Mercurial > hg > audiostuff
view spandsp-0.0.6pre17/src/vector_float.c @ 6:22a74b01a099 default tip
implement more meaningful test program
author | Peter Meerwald <pmeerw@cosy.sbg.ac.at> |
---|---|
date | Fri, 25 Jun 2010 16:14:50 +0200 |
parents | 26cd8f1ef0b1 |
children |
line wrap: on
line source
/* * SpanDSP - a series of DSP components for telephony * * vector_float.c - Floating vector arithmetic routines. * * Written by Steve Underwood <steveu@coppice.org> * * Copyright (C) 2006 Steve Underwood * * All rights reserved. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU Lesser General Public License version 2.1, * as published by the Free Software Foundation. * * 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 Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this program; if not, write to the Free Software * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. * * $Id: vector_float.c,v 1.22 2009/07/12 09:23:09 steveu Exp $ */ /*! \file */ #if defined(HAVE_CONFIG_H) #include "config.h" #endif #include <inttypes.h> #include <stdlib.h> #include <stdio.h> #include <string.h> #if defined(HAVE_TGMATH_H) #include <tgmath.h> #endif #if defined(HAVE_MATH_H) #include <math.h> #endif #include <assert.h> #include "floating_fudge.h" #include "mmx_sse_decs.h" #include "spandsp/telephony.h" #include "spandsp/vector_float.h" #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_copyf(float z[], const float x[], int n) { int i; __m128 n1; if ((i = n & ~3)) { for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); _mm_storeu_ps(z + i, n1); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3]; case 2: z[n - 2] = x[n - 2]; case 1: z[n - 1] = x[n - 1]; } } #else SPAN_DECLARE(void) vec_copyf(float z[], const float x[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_copy(double z[], const double x[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_copyl(long double z[], const long double x[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_negatef(float z[], const float x[], int n) { int i; static const uint32_t mask = 0x80000000; static const float *fmask = (float *) &mask; __m128 n1; __m128 n2; if ((i = n & ~3)) { n2 = _mm_set1_ps(*fmask); for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n1 = _mm_xor_ps(n1, n2); _mm_storeu_ps(z + i, n1); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = -x[n - 3]; case 2: z[n - 2] = -x[n - 2]; case 1: z[n - 1] = -x[n - 1]; } } #else SPAN_DECLARE(void) vec_negatef(float z[], const float x[], int n) { int i; for (i = 0; i < n; i++) z[i] = -x[i]; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_negate(double z[], const double x[], int n) { int i; for (i = 0; i < n; i++) z[i] = -x[i]; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_negatel(long double z[], const long double x[], int n) { int i; for (i = 0; i < n; i++) z[i] = -x[i]; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_zerof(float z[], int n) { int i; __m128 n1; if ((i = n & ~3)) { n1 = _mm_setzero_ps(); for (i -= 4; i >= 0; i -= 4) _mm_storeu_ps(z + i, n1); } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = 0; case 2: z[n - 2] = 0; case 1: z[n - 1] = 0; } } #else SPAN_DECLARE(void) vec_zerof(float z[], int n) { int i; for (i = 0; i < n; i++) z[i] = 0.0f; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_zero(double z[], int n) { int i; for (i = 0; i < n; i++) z[i] = 0.0; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_zerol(long double z[], int n) { int i; for (i = 0; i < n; i++) z[i] = 0.0L; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_setf(float z[], float x, int n) { int i; __m128 n1; if ((i = n & ~3)) { n1 = _mm_set1_ps(x); for (i -= 4; i >= 0; i -= 4) _mm_storeu_ps(z + i, n1); } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x; case 2: z[n - 2] = x; case 1: z[n - 1] = x; } } #else SPAN_DECLARE(void) vec_setf(float z[], float x, int n) { int i; for (i = 0; i < n; i++) z[i] = x; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_set(double z[], double x, int n) { int i; for (i = 0; i < n; i++) z[i] = x; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_setl(long double z[], long double x, int n) { int i; for (i = 0; i < n; i++) z[i] = x; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_addf(float z[], const float x[], const float y[], int n) { int i; __m128 n1; __m128 n2; if ((i = n & ~3)) { for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n2 = _mm_loadu_ps(y + i); n2 = _mm_add_ps(n1, n2); _mm_storeu_ps(z + i, n2); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3] + y[n - 3]; case 2: z[n - 2] = x[n - 2] + y[n - 2]; case 1: z[n - 1] = x[n - 1] + y[n - 1]; } } #else SPAN_DECLARE(void) vec_addf(float z[], const float x[], const float y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y[i]; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_add(double z[], const double x[], const double y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y[i]; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_addl(long double z[], const long double x[], const long double y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y[i]; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_scaledxy_addf(float z[], const float x[], float x_scale, const float y[], float y_scale, int n) { int i; __m128 n1; __m128 n2; __m128 n3; __m128 n4; if ((i = n & ~3)) { n3 = _mm_set1_ps(x_scale); n4 = _mm_set1_ps(y_scale); for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n2 = _mm_loadu_ps(y + i); n1 = _mm_mul_ps(n1, n3); n2 = _mm_mul_ps(n2, n4); n2 = _mm_add_ps(n1, n2); _mm_storeu_ps(z + i, n2); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3]*x_scale + y[n - 3]*y_scale; case 2: z[n - 2] = x[n - 2]*x_scale + y[n - 2]*y_scale; case 1: z[n - 1] = x[n - 1]*x_scale + y[n - 1]*y_scale; } } #else SPAN_DECLARE(void) vec_scaledxy_addf(float z[], const float x[], float x_scale, const float y[], float y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*x_scale + y[i]*y_scale; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_scaledxy_add(double z[], const double x[], double x_scale, const double y[], double y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*x_scale + y[i]*y_scale; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_scaledxy_addl(long double z[], const long double x[], long double x_scale, const long double y[], long double y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*x_scale + y[i]*y_scale; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_scaledy_addf(float z[], const float x[], const float y[], float y_scale, int n) { int i; __m128 n1; __m128 n2; __m128 n3; if ((i = n & ~3)) { n3 = _mm_set1_ps(y_scale); for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n2 = _mm_loadu_ps(y + i); n2 = _mm_mul_ps(n2, n3); n2 = _mm_add_ps(n1, n2); _mm_storeu_ps(z + i, n2); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3] + y[n - 3]*y_scale; case 2: z[n - 2] = x[n - 2] + y[n - 2]*y_scale; case 1: z[n - 1] = x[n - 1] + y[n - 1]*y_scale; } } #else SPAN_DECLARE(void) vec_scaledy_addf(float z[], const float x[], const float y[], float y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y[i]*y_scale; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_scaledy_add(double z[], const double x[], const double y[], double y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y[i]*y_scale; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_scaledy_addl(long double z[], const long double x[], const long double y[], long double y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y[i]*y_scale; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_subf(float z[], const float x[], const float y[], int n) { int i; __m128 n1; __m128 n2; if ((i = n & ~3)) { for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n2 = _mm_loadu_ps(y + i); n2 = _mm_sub_ps(n1, n2); _mm_storeu_ps(z + i, n2); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3] - y[n - 3]; case 2: z[n - 2] = x[n - 2] - y[n - 2]; case 1: z[n - 1] = x[n - 1] - y[n - 1]; } } #else SPAN_DECLARE(void) vec_subf(float z[], const float x[], const float y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] - y[i]; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_sub(double z[], const double x[], const double y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] - y[i]; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_subl(long double z[], const long double x[], const long double y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] - y[i]; } /*- End of function --------------------------------------------------------*/ #endif SPAN_DECLARE(void) vec_scaledxy_subf(float z[], const float x[], float x_scale, const float y[], float y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*x_scale - y[i]*y_scale; } /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_scaledxy_sub(double z[], const double x[], double x_scale, const double y[], double y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*x_scale - y[i]*y_scale; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_scaledxy_subl(long double z[], const long double x[], long double x_scale, const long double y[], long double y_scale, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*x_scale - y[i]*y_scale; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_scalar_mulf(float z[], const float x[], float y, int n) { int i; __m128 n1; __m128 n2; if ((i = n & ~3)) { n2 = _mm_set1_ps(y); for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n1 = _mm_mul_ps(n1, n2); _mm_storeu_ps(z + i, n1); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3]*y; case 2: z[n - 2] = x[n - 2]*y; case 1: z[n - 1] = x[n - 1]*y; } } #else SPAN_DECLARE(void) vec_scalar_mulf(float z[], const float x[], float y, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*y; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_scalar_mul(double z[], const double x[], double y, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*y; } /*- End of function --------------------------------------------------------*/ #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_scalar_addf(float z[], const float x[], float y, int n) { int i; __m128 n1; __m128 n2; if ((i = n & ~3)) { n2 = _mm_set1_ps(y); for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n1 = _mm_add_ps(n1, n2); _mm_storeu_ps(z + i, n1); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3] + y; case 2: z[n - 2] = x[n - 2] + y; case 1: z[n - 1] = x[n - 1] + y; } } #else SPAN_DECLARE(void) vec_scalar_addf(float z[], const float x[], float y, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_scalar_add(double z[], const double x[], double y, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_scalar_addl(long double z[], const long double x[], long double y, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] + y; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_scalar_subf(float z[], const float x[], float y, int n) { int i; __m128 n1; __m128 n2; if ((i = n & ~3)) { n2 = _mm_set1_ps(y); for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n1 = _mm_sub_ps(n1, n2); _mm_storeu_ps(z + i, n1); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3] - y; case 2: z[n - 2] = x[n - 2] - y; case 1: z[n - 1] = x[n - 1] - y; } } #else SPAN_DECLARE(void) vec_scalar_subf(float z[], const float x[], float y, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] - y; } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_scalar_sub(double z[], const double x[], double y, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] - y; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_scalar_subl(long double z[], const long double x[], long double y, int n) { int i; for (i = 0; i < n; i++) z[i] = x[i] - y; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_mulf(float z[], const float x[], const float y[], int n) { int i; __m128 n1; __m128 n2; __m128 n3; if ((i = n & ~3)) { for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n2 = _mm_loadu_ps(y + i); n3 = _mm_mul_ps(n1, n2); _mm_storeu_ps(z + i, n3); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z[n - 3] = x[n - 3]*y[n - 3]; case 2: z[n - 2] = x[n - 2]*y[n - 2]; case 1: z[n - 1] = x[n - 1]*y[n - 1]; } } #else SPAN_DECLARE(void) vec_mulf(float z[], const float x[], const float y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*y[i]; } /*- End of function --------------------------------------------------------*/ #endif SPAN_DECLARE(void) vec_mul(double z[], const double x[], const double y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*y[i]; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(void) vec_mull(long double z[], const long double x[], const long double y[], int n) { int i; for (i = 0; i < n; i++) z[i] = x[i]*y[i]; } /*- End of function --------------------------------------------------------*/ #endif #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(float) vec_dot_prodf(const float x[], const float y[], int n) { int i; float z; __m128 n1; __m128 n2; __m128 n3; __m128 n4; z = 0.0f; if ((i = n & ~3)) { n4 = _mm_setzero_ps(); //sets sum to zero for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n2 = _mm_loadu_ps(y + i); n3 = _mm_mul_ps(n1, n2); n4 = _mm_add_ps(n4, n3); } n4 = _mm_add_ps(_mm_movehl_ps(n4, n4), n4); n4 = _mm_add_ss(_mm_shuffle_ps(n4, n4, 1), n4); _mm_store_ss(&z, n4); } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: z += x[n - 3]*y[n - 3]; case 2: z += x[n - 2]*y[n - 2]; case 1: z += x[n - 1]*y[n - 1]; } return z; } #else SPAN_DECLARE(float) vec_dot_prodf(const float x[], const float y[], int n) { int i; float z; z = 0.0f; for (i = 0; i < n; i++) z += x[i]*y[i]; return z; } /*- End of function --------------------------------------------------------*/ #endif SPAN_DECLARE(double) vec_dot_prod(const double x[], const double y[], int n) { int i; double z; z = 0.0; for (i = 0; i < n; i++) z += x[i]*y[i]; return z; } /*- End of function --------------------------------------------------------*/ #if defined(HAVE_LONG_DOUBLE) SPAN_DECLARE(long double) vec_dot_prodl(const long double x[], const long double y[], int n) { int i; long double z; z = 0.0L; for (i = 0; i < n; i++) z += x[i]*y[i]; return z; } /*- End of function --------------------------------------------------------*/ #endif SPAN_DECLARE(float) vec_circular_dot_prodf(const float x[], const float y[], int n, int pos) { float z; z = vec_dot_prodf(&x[pos], &y[0], n - pos); z += vec_dot_prodf(&x[0], &y[n - pos], pos); return z; } /*- End of function --------------------------------------------------------*/ #define LMS_LEAK_RATE 0.9999f #if defined(__GNUC__) && defined(SPANDSP_USE_SSE2) SPAN_DECLARE(void) vec_lmsf(const float x[], float y[], int n, float error) { int i; __m128 n1; __m128 n2; __m128 n3; __m128 n4; if ((i = n & ~3)) { n3 = _mm_set1_ps(error); n4 = _mm_set1_ps(LMS_LEAK_RATE); for (i -= 4; i >= 0; i -= 4) { n1 = _mm_loadu_ps(x + i); n2 = _mm_loadu_ps(y + i); n1 = _mm_mul_ps(n1, n3); n2 = _mm_mul_ps(n2, n4); n1 = _mm_add_ps(n1, n2); _mm_storeu_ps(y + i, n1); } } /* Now deal with the last 1 to 3 elements, which don't fill an SSE2 register */ switch (n & 3) { case 3: y[n - 3] = y[n - 3]*LMS_LEAK_RATE + x[n - 3]*error; case 2: y[n - 2] = y[n - 2]*LMS_LEAK_RATE + x[n - 2]*error; case 1: y[n - 1] = y[n - 1]*LMS_LEAK_RATE + x[n - 1]*error; } } #else SPAN_DECLARE(void) vec_lmsf(const float x[], float y[], int n, float error) { int i; for (i = 0; i < n; i++) { /* Leak a little to tame uncontrolled wandering */ y[i] = y[i]*LMS_LEAK_RATE + x[i]*error; } } #endif /*- End of function --------------------------------------------------------*/ SPAN_DECLARE(void) vec_circular_lmsf(const float x[], float y[], int n, int pos, float error) { vec_lmsf(&x[pos], &y[0], n - pos, error); vec_lmsf(&x[0], &y[n - pos], pos, error); } /*- End of function --------------------------------------------------------*/ /*- End of file ------------------------------------------------------------*/