From 1e9cfa5bb0253784798029bda377823ab5d1a529 Mon Sep 17 00:00:00 2001 From: Hubert Mazur Date: Fri, 28 Oct 2022 11:34:37 +0000 Subject: [PATCH] sw_scale: Add specializations for hscale 8 to 19 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add arm64 neon implementations for hscale 8 to 19 with filter sizes 4, 4X and 8. Both implementations are based on very similar ones dedicated to hscale 8 to 15. The major changes refer to saving the data - instead of writing the result as int16_t it is done with int32_t. These functions are heavily inspired on patches provided by J. Swinney and M. Storsjö for hscale8to15 which were slightly adapted for hscale8to19. The tests and benchmarks run on AWS Graviton 2 instances. The results from a checkasm tool shown below. hscale_8_to_19__fs_4_dstW_512_c: 5663.2 hscale_8_to_19__fs_4_dstW_512_neon: 1259.7 hscale_8_to_19__fs_8_dstW_512_c: 9306.0 hscale_8_to_19__fs_8_dstW_512_neon: 2020.2 hscale_8_to_19__fs_12_dstW_512_c: 12932.7 hscale_8_to_19__fs_12_dstW_512_neon: 2462.5 hscale_8_to_19__fs_16_dstW_512_c: 16844.2 hscale_8_to_19__fs_16_dstW_512_neon: 4671.2 hscale_8_to_19__fs_32_dstW_512_c: 32803.7 hscale_8_to_19__fs_32_dstW_512_neon: 5474.2 hscale_8_to_19__fs_40_dstW_512_c: 40948.0 hscale_8_to_19__fs_40_dstW_512_neon: 6669.7 Signed-off-by: Hubert Mazur Signed-off-by: Martin Storsjö --- libswscale/aarch64/hscale.S | 291 +++++++++++++++++++++++++++++++++++ libswscale/aarch64/swscale.c | 13 +- 2 files changed, 300 insertions(+), 4 deletions(-) diff --git a/libswscale/aarch64/hscale.S b/libswscale/aarch64/hscale.S index a16d3dca42..35e25462ce 100644 --- a/libswscale/aarch64/hscale.S +++ b/libswscale/aarch64/hscale.S @@ -347,3 +347,294 @@ function ff_hscale8to15_4_neon, export=1 add sp, sp, #32 // clean up stack ret endfunc + +function ff_hscale8to19_4_neon, export=1 + // x0 SwsContext *c (unused) + // x1 int32_t *dst + // w2 int dstW + // x3 const uint8_t *src // treat it as uint16_t *src + // x4 const uint16_t *filter + // x5 const int32_t *filterPos + // w6 int filterSize + + movi v18.4s, #1 + movi v17.4s, #1 + shl v18.4s, v18.4s, #19 + sub v18.4s, v18.4s, v17.4s // max allowed value + + cmp w2, #16 + b.lt 2f // move to last block + + ldp w8, w9, [x5] // filterPos[0], filterPos[1] + ldp w10, w11, [x5, #8] // filterPos[2], filterPos[3] + ldp w12, w13, [x5, #16] // filterPos[4], filterPos[5] + ldp w14, w15, [x5, #24] // filterPos[6], filterPos[7] + add x5, x5, #32 + + // load data from + ldr w8, [x3, w8, UXTW] + ldr w9, [x3, w9, UXTW] + ldr w10, [x3, w10, UXTW] + ldr w11, [x3, w11, UXTW] + ldr w12, [x3, w12, UXTW] + ldr w13, [x3, w13, UXTW] + ldr w14, [x3, w14, UXTW] + ldr w15, [x3, w15, UXTW] + + sub sp, sp, #32 + + stp w8, w9, [sp] + stp w10, w11, [sp, #8] + stp w12, w13, [sp, #16] + stp w14, w15, [sp, #24] + +1: + ld4 {v0.8b, v1.8b, v2.8b, v3.8b}, [sp] + ld4 {v28.8h, v29.8h, v30.8h, v31.8h}, [x4], #64 // filter[0..7] + // load filterPositions into registers for next iteration + + ldp w8, w9, [x5] // filterPos[0], filterPos[1] + ldp w10, w11, [x5, #8] // filterPos[2], filterPos[3] + ldp w12, w13, [x5, #16] // filterPos[4], filterPos[5] + ldp w14, w15, [x5, #24] // filterPos[6], filterPos[7] + add x5, x5, #32 + uxtl v0.8h, v0.8b + ldr w8, [x3, w8, UXTW] + smull v5.4s, v0.4h, v28.4h // multiply first column of src + ldr w9, [x3, w9, UXTW] + smull2 v6.4s, v0.8h, v28.8h + stp w8, w9, [sp] + + uxtl v1.8h, v1.8b + ldr w10, [x3, w10, UXTW] + smlal v5.4s, v1.4h, v29.4h // multiply second column of src + ldr w11, [x3, w11, UXTW] + smlal2 v6.4s, v1.8h, v29.8h + stp w10, w11, [sp, #8] + + uxtl v2.8h, v2.8b + ldr w12, [x3, w12, UXTW] + smlal v5.4s, v2.4h, v30.4h // multiply third column of src + ldr w13, [x3, w13, UXTW] + smlal2 v6.4s, v2.8h, v30.8h + stp w12, w13, [sp, #16] + + uxtl v3.8h, v3.8b + ldr w14, [x3, w14, UXTW] + smlal v5.4s, v3.4h, v31.4h // multiply fourth column of src + ldr w15, [x3, w15, UXTW] + smlal2 v6.4s, v3.8h, v31.8h + stp w14, w15, [sp, #24] + + sub w2, w2, #8 + sshr v5.4s, v5.4s, #3 + sshr v6.4s, v6.4s, #3 + smin v5.4s, v5.4s, v18.4s + smin v6.4s, v6.4s, v18.4s + + st1 {v5.4s, v6.4s}, [x1], #32 + cmp w2, #16 + b.ge 1b + + // here we make last iteration, without updating the registers + ld4 {v0.8b, v1.8b, v2.8b, v3.8b}, [sp] + ld4 {v28.8h, v29.8h, v30.8h, v31.8h}, [x4], #64 // filter[0..7] + + uxtl v0.8h, v0.8b + uxtl v1.8h, v1.8b + smull v5.4s, v0.4h, v28.4h + smull2 v6.4s, v0.8h, v28.8h + uxtl v2.8h, v2.8b + smlal v5.4s, v1.4h, v29.4H + smlal2 v6.4s, v1.8h, v29.8H + uxtl v3.8h, v3.8b + smlal v5.4s, v2.4h, v30.4H + smlal2 v6.4s, v2.8h, v30.8H + smlal v5.4s, v3.4h, v31.4H + smlal2 v6.4s, v3.8h, v31.8h + + sshr v5.4s, v5.4s, #3 + sshr v6.4s, v6.4s, #3 + + smin v5.4s, v5.4s, v18.4s + smin v6.4s, v6.4s, v18.4s + + sub w2, w2, #8 + st1 {v5.4s, v6.4s}, [x1], #32 + add sp, sp, #32 // restore stack + cbnz w2, 2f + + ret + +2: + ldr w8, [x5], #4 // load filterPos + add x9, x3, w8, UXTW // src + filterPos + ld1 {v0.s}[0], [x9] // load 4 * uint8_t* into one single + ld1 {v31.4h}, [x4], #8 + uxtl v0.8h, v0.8b + smull v5.4s, v0.4h, v31.4H + saddlv d0, v5.4S + sqshrn s0, d0, #3 + smin v0.4s, v0.4s, v18.4s + st1 {v0.s}[0], [x1], #4 + sub w2, w2, #1 + cbnz w2, 2b // if iterations remain jump to beginning + + ret +endfunc + +function ff_hscale8to19_X8_neon, export=1 + movi v20.4s, #1 + movi v17.4s, #1 + shl v20.4s, v20.4s, #19 + sub v20.4s, v20.4s, v17.4s + + sbfiz x7, x6, #1, #32 // filterSize*2 (*2 because int16) +1: + mov x16, x4 // filter0 = filter + ldr w8, [x5], #4 // filterPos[idx] + add x12, x16, x7 // filter1 = filter0 + filterSize*2 + ldr w0, [x5], #4 // filterPos[idx + 1] + add x13, x12, x7 // filter2 = filter1 + filterSize*2 + ldr w11, [x5], #4 // filterPos[idx + 2] + add x4, x13, x7 // filter3 = filter2 + filterSize*2 + ldr w9, [x5], #4 // filterPos[idx + 3] + movi v0.2D, #0 // val sum part 1 (for dst[0]) + movi v1.2D, #0 // val sum part 2 (for dst[1]) + movi v2.2D, #0 // val sum part 3 (for dst[2]) + movi v3.2D, #0 // val sum part 4 (for dst[3]) + add x17, x3, w8, UXTW // srcp + filterPos[0] + add x8, x3, w0, UXTW // srcp + filterPos[1] + add x0, x3, w11, UXTW // srcp + filterPos[2] + add x11, x3, w9, UXTW // srcp + filterPos[3] + mov w15, w6 // filterSize counter +2: ld1 {v4.8B}, [x17], #8 // srcp[filterPos[0] + {0..7}] + ld1 {v5.8H}, [x16], #16 // load 8x16-bit filter values, part 1 + uxtl v4.8H, v4.8B // unpack part 1 to 16-bit + smlal v0.4S, v4.4H, v5.4H // v0 accumulates srcp[filterPos[0] + {0..3}] * filter[{0..3}] + ld1 {v6.8B}, [x8], #8 // srcp[filterPos[1] + {0..7}] + smlal2 v0.4S, v4.8H, v5.8H // v0 accumulates srcp[filterPos[0] + {4..7}] * filter[{4..7}] + ld1 {v7.8H}, [x12], #16 // load 8x16-bit at filter+filterSize + ld1 {v16.8B}, [x0], #8 // srcp[filterPos[2] + {0..7}] + uxtl v6.8H, v6.8B // unpack part 2 to 16-bit + ld1 {v17.8H}, [x13], #16 // load 8x16-bit at filter+2*filterSize + uxtl v16.8H, v16.8B // unpack part 3 to 16-bit + smlal v1.4S, v6.4H, v7.4H // v1 accumulates srcp[filterPos[1] + {0..3}] * filter[{0..3}] + ld1 {v18.8B}, [x11], #8 // srcp[filterPos[3] + {0..7}] + smlal v2.4S, v16.4H, v17.4H // v2 accumulates srcp[filterPos[2] + {0..3}] * filter[{0..3}] + ld1 {v19.8H}, [x4], #16 // load 8x16-bit at filter+3*filterSize + smlal2 v2.4S, v16.8H, v17.8H // v2 accumulates srcp[filterPos[2] + {4..7}] * filter[{4..7}] + uxtl v18.8H, v18.8B // unpack part 4 to 16-bit + smlal2 v1.4S, v6.8H, v7.8H // v1 accumulates srcp[filterPos[1] + {4..7}] * filter[{4..7}] + smlal v3.4S, v18.4H, v19.4H // v3 accumulates srcp[filterPos[3] + {0..3}] * filter[{0..3}] + subs w15, w15, #8 // j -= 8: processed 8/filterSize + smlal2 v3.4S, v18.8H, v19.8H // v3 accumulates srcp[filterPos[3] + {4..7}] * filter[{4..7}] + b.gt 2b // inner loop if filterSize not consumed completely + addp v0.4S, v0.4S, v1.4S // part01 horizontal pair adding + addp v2.4S, v2.4S, v3.4S // part23 horizontal pair adding + addp v0.4S, v0.4S, v2.4S // part0123 horizontal pair adding + subs w2, w2, #4 // dstW -= 4 + sshr v0.4s, v0.4S, #3 // shift and clip the 2x16-bit final values + smin v0.4s, v0.4s, v20.4s + st1 {v0.4s}, [x1], #16 // write to destination part0123 + b.gt 1b // loop until end of line + ret +endfunc + +function ff_hscale8to19_X4_neon, export=1 + // x0 SwsContext *c (not used) + // x1 int16_t *dst + // w2 int dstW + // x3 const uint8_t *src + // x4 const int16_t *filter + // x5 const int32_t *filterPos + // w6 int filterSize + + movi v20.4s, #1 + movi v17.4s, #1 + shl v20.4s, v20.4s, #19 + sub v20.4s, v20.4s, v17.4s + + lsl w7, w6, #1 +1: + ldp w8, w9, [x5] + ldp w10, w11, [x5, #8] + + movi v16.2d, #0 // initialize accumulator for idx + 0 + movi v17.2d, #0 // initialize accumulator for idx + 1 + movi v18.2d, #0 // initialize accumulator for idx + 2 + movi v19.2d, #0 // initialize accumulator for idx + 3 + + mov x12, x4 // filter + 0 + add x13, x4, x7 // filter + 1 + add x8, x3, w8, UXTW // srcp + filterPos 0 + add x14, x13, x7 // filter + 2 + add x9, x3, w9, UXTW // srcp + filterPos 1 + add x15, x14, x7 // filter + 3 + add x10, x3, w10, UXTW // srcp + filterPos 2 + mov w0, w6 // save the filterSize to temporary variable + add x11, x3, w11, UXTW // srcp + filterPos 3 + add x5, x5, #16 // advance filter position + mov x16, xzr // clear the register x16 used for offsetting the filter values + +2: + ldr d4, [x8], #8 // load src values for idx 0 + ldr q31, [x12, x16] // load filter values for idx 0 + uxtl v4.8h, v4.8b // extend type to match the filter' size + ldr d5, [x9], #8 // load src values for idx 1 + smlal v16.4s, v4.4h, v31.4h // multiplication of lower half for idx 0 + uxtl v5.8h, v5.8b // extend type to match the filter' size + ldr q30, [x13, x16] // load filter values for idx 1 + smlal2 v16.4s, v4.8h, v31.8h // multiplication of upper half for idx 0 + ldr d6, [x10], #8 // load src values for idx 2 + ldr q29, [x14, x16] // load filter values for idx 2 + smlal v17.4s, v5.4h, v30.4H // multiplication of lower half for idx 1 + ldr d7, [x11], #8 // load src values for idx 3 + smlal2 v17.4s, v5.8h, v30.8H // multiplication of upper half for idx 1 + uxtl v6.8h, v6.8B // extend tpye to matchi the filter's size + ldr q28, [x15, x16] // load filter values for idx 3 + smlal v18.4s, v6.4h, v29.4h // multiplication of lower half for idx 2 + uxtl v7.8h, v7.8B + smlal2 v18.4s, v6.8h, v29.8H // multiplication of upper half for idx 2 + sub w0, w0, #8 + smlal v19.4s, v7.4h, v28.4H // multiplication of lower half for idx 3 + cmp w0, #8 + smlal2 v19.4s, v7.8h, v28.8h // multiplication of upper half for idx 3 + add x16, x16, #16 // advance filter values indexing + + b.ge 2b + + + // 4 iterations left + + sub x17, x7, #8 // step back to wrap up the filter pos for last 4 elements + + ldr s4, [x8] // load src values for idx 0 + ldr d31, [x12, x17] // load filter values for idx 0 + uxtl v4.8h, v4.8b // extend type to match the filter' size + ldr s5, [x9] // load src values for idx 1 + smlal v16.4s, v4.4h, v31.4h + ldr d30, [x13, x17] // load filter values for idx 1 + uxtl v5.8h, v5.8b // extend type to match the filter' size + ldr s6, [x10] // load src values for idx 2 + smlal v17.4s, v5.4h, v30.4h + uxtl v6.8h, v6.8B // extend type to match the filter's size + ldr d29, [x14, x17] // load filter values for idx 2 + ldr s7, [x11] // load src values for idx 3 + addp v16.4s, v16.4s, v17.4s + uxtl v7.8h, v7.8B + ldr d28, [x15, x17] // load filter values for idx 3 + smlal v18.4s, v6.4h, v29.4h + smlal v19.4s, v7.4h, v28.4h + subs w2, w2, #4 + addp v18.4s, v18.4s, v19.4s + addp v16.4s, v16.4s, v18.4s + sshr v16.4s, v16.4s, #3 + smin v16.4s, v16.4s, v20.4s + + st1 {v16.4s}, [x1], #16 + add x4, x4, x7, lsl #2 + b.gt 1b + ret + +endfunc diff --git a/libswscale/aarch64/swscale.c b/libswscale/aarch64/swscale.c index d1312c6658..171fb51329 100644 --- a/libswscale/aarch64/swscale.c +++ b/libswscale/aarch64/swscale.c @@ -29,7 +29,8 @@ void ff_hscale ## from_bpc ## to ## to_bpc ## _ ## filter_n ## _ ## opt( \ const int16_t *filter, \ const int32_t *filterPos, int filterSize) #define SCALE_FUNCS(filter_n, opt) \ - SCALE_FUNC(filter_n, 8, 15, opt); + SCALE_FUNC(filter_n, 8, 15, opt); \ + SCALE_FUNC(filter_n, 8, 19, opt); #define ALL_SCALE_FUNCS(opt) \ SCALE_FUNCS(4, opt); \ SCALE_FUNCS(X8, opt); \ @@ -48,9 +49,13 @@ void ff_yuv2plane1_8_neon( int offset); #define ASSIGN_SCALE_FUNC2(hscalefn, filtersize, opt) do { \ - if (c->srcBpc == 8 && c->dstBpc <= 14) { \ - hscalefn = \ - ff_hscale8to15_ ## filtersize ## _ ## opt; \ + if (c->srcBpc == 8) { \ + if(c->dstBpc <= 14) { \ + hscalefn = \ + ff_hscale8to15_ ## filtersize ## _ ## opt; \ + } else \ + hscalefn = \ + ff_hscale8to19_ ## filtersize ## _ ## opt; \ } \ } while (0)