From: Scott LaVarnway Date: Sat, 12 Jan 2013 01:11:04 +0000 (-0800) Subject: WIP: Added sse2 version of vp9_mb_lpf_horizontal_edge_w X-Git-Tag: v1.3.0~1210^2~12^2 X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=b20ce07d76d9b8fa2230bfccdd11324b565e39c0;p=libvpx WIP: Added sse2 version of vp9_mb_lpf_horizontal_edge_w and vp9_mb_lpf_vertical_edge_w_sse2. This was quickly done so we can run some tests over the weekend. Future commits will optimize/refactor these functions further. The decoder performance improved by ~17% for the clip used. Change-Id: I612687cd5a7670ee840a0cbc3c68dc2b84d4af76 --- diff --git a/vp9/common/x86/vp9_loopfilter_x86.c b/vp9/common/x86/vp9_loopfilter_x86.c index 9f46eec4c..3cf10ba20 100644 --- a/vp9/common/x86/vp9_loopfilter_x86.c +++ b/vp9/common/x86/vp9_loopfilter_x86.c @@ -85,6 +85,482 @@ void vp9_loop_filter_bvs_mmx(unsigned char *y_ptr, int y_stride, #endif #if HAVE_SSE2 + +#if CONFIG_WIDERLPF +void vp9_mb_lpf_horizontal_edge_w_sse2(unsigned char *s, + int p, + const unsigned char *_blimit, + const unsigned char *_limit, + const unsigned char *_thresh) { + DECLARE_ALIGNED(16, unsigned char, flat2_op[7][16]); + DECLARE_ALIGNED(16, unsigned char, flat2_oq[7][16]); + + DECLARE_ALIGNED(16, unsigned char, flat_op2[16]); + DECLARE_ALIGNED(16, unsigned char, flat_op1[16]); + DECLARE_ALIGNED(16, unsigned char, flat_op0[16]); + DECLARE_ALIGNED(16, unsigned char, flat_oq0[16]); + DECLARE_ALIGNED(16, unsigned char, flat_oq1[16]); + DECLARE_ALIGNED(16, unsigned char, flat_oq2[16]); + __m128i mask, hev, flat, flat2; + const __m128i zero = _mm_set1_epi16(0); + __m128i p7, p6, p5; + __m128i p4, p3, p2, p1, p0, q0, q1, q2, q3, q4; + __m128i q5, q6, q7; + int i = 0; + const unsigned int extended_thresh = _thresh[0] * 0x01010101u; + const unsigned int extended_limit = _limit[0] * 0x01010101u; + const unsigned int extended_blimit = _blimit[0] * 0x01010101u; + const __m128i thresh = + _mm_shuffle_epi32(_mm_cvtsi32_si128((int)extended_thresh), 0); + const __m128i limit = + _mm_shuffle_epi32(_mm_cvtsi32_si128((int)extended_limit), 0); + const __m128i blimit = + _mm_shuffle_epi32(_mm_cvtsi32_si128((int)extended_blimit), 0); + + p4 = _mm_loadu_si128((__m128i *)(s - 5 * p)); + p3 = _mm_loadu_si128((__m128i *)(s - 4 * p)); + p2 = _mm_loadu_si128((__m128i *)(s - 3 * p)); + p1 = _mm_loadu_si128((__m128i *)(s - 2 * p)); + p0 = _mm_loadu_si128((__m128i *)(s - 1 * p)); + q0 = _mm_loadu_si128((__m128i *)(s - 0 * p)); + q1 = _mm_loadu_si128((__m128i *)(s + 1 * p)); + q2 = _mm_loadu_si128((__m128i *)(s + 2 * p)); + q3 = _mm_loadu_si128((__m128i *)(s + 3 * p)); + q4 = _mm_loadu_si128((__m128i *)(s + 4 * p)); + { + const __m128i abs_p1p0 = _mm_or_si128(_mm_subs_epu8(p1, p0), + _mm_subs_epu8(p0, p1)); + const __m128i abs_q1q0 = _mm_or_si128(_mm_subs_epu8(q1, q0), + _mm_subs_epu8(q0, q1)); + const __m128i one = _mm_set1_epi8(1); + const __m128i fe = _mm_set1_epi8(0xfe); + const __m128i ff = _mm_cmpeq_epi8(abs_p1p0, abs_p1p0); + __m128i abs_p0q0 = _mm_or_si128(_mm_subs_epu8(p0, q0), + _mm_subs_epu8(q0, p0)); + __m128i abs_p1q1 = _mm_or_si128(_mm_subs_epu8(p1, q1), + _mm_subs_epu8(q1, p1)); + __m128i work; + flat = _mm_max_epu8(abs_p1p0, abs_q1q0); + hev = _mm_subs_epu8(flat, thresh); + hev = _mm_xor_si128(_mm_cmpeq_epi8(hev, zero), ff); + + abs_p0q0 =_mm_adds_epu8(abs_p0q0, abs_p0q0); + abs_p1q1 = _mm_srli_epi16(_mm_and_si128(abs_p1q1, fe), 1); + mask = _mm_subs_epu8(_mm_adds_epu8(abs_p0q0, abs_p1q1), blimit); + mask = _mm_xor_si128(_mm_cmpeq_epi8(mask, zero), ff); + // mask |= (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 > blimit) * -1; + mask = _mm_max_epu8(flat, mask); + // mask |= (abs(p1 - p0) > limit) * -1; + // mask |= (abs(q1 - q0) > limit) * -1; + work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(p2, p1), + _mm_subs_epu8(p1, p2)), + _mm_or_si128(_mm_subs_epu8(p3, p2), + _mm_subs_epu8(p2, p3))); + mask = _mm_max_epu8(work, mask); + work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(q2, q1), + _mm_subs_epu8(q1, q2)), + _mm_or_si128(_mm_subs_epu8(q3, q2), + _mm_subs_epu8(q2, q3))); + mask = _mm_max_epu8(work, mask); + mask = _mm_subs_epu8(mask, limit); + mask = _mm_cmpeq_epi8(mask, zero); + + work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(p2, p0), + _mm_subs_epu8(p0, p2)), + _mm_or_si128(_mm_subs_epu8(q2, q0), + _mm_subs_epu8(q0, q2))); + flat = _mm_max_epu8(work, flat); + work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(p3, p0), + _mm_subs_epu8(p0, p3)), + _mm_or_si128(_mm_subs_epu8(q3, q0), + _mm_subs_epu8(q0, q3))); + flat = _mm_max_epu8(work, flat); + work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(p4, p0), + _mm_subs_epu8(p0, p4)), + _mm_or_si128(_mm_subs_epu8(q4, q0), + _mm_subs_epu8(q0, q4))); + flat = _mm_max_epu8(work, flat); + flat = _mm_subs_epu8(flat, one); + flat = _mm_cmpeq_epi8(flat, zero); + flat = _mm_and_si128(flat, mask); + } + + // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + // calculate flat2 + p4 = _mm_loadu_si128((__m128i *)(s - 8 * p)); + p3 = _mm_loadu_si128((__m128i *)(s - 7 * p)); + p2 = _mm_loadu_si128((__m128i *)(s - 6 * p)); + p1 = _mm_loadu_si128((__m128i *)(s - 5 * p)); +// p0 = _mm_loadu_si128((__m128i *)(s - 1 * p)); +// q0 = _mm_loadu_si128((__m128i *)(s - 0 * p)); + q1 = _mm_loadu_si128((__m128i *)(s + 4 * p)); + q2 = _mm_loadu_si128((__m128i *)(s + 5 * p)); + q3 = _mm_loadu_si128((__m128i *)(s + 6 * p)); + q4 = _mm_loadu_si128((__m128i *)(s + 7 * p)); + + { + const __m128i abs_p1p0 = _mm_or_si128(_mm_subs_epu8(p1, p0), + _mm_subs_epu8(p0, p1)); + const __m128i abs_q1q0 = _mm_or_si128(_mm_subs_epu8(q1, q0), + _mm_subs_epu8(q0, q1)); + const __m128i one = _mm_set1_epi8(1); + __m128i work; + flat2 = _mm_max_epu8(abs_p1p0, abs_q1q0); + work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(p2, p0), + _mm_subs_epu8(p0, p2)), + _mm_or_si128(_mm_subs_epu8(q2, q0), + _mm_subs_epu8(q0, q2))); + flat2 = _mm_max_epu8(work, flat2); + work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(p3, p0), + _mm_subs_epu8(p0, p3)), + _mm_or_si128(_mm_subs_epu8(q3, q0), + _mm_subs_epu8(q0, q3))); + flat2 = _mm_max_epu8(work, flat2); + work = _mm_max_epu8(_mm_or_si128(_mm_subs_epu8(p4, p0), + _mm_subs_epu8(p0, p4)), + _mm_or_si128(_mm_subs_epu8(q4, q0), + _mm_subs_epu8(q0, q4))); + flat2 = _mm_max_epu8(work, flat2); + flat2 = _mm_subs_epu8(flat2, one); + flat2 = _mm_cmpeq_epi8(flat2, zero); + flat2 = _mm_and_si128(flat2, flat); // flat2 & flat & mask + } + // calculate flat2 + // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + + { + const __m128i four = _mm_set1_epi16(4); + unsigned char *src = s; + i = 0; + do { + __m128i workp_a, workp_b, workp_shft; + p4 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 5 * p)), zero); + p3 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 4 * p)), zero); + p2 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 3 * p)), zero); + p1 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 2 * p)), zero); + p0 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 1 * p)), zero); + q0 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 0 * p)), zero); + q1 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 1 * p)), zero); + q2 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 2 * p)), zero); + q3 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 3 * p)), zero); + q4 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 4 * p)), zero); + + workp_a = _mm_add_epi16(_mm_add_epi16(p4, p3), _mm_add_epi16(p2, p1)); + workp_a = _mm_add_epi16(_mm_add_epi16(workp_a, four), p0); + workp_b = _mm_add_epi16(_mm_add_epi16(q0, p2), p4); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 3); + _mm_storel_epi64((__m128i *)&flat_op2[i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_b = _mm_add_epi16(_mm_add_epi16(q0, q1), p1); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 3); + _mm_storel_epi64((__m128i *)&flat_op1[i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p4), q2); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p1), p0); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 3); + _mm_storel_epi64((__m128i *)&flat_op0[i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p3), q3); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p0), q0); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 3); + _mm_storel_epi64((__m128i *)&flat_oq0[i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p2), q4); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, q0), q1); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 3); + _mm_storel_epi64((__m128i *)&flat_oq1[i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p1), q4); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, q1), q2); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 3); + _mm_storel_epi64((__m128i *)&flat_oq2[i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + src += 8; + } while (++i < 2); + } + // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + // wide flat + // TODO(slavarnway): interleave with the flat pixel calculations (see above) + { + const __m128i eight = _mm_set1_epi16(8); + unsigned char *src = s; + int i = 0; + do { + __m128i workp_a, workp_b, workp_shft; + p7 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 8 * p)), zero); + p6 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 7 * p)), zero); + p5 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 6 * p)), zero); + p4 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 5 * p)), zero); + p3 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 4 * p)), zero); + p2 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 3 * p)), zero); + p1 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 2 * p)), zero); + p0 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 1 * p)), zero); + q0 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src - 0 * p)), zero); + q1 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 1 * p)), zero); + q2 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 2 * p)), zero); + q3 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 3 * p)), zero); + q4 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 4 * p)), zero); + q5 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 5 * p)), zero); + q6 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 6 * p)), zero); + q7 = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i *)(src + 7 * p)), zero); + + + workp_a = _mm_sub_epi16(_mm_slli_epi16(p7, 3), p7); // p7 * 7 + workp_a = _mm_add_epi16(_mm_slli_epi16(p6, 1), workp_a); + workp_b = _mm_add_epi16(_mm_add_epi16(p5, p4), _mm_add_epi16(p3, p2)); + workp_a = _mm_add_epi16(_mm_add_epi16(p1, p0), workp_a); + workp_b = _mm_add_epi16(_mm_add_epi16(q0, eight), workp_b); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_op[6][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p7), p5); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p6), q1); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_op[5][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p7), p4); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p5), q2); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_op[4][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p7), p3); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p4), q3); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_op[3][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p7), p2); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p3), q4); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_op[2][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p7), p1); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p2), q5); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_op[1][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p7), p0); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p1), q6); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_op[0][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p7), q0); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, p0), q7); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_oq[0][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p6), q1); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, q0), q7); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_oq[1][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p5), q2); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, q1), q7); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_oq[2][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p4), q3); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, q2), q7); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_oq[3][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p3), q4); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, q3), q7); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_oq[4][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p2), q5); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, q4), q7); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_oq[5][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + workp_a = _mm_add_epi16(_mm_sub_epi16(workp_a, p1), q6); + workp_b = _mm_add_epi16(_mm_sub_epi16(workp_b, q5), q7); + workp_shft = _mm_srli_epi16(_mm_add_epi16(workp_a, workp_b), 4); + _mm_storel_epi64((__m128i *)&flat2_oq[6][i*8], + _mm_packus_epi16(workp_shft, workp_shft)); + + src += 8; + } while (++i < 2); + } + // wide flat + // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + + // lp filter + { + const __m128i t4 = _mm_set1_epi8(4); + const __m128i t3 = _mm_set1_epi8(3); + const __m128i t80 = _mm_set1_epi8(0x80); + const __m128i te0 = _mm_set1_epi8(0xe0); + const __m128i t1f = _mm_set1_epi8(0x1f); + const __m128i t1 = _mm_set1_epi8(0x1); + const __m128i t7f = _mm_set1_epi8(0x7f); + + __m128i ps1 = _mm_xor_si128(_mm_loadu_si128((__m128i *)(s - 2 * p)), + t80); + __m128i ps0 = _mm_xor_si128(_mm_loadu_si128((__m128i *)(s - 1 * p)), + t80); + __m128i qs0 = _mm_xor_si128(_mm_loadu_si128((__m128i *)(s + 0 * p)), + t80); + __m128i qs1 = _mm_xor_si128(_mm_loadu_si128((__m128i *)(s + 1 * p)), + t80); + __m128i filt; + __m128i work_a; + __m128i filter1, filter2; + + filt = _mm_and_si128(_mm_subs_epi8(ps1, qs1), hev); + work_a = _mm_subs_epi8(qs0, ps0); + filt = _mm_adds_epi8(filt, work_a); + filt = _mm_adds_epi8(filt, work_a); + filt = _mm_adds_epi8(filt, work_a); + /* (vp9_filter + 3 * (qs0 - ps0)) & mask */ + filt = _mm_and_si128(filt, mask); + + filter1 = _mm_adds_epi8(filt, t4); + filter2 = _mm_adds_epi8(filt, t3); + + /* Filter1 >> 3 */ + work_a = _mm_cmpgt_epi8(zero, filter1); + filter1 = _mm_srli_epi16(filter1, 3); + work_a = _mm_and_si128(work_a, te0); + filter1 = _mm_and_si128(filter1, t1f); + filter1 = _mm_or_si128(filter1, work_a); + + /* Filter2 >> 3 */ + work_a = _mm_cmpgt_epi8(zero, filter2); + filter2 = _mm_srli_epi16(filter2, 3); + work_a = _mm_and_si128(work_a, te0); + filter2 = _mm_and_si128(filter2, t1f); + filter2 = _mm_or_si128(filter2, work_a); + + /* filt >> 1 */ + filt = _mm_adds_epi8(filter1, t1); + work_a = _mm_cmpgt_epi8(zero, filt); + filt = _mm_srli_epi16(filt, 1); + work_a = _mm_and_si128(work_a, t80); + filt = _mm_and_si128(filt, t7f); + filt = _mm_or_si128(filt, work_a); + + filt = _mm_andnot_si128(hev, filt); + + ps0 = _mm_xor_si128(_mm_adds_epi8(ps0, filter2), t80); + ps1 = _mm_xor_si128(_mm_adds_epi8(ps1, filt), t80); + qs0 = _mm_xor_si128(_mm_subs_epi8(qs0, filter1), t80); + qs1 = _mm_xor_si128(_mm_subs_epi8(qs1, filt), t80); + + // write out op6 - op3 + { + unsigned char *dst = (s - 7 * p); + for (i = 6; i > 2; i--) { + __m128i flat2_output; + work_a = _mm_loadu_si128((__m128i *)dst); + flat2_output = _mm_load_si128((__m128i *)flat2_op[i]); + work_a = _mm_andnot_si128(flat2, work_a); + flat2_output = _mm_and_si128(flat2, flat2_output); + work_a = _mm_or_si128(work_a, flat2_output); + _mm_storeu_si128((__m128i *)dst, work_a); + dst += p; + } + } + + work_a = _mm_loadu_si128((__m128i *)(s - 3 * p)); + p2 = _mm_load_si128((__m128i *)flat_op2); + work_a = _mm_andnot_si128(flat, work_a); + p2 = _mm_and_si128(flat, p2); + work_a = _mm_or_si128(work_a, p2); + p2 = _mm_load_si128((__m128i *)flat2_op[2]); + work_a = _mm_andnot_si128(flat2, work_a); + p2 = _mm_and_si128(flat2, p2); + p2 = _mm_or_si128(work_a, p2); + _mm_storeu_si128((__m128i *)(s - 3 * p), p2); + + p1 = _mm_load_si128((__m128i *)flat_op1); + work_a = _mm_andnot_si128(flat, ps1); + p1 = _mm_and_si128(flat, p1); + work_a = _mm_or_si128(work_a, p1); + p1 = _mm_load_si128((__m128i *)flat2_op[1]); + work_a = _mm_andnot_si128(flat2, work_a); + p1 = _mm_and_si128(flat2, p1); + p1 = _mm_or_si128(work_a, p1); + _mm_storeu_si128((__m128i *)(s - 2 * p), p1); + + p0 = _mm_load_si128((__m128i *)flat_op0); + work_a = _mm_andnot_si128(flat, ps0); + p0 = _mm_and_si128(flat, p0); + work_a = _mm_or_si128(work_a, p0); + p0 = _mm_load_si128((__m128i *)flat2_op[0]); + work_a = _mm_andnot_si128(flat2, work_a); + p0 = _mm_and_si128(flat2, p0); + p0 = _mm_or_si128(work_a, p0); + _mm_storeu_si128((__m128i *)(s - 1 * p), p0); + + q0 = _mm_load_si128((__m128i *)flat_oq0); + work_a = _mm_andnot_si128(flat, qs0); + q0 = _mm_and_si128(flat, q0); + work_a = _mm_or_si128(work_a, q0); + q0 = _mm_load_si128((__m128i *)flat2_oq[0]); + work_a = _mm_andnot_si128(flat2, work_a); + q0 = _mm_and_si128(flat2, q0); + q0 = _mm_or_si128(work_a, q0); + _mm_storeu_si128((__m128i *)(s - 0 * p), q0); + + q1 = _mm_load_si128((__m128i *)flat_oq1); + work_a = _mm_andnot_si128(flat, qs1); + q1 = _mm_and_si128(flat, q1); + work_a = _mm_or_si128(work_a, q1); + q1 = _mm_load_si128((__m128i *)flat2_oq[1]); + work_a = _mm_andnot_si128(flat2, work_a); + q1 = _mm_and_si128(flat2, q1); + q1 = _mm_or_si128(work_a, q1); + _mm_storeu_si128((__m128i *)(s + 1 * p), q1); + + work_a = _mm_loadu_si128((__m128i *)(s + 2 * p)); + q2 = _mm_load_si128((__m128i *)flat_oq2); + work_a = _mm_andnot_si128(flat, work_a); + q2 = _mm_and_si128(flat, q2); + work_a = _mm_or_si128(work_a, q2); + q2 = _mm_load_si128((__m128i *)flat2_oq[2]); + work_a = _mm_andnot_si128(flat2, work_a); + q2 = _mm_and_si128(flat2, q2); + q2 = _mm_or_si128(work_a, q2); + _mm_storeu_si128((__m128i *)(s + 2 * p), q2); + + // write out oq3 - oq7 + { + unsigned char *dst = (s + 3 * p); + for (i = 3; i < 7; i++) { + __m128i flat2_output; + work_a = _mm_loadu_si128((__m128i *)dst); + flat2_output = _mm_load_si128((__m128i *)flat2_oq[i]); + work_a = _mm_andnot_si128(flat2, work_a); + flat2_output = _mm_and_si128(flat2, flat2_output); + work_a = _mm_or_si128(work_a, flat2_output); + _mm_storeu_si128((__m128i *)dst, work_a); + dst += p; + } + } + } +} +#endif + void vp9_mbloop_filter_horizontal_edge_sse2(unsigned char *s, int p, const unsigned char *_blimit, @@ -562,6 +1038,39 @@ void vp9_mbloop_filter_vertical_edge_sse2(unsigned char *s, transpose(src, 16, dst, p, 2); } +#if CONFIG_WIDERLPF +void vp9_mb_lpf_vertical_edge_w_sse2(unsigned char *s, + int p, + const unsigned char *blimit, + const unsigned char *limit, + const unsigned char *thresh) { + DECLARE_ALIGNED_ARRAY(16, unsigned char, t_dst, 256); + unsigned char *src[4]; + unsigned char *dst[4]; + + /* Transpose 16x16 */ + transpose8x16(s - 8, s - 8 + p * 8, p, t_dst, 16); + transpose8x16(s, s + p * 8, p, t_dst + 16 * 8, 16); + + /* Loop filtering */ + vp9_mb_lpf_horizontal_edge_w_sse2(t_dst + 8 * 16, 16, blimit, limit, + thresh); + + src[0] = t_dst; + src[1] = t_dst + 8 * 16; + src[2] = t_dst + 8; + src[3] = t_dst + 8 * 16 + 8; + + dst[0] = s - 8; + dst[1] = s - 8 + 8; + dst[2] = s - 8 + p * 8; + dst[3] = s - 8 + p * 8 + 8; + + /* Transpose 16x16 */ + transpose(src, 16, dst, p, 4); +} +#endif + void vp9_mbloop_filter_vertical_edge_uv_sse2(unsigned char *u, int p, const unsigned char *blimit, @@ -608,8 +1117,8 @@ void vp9_loop_filter_mbh_sse2(unsigned char *y_ptr, void vp9_lpf_mbh_w_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi) { - vp9_mb_lpf_horizontal_edge_w(y_ptr, y_stride, - lfi->mblim, lfi->lim, lfi->hev_thr, 2); + vp9_mb_lpf_horizontal_edge_w_sse2(y_ptr, y_stride, + lfi->mblim, lfi->lim, lfi->hev_thr); /* u,v */ if (u_ptr) @@ -642,8 +1151,8 @@ void vp9_loop_filter_mbv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, void vp9_lpf_mbv_w_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi) { - vp9_mb_lpf_vertical_edge_w(y_ptr, y_stride, - lfi->mblim, lfi->lim, lfi->hev_thr, 2); + vp9_mb_lpf_vertical_edge_w_sse2(y_ptr, y_stride, + lfi->mblim, lfi->lim, lfi->hev_thr); /* u,v */ if (u_ptr)