Philipp Maier | e8ae9fc | 2017-03-20 12:08:42 +0100 | [diff] [blame] | 1 | /* |
| 2 | * SSE type conversions |
| 3 | * Copyright (C) 2013 Thomas Tsou <tom@tsou.cc> |
| 4 | * |
| 5 | * This library is free software; you can redistribute it and/or |
| 6 | * modify it under the terms of the GNU Lesser General Public |
| 7 | * License as published by the Free Software Foundation; either |
| 8 | * version 2.1 of the License, or (at your option) any later version. |
| 9 | * |
| 10 | * This library is distributed in the hope that it will be useful, |
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 13 | * Lesser General Public License for more details. |
| 14 | * |
| 15 | * You should have received a copy of the GNU Lesser General Public |
| 16 | * License along with this library; if not, write to the Free Software |
| 17 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 18 | */ |
| 19 | |
| 20 | #include <malloc.h> |
| 21 | #include <string.h> |
| 22 | #include "convert_sse_4_1.h" |
| 23 | |
| 24 | #ifdef HAVE_CONFIG_H |
| 25 | #include "config.h" |
| 26 | #endif |
| 27 | |
| 28 | #ifdef HAVE_SSE4_1 |
| 29 | #include <smmintrin.h> |
| 30 | |
| 31 | /* 16*N 16-bit signed integer converted to single precision floats */ |
| 32 | void _sse_convert_si16_ps_16n(float *restrict out, |
| 33 | const short *restrict in, int len) |
| 34 | { |
| 35 | __m128i m0, m1, m2, m3, m4, m5; |
| 36 | __m128 m6, m7, m8, m9; |
| 37 | |
| 38 | for (int i = 0; i < len / 16; i++) { |
| 39 | /* Load (unaligned) packed floats */ |
| 40 | m0 = _mm_loadu_si128((__m128i *) & in[16 * i + 0]); |
| 41 | m1 = _mm_loadu_si128((__m128i *) & in[16 * i + 8]); |
| 42 | |
| 43 | /* Unpack */ |
| 44 | m2 = _mm_cvtepi16_epi32(m0); |
| 45 | m4 = _mm_cvtepi16_epi32(m1); |
| 46 | m0 = _mm_shuffle_epi32(m0, _MM_SHUFFLE(1, 0, 3, 2)); |
| 47 | m1 = _mm_shuffle_epi32(m1, _MM_SHUFFLE(1, 0, 3, 2)); |
| 48 | m3 = _mm_cvtepi16_epi32(m0); |
| 49 | m5 = _mm_cvtepi16_epi32(m1); |
| 50 | |
| 51 | /* Convert */ |
| 52 | m6 = _mm_cvtepi32_ps(m2); |
| 53 | m7 = _mm_cvtepi32_ps(m3); |
| 54 | m8 = _mm_cvtepi32_ps(m4); |
| 55 | m9 = _mm_cvtepi32_ps(m5); |
| 56 | |
| 57 | /* Store */ |
| 58 | _mm_storeu_ps(&out[16 * i + 0], m6); |
| 59 | _mm_storeu_ps(&out[16 * i + 4], m7); |
| 60 | _mm_storeu_ps(&out[16 * i + 8], m8); |
| 61 | _mm_storeu_ps(&out[16 * i + 12], m9); |
| 62 | } |
| 63 | } |
| 64 | |
| 65 | /* 16*N 16-bit signed integer conversion with remainder */ |
| 66 | void _sse_convert_si16_ps(float *restrict out, |
| 67 | const short *restrict in, int len) |
| 68 | { |
| 69 | int start = len / 16 * 16; |
| 70 | |
| 71 | _sse_convert_si16_ps_16n(out, in, len); |
| 72 | |
| 73 | for (int i = 0; i < len % 16; i++) |
| 74 | out[start + i] = in[start + i]; |
| 75 | } |
| 76 | |
| 77 | #endif |