5
+ dup v30.4s, w6
+ dup v31.4s, w10
+ dup v29.4s, w7
+.endm
+
+
+function ff_hevc_put_hevc_epel_uni_w_h4_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+1:
+ ld1 {v0.8b}, [x2], x3
+ subs w4, w4, #1
+ ext v1.8b, v0.8b, v0.8b, #1
+ ext v2.8b, v0.8b, v0.8b, #2
+ ext v3.8b, v0.8b, v0.8b, #3
+ trn1 v0.2s, v0.2s, v2.2s
+ trn1 v1.2s, v1.2s, v3.2s
+ zip1 v0.4s, v0.4s, v1.4s
+ movi v16.2d, #0
+ usdot v16.4s, v0.16b, v28.16b
+ mul v16.4s, v16.4s, v30.4s
+ sqrshl v16.4s, v16.4s, v31.4s
+ sqadd v16.4s, v16.4s, v29.4s
+ sqxtn v16.4h, v16.4s
+ sqxtun v16.8b, v16.8h
+ str s16, [x0]
+ add x0, x0, x1
+ b.hi 1b
+ ret
+endfunc
+
+
+function ff_hevc_put_hevc_epel_uni_w_h6_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+ sub x1, x1, #4
+1:
+ ld1 {v0.16b}, [x2], x3
+ subs w4, w4, #1
+ ext v1.16b, v0.16b, v0.16b, #1
+ ext v2.16b, v0.16b, v0.16b, #2
+ ext v3.16b, v0.16b, v0.16b, #3
+ trn1 v4.2s, v0.2s, v1.2s
+ trn2 v6.2s, v0.2s, v1.2s
+ trn1 v5.2s, v2.2s, v3.2s
+ zip1 v4.2d, v4.2d, v5.2d
+ movi v16.2d, #0
+ movi v17.2d, #0
+ usdot v16.4s, v4.16b, v28.16b
+ usdot v17.2s, v6.8b, v28.8b
+ mul v16.4s, v16.4s, v30.4s
+ mul v17.2s, v17.2s, v30.2s
+ sqrshl v16.4s, v16.4s, v31.4s
+ sqrshl v17.2s, v17.2s, v31.2s
+ sqadd v16.4s, v16.4s, v29.4s
+ sqadd v17.2s, v17.2s, v29.2s
+ sqxtn v16.4h, v16.4s
+ sqxtn2 v16.8h, v17.4s
+ sqxtun v16.8b, v16.8h
+ str s16, [x0], #4
+ st1 {v16.h}[2], [x0], x1
+ b.hi 1b
+ ret
+endfunc
+
+.macro EPEL_UNI_W_H_CALC s0, s1, d0, d1
+ movi \d0\().2d, #0
+ movi \d1\().2d, #0
+ usdot \d0\().4s, \s0\().16b, v28.16b
+ usdot \d1\().4s, \s1\().16b, v28.16b
+ mul \d0\().4s, \d0\().4s, v30.4s
+ mul \d1\().4s, \d1\().4s, v30.4s
+ sqrshl \d0\().4s, \d0\().4s, v31.4s
+ sqrshl \d1\().4s, \d1\().4s, v31.4s
+ sqadd \d0\().4s, \d0\().4s, v29.4s
+ sqadd \d1\().4s, \d1\().4s, v29.4s
+.endm
+
+function ff_hevc_put_hevc_epel_uni_w_h8_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+1:
+ ld1 {v0.16b}, [x2], x3
+ subs w4, w4, #1
+ ext v1.16b, v0.16b, v0.16b, #1
+ ext v2.16b, v0.16b, v0.16b, #2
+ ext v3.16b, v0.16b, v0.16b, #3
+ zip1 v4.4s, v0.4s, v2.4s
+ zip1 v5.4s, v1.4s, v3.4s
+ EPEL_UNI_W_H_CALC v4, v5, v16, v17
+ sqxtn v16.4h, v16.4s
+ sqxtn v17.4h, v17.4s
+ zip1 v16.8h, v16.8h, v17.8h
+ sqxtun v16.8b, v16.8h
+ str d16, [x0]
+ add x0, x0, x1
+ b.hi 1b
+ ret
+endfunc
+
+function ff_hevc_put_hevc_epel_uni_w_h12_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+1:
+ ld1 {v0.16b}, [x2], x3
+ subs w4, w4, #1
+ ext v1.16b, v0.16b, v0.16b, #1
+ ext v2.16b, v0.16b, v0.16b, #2
+ ext v3.16b, v0.16b, v0.16b, #3
+ zip1 v4.4s, v0.4s, v2.4s
+ zip1 v5.4s, v1.4s, v3.4s
+ zip2 v6.4s, v0.4s, v2.4s
+ zip2 v7.4s, v1.4s, v3.4s
+ zip1 v6.4s, v6.4s, v7.4s
+ EPEL_UNI_W_H_CALC v4, v5, v16, v17
+ movi v18.2d, #0
+ usdot v18.4s, v6.16b, v28.16b
+ mul v18.4s, v18.4s, v30.4s
+ sqrshl v18.4s, v18.4s, v31.4s
+ sqadd v18.4s, v18.4s, v29.4s
+ sqxtn v16.4h, v16.4s
+ sqxtn v17.4h, v17.4s
+ sqxtn v18.4h, v18.4s
+ zip1 v16.8h, v16.8h, v17.8h
+ sqxtun v16.8b, v16.8h
+ sqxtun v18.8b, v18.8h
+ str d16, [x0]
+ str s18, [x0, #8]
+ add x0, x0, x1
+ b.hi 1b
+ ret
+endfunc
+
+function ff_hevc_put_hevc_epel_uni_w_h16_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+1:
+ ld1 {v0.16b, v1.16b}, [x2], x3
+ subs w4, w4, #1
+ ext v4.16b, v0.16b, v1.16b, #1
+ ext v5.16b, v0.16b, v1.16b, #2
+ ext v6.16b, v0.16b, v1.16b, #3
+ zip1 v20.4s, v0.4s, v5.4s
+ zip1 v21.4s, v4.4s, v6.4s
+ zip2 v22.4s, v0.4s, v5.4s
+ zip2 v23.4s, v4.4s, v6.4s
+ EPEL_UNI_W_H_CALC v20, v21, v16, v17
+ EPEL_UNI_W_H_CALC v22, v23, v18, v19
+ sqxtn v16.4h, v16.4s
+ sqxtn v17.4h, v17.4s
+ sqxtn2 v16.8h, v18.4s
+ sqxtn2 v17.8h, v19.4s
+ sqxtun v16.8b, v16.8h
+ sqxtun v17.8b, v17.8h
+ st2 {v16.8b, v17.8b}, [x0], x1
+ b.hi 1b
+ ret
+endfunc
+
+function ff_hevc_put_hevc_epel_uni_w_h24_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+1:
+ ld1 {v0.16b, v1.16b}, [x2], x3
+ subs w4, w4, #1
+ ext v2.16b, v0.16b, v1.16b, #1
+ ext v3.16b, v0.16b, v1.16b, #2
+ ext v4.16b, v0.16b, v1.16b, #3
+ ext v5.16b, v1.16b, v1.16b, #1
+ ext v6.16b, v1.16b, v1.16b, #2
+ ext v7.16b, v1.16b, v1.16b, #3
+ zip1 v20.4s, v0.4s, v3.4s
+ zip1 v21.4s, v2.4s, v4.4s
+ zip2 v22.4s, v0.4s, v3.4s
+ zip2 v23.4s, v2.4s, v4.4s
+ zip1 v24.4s, v1.4s, v6.4s
+ zip1 v25.4s, v5.4s, v7.4s
+ EPEL_UNI_W_H_CALC v20, v21, v16, v17
+ EPEL_UNI_W_H_CALC v22, v23, v18, v19
+ EPEL_UNI_W_H_CALC v24, v25, v26, v27
+ sqxtn v16.4h, v16.4s
+ sqxtn v17.4h, v17.4s
+ sqxtn v18.4h, v18.4s
+ sqxtn v19.4h, v19.4s
+ sqxtn v26.4h, v26.4s
+ sqxtn v27.4h, v27.4s
+ zip1 v16.8h, v16.8h, v17.8h
+ zip1 v18.8h, v18.8h, v19.8h
+ zip1 v26.8h, v26.8h, v27.8h
+ sqxtun v16.8b, v16.8h
+ sqxtun2 v16.16b, v18.8h
+ sqxtun v26.8b, v26.8h
+ str q16, [x0]
+ str d26, [x0, #16]
+ add x0, x0, x1
+ b.hi 1b
+ ret
+endfunc
+
+function ff_hevc_put_hevc_epel_uni_w_h32_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+1:
+ ld1 {v0.16b, v1.16b, v2.16b}, [x2], x3
+ subs w4, w4, #1
+ ext v3.16b, v0.16b, v1.16b, #1
+ ext v4.16b, v0.16b, v1.16b, #2
+ ext v5.16b, v0.16b, v1.16b, #3
+ ext v16.16b, v1.16b, v2.16b, #1
+ ext v17.16b, v1.16b, v2.16b, #2
+ ext v18.16b, v1.16b, v2.16b, #3
+ EPEL_UNI_W_H_CALC v0, v3, v6, v7
+ EPEL_UNI_W_H_CALC v4, v5, v19, v20
+ EPEL_UNI_W_H_CALC v1, v16, v21, v22
+ EPEL_UNI_W_H_CALC v17, v18, v23, v24
+ sqxtn v6.4h, v6.4s
+ sqxtn2 v6.8h, v21.4s
+ sqxtn v7.4h, v7.4s
+ sqxtn2 v7.8h, v22.4s
+ sqxtn v19.4h, v19.4s
+ sqxtn2 v19.8h, v23.4s
+ sqxtn v20.4h, v20.4s
+ sqxtn2 v20.8h, v24.4s
+ sqxtun v0.8b, v6.8h
+ sqxtun v1.8b, v7.8h
+ sqxtun v2.8b, v19.8h
+ sqxtun v3.8b, v20.8h
+ st4 {v0.8b, v1.8b, v2.8b, v3.8b}, [x0], x1
+ b.hi 1b
+ ret
+endfunc
+
+
+
+function ff_hevc_put_hevc_epel_uni_w_h48_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+ sub x1, x1, #32
+1:
+ ld1 {v0.16b, v1.16b, v2.16b, v3.16b}, [x2], x3
+ subs w4, w4, #1
+ ext v4.16b, v0.16b, v1.16b, #1
+ ext v5.16b, v0.16b, v1.16b, #2
+ ext v6.16b, v0.16b, v1.16b, #3
+ ext v16.16b, v1.16b, v2.16b, #1
+ ext v17.16b, v1.16b, v2.16b, #2
+ ext v18.16b, v1.16b, v2.16b, #3
+ EPEL_UNI_W_H_CALC v0, v4, v19, v20
+ EPEL_UNI_W_H_CALC v5, v6, v21, v22
+ EPEL_UNI_W_H_CALC v1, v16, v23, v24
+ EPEL_UNI_W_H_CALC v17, v18, v25, v26
+ sqxtn v19.4h, v19.4s
+ sqxtn2 v19.8h, v23.4s
+ sqxtn v20.4h, v20.4s
+ sqxtn2 v20.8h, v24.4s
+ sqxtn v21.4h, v21.4s
+ sqxtn2 v21.8h, v25.4s
+ sqxtn v22.4h, v22.4s
+ sqxtn2 v22.8h, v26.4s
+ sqxtun v19.8b, v19.8h
+ sqxtun v20.8b, v20.8h
+ sqxtun v21.8b, v21.8h
+ sqxtun v22.8b, v22.8h
+ st4 {v19.8b, v20.8b, v21.8b, v22.8b}, [x0], #32
+ ext v5.16b, v2.16b, v3.16b, #1
+ ext v6.16b, v2.16b, v3.16b, #2
+ ext v7.16b, v2.16b, v3.16b, #3
+ EPEL_UNI_W_H_CALC v2, v5, v19, v20
+ EPEL_UNI_W_H_CALC v6, v7, v21, v22
+ sqxtn v19.4h, v19.4s
+ sqxtn v20.4h, v20.4s
+ sqxtn v21.4h, v21.4s
+ sqxtn v22.4h, v22.4s
+ zip1 v4.8h, v19.8h, v21.8h
+ zip1 v5.8h, v20.8h, v22.8h
+ sqxtun v4.8b, v4.8h
+ sqxtun v5.8b, v5.8h
+ st2 {v4.8b, v5.8b}, [x0], x1
+ b.hi 1b
+ ret
+endfunc
+
+
+function ff_hevc_put_hevc_epel_uni_w_h64_8_neon_i8mm, export=1
+ EPEL_UNI_W_H_HEADER
+ sub x1, x1, #32
+ sub x3, x3, #64
+1:
+ ld1 {v0.16b, v1.16b, v2.16b, v3.16b}, [x2], #64
+ subs w4, w4, #1
+ ext v4.16b, v0.16b, v1.16b, #1
+ ext v5.16b, v0.16b, v1.16b, #2
+ ext v6.16b, v0.16b, v1.16b, #3
+ ext v16.16b, v1.16b, v2.16b, #1
+ ext v17.16b, v1.16b, v2.16b, #2
+ ext v18.16b, v1.16b, v2.16b, #3
+ EPEL_UNI_W_H_CALC v0, v4, v19, v20
+ EPEL_UNI_W_H_CALC v5, v6, v21, v22
+ EPEL_UNI_W_H_CALC v1, v16, v23, v24
+ EPEL_UNI_W_H_CALC v17, v18, v25, v26
+ sqxtn v19.4h, v19.4s
+ sqxtn2 v19.8h, v23.4s
+ sqxtn v20.4h, v20.4s
+ sqxtn2 v20.8h, v24.4s
+ sqxtn v21.4h, v21.4s
+ sqxtn2 v21.8h, v25.4s
+ sqxtn v22.4h, v22.4s
+ sqxtn2 v22.8h, v26.4s
+ sqxtun v19.8b, v19.8h
+ sqxtun v20.8b, v20.8h
+ sqxtun v21.8b, v21.8h
+ sqxtun v22.8b, v22.8h
+ st4 {v19.8b, v20.8b, v21.8b, v22.8b}, [x0], #32
+ ld1 {v7.8b}, [x2], x3
+ ext v4.16b, v2.16b, v3.16b, #1
+ ext v5.16b, v2.16b, v3.16b, #2
+ ext v6.16b, v2.16b, v3.16b, #3
+ ext v16.16b, v3.16b, v7.16b, #1
+ ext v17.16b, v3.16b, v7.16b, #2
+ ext v18.16b, v3.16b, v7.16b, #3
+ EPEL_UNI_W_H_CALC v2, v4, v19, v20
+ EPEL_UNI_W_H_CALC v5, v6, v21, v22
+ EPEL_UNI_W_H_CALC v3, v16, v23, v24
+ EPEL_UNI_W_H_CALC v17, v18, v25, v26
+ sqxtn v19.4h, v19.4s
+ sqxtn2 v19.8h, v23.4s
+ sqxtn v20.4h, v20.4s
+ sqxtn2 v20.8h, v24.4s
+ sqxtn v21.4h, v21.4s
+ sqxtn2 v21.8h, v25.4s
+ sqxtn v22.4h, v22.4s
+ sqxtn2 v22.8h, v26.4s
+ sqxtun v19.8b, v19.8h
+ sqxtun v20.8b, v20.8h
+ sqxtun v21.8b, v21.8h
+ sqxtun v22.8b, v22.8h
+ st4 {v19.8b, v20.8b, v21.8b, v22.8b}, [x0], x1
+ b.hi 1b
+ ret
+endfunc
+
+#endif
diff --git a/libavcodec/aarch64/hevcdsp_init_aarch64.c b/libavcodec/aarch64/hevcdsp_init_aarch64.c
index 5a1d520eec..8af0a2b4b9 100644
--- a/libavcodec/aarch64/hevcdsp_init_aarch64.c
+++ b/libavcodec/aarch64/hevcdsp_init_aarch64.c
@@ -166,6 +166,10 @@ NEON8_FNPROTO_PARTIAL_4(qpel_uni_w_v, (uint8_t *_dst, ptrdiff_t _dststride,
int height, int denom, int wx, int ox,
intptr_t mx, intptr_t my, int width),);
+NEON8_FNPROTO(epel_uni_w_h, (uint8_t *_dst, ptrdiff_t _dststride,
+ const uint8_t *_src, ptrdiff_t _srcstride,
+ int height, int denom, int wx, int ox,
+ intptr_t mx, intptr_t my, int width), _i8mm);
NEON8_FNPROTO(qpel_h, (int16_t *dst,
const uint8_t *_src, ptrdiff_t _srcstride,
@@ -273,8 +277,9 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth)
NEON8_FNASSIGN_PARTIAL_4(c->put_hevc_qpel_uni_w, 1, 0, qpel_uni_w_v,);
if (have_i8mm(cpu_flags)) {
- NEON8_FNASSIGN(c->put_hevc_qpel_uni_w, 0, 1, qpel_uni_w_h, _i8mm);
+ NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 0, 1, epel_uni_w_h ,_i8mm);
NEON8_FNASSIGN(c->put_hevc_qpel, 0, 1, qpel_h, _i8mm);
+ NEON8_FNASSIGN(c->put_hevc_qpel_uni_w, 0, 1, qpel_uni_w_h, _i8mm);
NEON8_FNASSIGN_PARTIAL_5(c->put_hevc_qpel_uni_w, 1, 1, qpel_uni_w_hv, _i8mm);
}
--
2.38.0.windows.1
[-- Attachment #3: Type: text/plain, Size: 251 bytes --]
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
next prev parent reply other threads:[~2023-06-18 8:21 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-04 4:17 [FFmpeg-devel] [PATCH 1/5] lavc/aarch64: new optimization for 8-bit hevc_pel_uni_pixels Logan.Lyu
2023-06-04 4:17 ` [FFmpeg-devel] [PATCH 2/5] lavc/aarch64: new optimization for 8-bit hevc_epel_uni_w_h Logan.Lyu
2023-06-12 7:59 ` Martin Storsjö
2023-06-18 8:21 ` Logan.Lyu [this message]
2023-06-04 4:17 ` [FFmpeg-devel] [PATCH 3/5] lavc/aarch64: new optimization for 8-bit hevc_epel_uni_w_v Logan.Lyu
2023-06-12 8:09 ` Martin Storsjö
2023-06-12 9:08 ` Martin Storsjö
2023-06-18 8:22 ` Logan.Lyu
2023-07-01 21:21 ` Martin Storsjö
2023-06-04 4:17 ` [FFmpeg-devel] [PATCH 4/5] lavc/aarch64: new optimization for 8-bit hevc_epel_h Logan.Lyu
2023-06-12 8:12 ` Martin Storsjö
2023-06-18 8:23 ` Logan.Lyu
2023-06-18 8:26 ` Logan.Lyu
2023-06-04 4:17 ` [FFmpeg-devel] [PATCH 5/5] lavc/aarch64: new optimization for 8-bit hevc_epel_uni_w_hv Logan.Lyu
2023-06-12 8:19 ` Martin Storsjö
2023-06-18 8:25 ` Logan.Lyu
2023-07-01 21:28 ` Martin Storsjö
2023-07-13 14:54 ` Logan.Lyu
2023-07-14 9:28 ` Martin Storsjö
2023-06-12 7:47 ` [FFmpeg-devel] [PATCH 1/5] lavc/aarch64: new optimization for 8-bit hevc_pel_uni_pixels Martin Storsjö
2023-06-18 8:29 ` Logan.Lyu
2023-07-01 21:16 ` Martin Storsjö
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=d588cdb0-6883-17cc-a832-d99a7e9ad59c@myais.com.cn \
--to=logan.lyu@myais.com.cn \
--cc=ffmpeg-devel@ffmpeg.org \
--cc=martin@martin.st \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
This inbox may be cloned and mirrored by anyone:
git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git
# If you have public-inbox 1.1+ installed, you may
# initialize and index your mirror using the following commands:
public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \
ffmpegdev@gitmailbox.com
public-inbox-index ffmpegdev
Example config snippet for mirrors.
AGPL code for this site: git clone https://public-inbox.org/public-inbox.git