From d46684710950a6695ee22c079327253a91eb268d Mon Sep 17 00:00:00 2001 From: Josh Blum Date: Wed, 8 Feb 2012 19:22:38 -0800 Subject: uhd: added sse2 conversions for fc64 to sc8 --- host/lib/convert/convert_fc64_to_sc8_with_sse2.cpp | 156 +++++++++++++++++++++ 1 file changed, 156 insertions(+) create mode 100644 host/lib/convert/convert_fc64_to_sc8_with_sse2.cpp (limited to 'host/lib/convert/convert_fc64_to_sc8_with_sse2.cpp') diff --git a/host/lib/convert/convert_fc64_to_sc8_with_sse2.cpp b/host/lib/convert/convert_fc64_to_sc8_with_sse2.cpp new file mode 100644 index 000000000..405850601 --- /dev/null +++ b/host/lib/convert/convert_fc64_to_sc8_with_sse2.cpp @@ -0,0 +1,156 @@ +// +// Copyright 2012 Ettus Research LLC +// +// This program is free software: you can redistribute it and/or modify +// it under the terms of the GNU General Public License as published by +// the Free Software Foundation, either version 3 of the License, or +// (at your option) any later version. +// +// This program 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 General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with this program. If not, see . +// + +#include "convert_common.hpp" +#include +#include + +using namespace uhd::convert; + +UHD_INLINE __m128i pack_sc8_item32_4x( + const __m128i &in0, const __m128i &in1, + const __m128i &in2, const __m128i &in3 +){ + const __m128i lo = _mm_packs_epi32(in0, in1); + const __m128i hi = _mm_packs_epi32(in2, in3); + return _mm_packs_epi16(lo, hi); +} + +UHD_INLINE __m128i pack_sc32_4x_be( + const __m128d &lo, const __m128d &hi, + const __m128d &scalar +){ + const __m128i tmpi_lo = _mm_cvttpd_epi32(_mm_mul_pd(hi, scalar)); + const __m128i tmpi_hi = _mm_cvttpd_epi32(_mm_mul_pd(lo, scalar)); + return _mm_unpacklo_epi64(tmpi_lo, tmpi_hi); +} + +UHD_INLINE __m128i pack_sc32_4x_le( + const __m128d &lo, const __m128d &hi, + const __m128d &scalar +){ + const __m128i tmpi_lo = _mm_cvttpd_epi32(_mm_mul_pd(lo, scalar)); + const __m128i tmpi_hi = _mm_cvttpd_epi32(_mm_mul_pd(hi, scalar)); + const __m128i tmpi = _mm_unpacklo_epi64(tmpi_lo, tmpi_hi); + return _mm_shuffle_epi32(tmpi, _MM_SHUFFLE(2, 3, 0, 1)); +} + +DECLARE_CONVERTER(fc64, 1, sc8_item32_be, 1, PRIORITY_SIMD){ + const fc64_t *input = reinterpret_cast(inputs[0]); + item32_t *output = reinterpret_cast(outputs[0]); + + const __m128d scalar = _mm_set1_pd(scale_factor); + + #define convert_fc64_1_to_sc8_item32_1_bswap_guts(_al_) \ + for (size_t j = 0; i+7 < nsamps; i+=8, j+=4){ \ + /* load from input */ \ + __m128d tmp0 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+0)); \ + __m128d tmp1 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+1)); \ + __m128d tmp2 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+2)); \ + __m128d tmp3 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+3)); \ + __m128d tmp4 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+4)); \ + __m128d tmp5 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+5)); \ + __m128d tmp6 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+6)); \ + __m128d tmp7 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+7)); \ + \ + /* interleave */ \ + const __m128i tmpi = pack_sc8_item32_4x( \ + pack_sc32_4x_be(tmp0, tmp1, scalar), \ + pack_sc32_4x_be(tmp2, tmp3, scalar), \ + pack_sc32_4x_be(tmp4, tmp5, scalar), \ + pack_sc32_4x_be(tmp6, tmp7, scalar) \ + ); \ + \ + /* store to output */ \ + _mm_storeu_si128(reinterpret_cast<__m128i *>(output+j), tmpi); \ + } \ + + size_t i = 0; + + //dispatch according to alignment + if ((size_t(input) & 0xf) == 0){ + convert_fc64_1_to_sc8_item32_1_bswap_guts(_) + } + else{ + convert_fc64_1_to_sc8_item32_1_bswap_guts(u_) + } + + //convert remainder + const size_t num_pairs = nsamps/2; + for (size_t j = i/2; j < num_pairs; j++, i+=2){ + const item32_t item = fc64_to_item32_sc8(input[i], input[i+1], scale_factor); + output[j] = uhd::byteswap(item); + } + + if (nsamps != num_pairs*2){ + const item32_t item = fc64_to_item32_sc8(input[nsamps-1], 0, scale_factor); + output[num_pairs] = uhd::byteswap(item); + } +} + +DECLARE_CONVERTER(fc64, 1, sc8_item32_le, 1, PRIORITY_SIMD){ + const fc64_t *input = reinterpret_cast(inputs[0]); + item32_t *output = reinterpret_cast(outputs[0]); + + const __m128d scalar = _mm_set1_pd(scale_factor); + + #define convert_fc64_1_to_sc8_item32_1_nswap_guts(_al_) \ + for (size_t j = 0; i+7 < nsamps; i+=8, j+=4){ \ + /* load from input */ \ + __m128d tmp0 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+0)); \ + __m128d tmp1 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+1)); \ + __m128d tmp2 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+2)); \ + __m128d tmp3 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+3)); \ + __m128d tmp4 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+4)); \ + __m128d tmp5 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+5)); \ + __m128d tmp6 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+6)); \ + __m128d tmp7 = _mm_load ## _al_ ## pd(reinterpret_cast(input+i+7)); \ + \ + /* interleave */ \ + const __m128i tmpi = pack_sc8_item32_4x( \ + pack_sc32_4x_le(tmp0, tmp1, scalar), \ + pack_sc32_4x_le(tmp2, tmp3, scalar), \ + pack_sc32_4x_le(tmp4, tmp5, scalar), \ + pack_sc32_4x_le(tmp6, tmp7, scalar) \ + ); \ + \ + /* store to output */ \ + _mm_storeu_si128(reinterpret_cast<__m128i *>(output+j), tmpi); \ + } \ + + size_t i = 0; + + //dispatch according to alignment + if ((size_t(input) & 0xf) == 0){ + convert_fc64_1_to_sc8_item32_1_nswap_guts(_) + } + else{ + convert_fc64_1_to_sc8_item32_1_nswap_guts(u_) + } + + //convert remainder + const size_t num_pairs = nsamps/2; + for (size_t j = i/2; j < num_pairs; j++, i+=2){ + const item32_t item = fc64_to_item32_sc8(input[i], input[i+1], scale_factor); + output[j] = (item); + } + + if (nsamps != num_pairs*2){ + const item32_t item = fc64_to_item32_sc8(input[nsamps-1], 0, scale_factor); + output[num_pairs] = (item); + } +} -- cgit v1.2.3