src/gromacs/simd/tests/simd_floatingpoint.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 #ifdef HAVE_CONFIG_H
  36 #include <config.h>
  37 #endif
  38
  39 #include <math.h>
  40 #include "gromacs/math/utilities.h"
  41
  42 #include "simd.h"
  43
  44 namespace gmx
  45 {
  46 namespace test
  47 {
  48 namespace
  49 {
  50
  51 /*! \cond internal */
  52 /*! \addtogroup module_simd */
  53 /*! \{ */
  54
  55 #ifdef GMX_SIMD_HAVE_REAL
  56
  57 /*! \brief Test fixture for floating-point tests (identical to the generic \ref SimdTest) */
  58 typedef SimdTest SimdFloatingpointTest;
  59
  60 TEST_F(SimdFloatingpointTest, gmxSimdSetZeroR)
  61 {
  62     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(0.0), gmx_simd_setzero_r());
  63 }
  64
  65 TEST_F(SimdFloatingpointTest, gmxSimdSet1R)
  66 {
  67     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(1.0), gmx_simd_set1_r(1.0));
  68 }
  69
  70 TEST_F(SimdFloatingpointTest, gmxSimdLoad1R)
  71 {
  72     real r = 2.0;
  73     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(r), gmx_simd_load1_r(&r));
  74 }
  75
  76 TEST_F(SimdFloatingpointTest, gmxSimdAddR)
  77 {
  78     GMX_EXPECT_SIMD_REAL_EQ(rSimd_5_7_9,
  79                             gmx_simd_add_r(rSimd_1_2_3, rSimd_4_5_6)); // 1+4=5, 2+5=7, 3+6=9
  80 }
  81
  82 TEST_F(SimdFloatingpointTest, gmxSimdSubR)
  83 {
  84     GMX_EXPECT_SIMD_REAL_EQ(rSimd_4_5_6,
  85                             gmx_simd_sub_r(rSimd_5_7_9, rSimd_1_2_3)); // 5-1=4, 7-2=5, 9-3=6
  86 }
  87
  88 TEST_F(SimdFloatingpointTest, gmxSimdMulR)
  89 {
  90     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(4, 10, 18),
  91                             gmx_simd_mul_r(rSimd_1_2_3, rSimd_4_5_6));
  92 }
  93
  94 TEST_F(SimdFloatingpointTest, gmxSimdFmaddR)
  95 {
  96     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(11, 18, 27),
  97                             gmx_simd_fmadd_r(rSimd_1_2_3, rSimd_4_5_6, rSimd_7_8_9)); // 1*4+7, etc.
  98 }
  99
 100 TEST_F(SimdFloatingpointTest, gmxSimdFmsubR)
 101 {
 102     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(-3, 2, 9),
 103                             gmx_simd_fmsub_r(rSimd_1_2_3, rSimd_4_5_6, rSimd_7_8_9)); // 1*4-7, etc.
 104 }
 105
 106 TEST_F(SimdFloatingpointTest, gmxSimdFnmaddR)
 107 {
 108     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(3, -2, -9),
 109                             gmx_simd_fnmadd_r(rSimd_1_2_3, rSimd_4_5_6, rSimd_7_8_9)); // -1*4+7, etc.
 110 }
 111
 112 TEST_F(SimdFloatingpointTest, gmxSimdFnmsubR)
 113 {
 114     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(-11, -18, -27),
 115                             gmx_simd_fnmsub_r(rSimd_1_2_3, rSimd_4_5_6, rSimd_7_8_9)); // -1*4-7, etc.
 116 }
 117
 118 TEST_F(SimdFloatingpointTest, gmxSimdFabsR)
 119 {
 120     GMX_EXPECT_SIMD_REAL_EQ(rSimd_1_2_3, gmx_simd_fabs_r(rSimd_1_2_3));    // fabs(x)=x
 121     GMX_EXPECT_SIMD_REAL_EQ(rSimd_1_2_3, gmx_simd_fabs_r(rSimd_m1_m2_m3)); // fabs(-x)=x
 122 }
 123
 124 TEST_F(SimdFloatingpointTest, gmxSimdFnegR)
 125 {
 126     GMX_EXPECT_SIMD_REAL_EQ(rSimd_m1_m2_m3, gmx_simd_fneg_r(rSimd_1_2_3));    // fneg(x)=-x
 127     GMX_EXPECT_SIMD_REAL_EQ(rSimd_1_2_3,    gmx_simd_fneg_r(rSimd_m1_m2_m3)); // fneg(-x)=x
 128 }
 129
 130 #ifdef GMX_SIMD_HAVE_LOGICAL
 131 TEST_F(SimdFloatingpointTest, gmxSimdAndR)
 132 {
 133     GMX_EXPECT_SIMD_REAL_EQ(rSimd_Bits3, gmx_simd_and_r(rSimd_Bits1, rSimd_Bits2)); // Bits1 & Bits2 = Bits3
 134 }
 135
 136 TEST_F(SimdFloatingpointTest, gmxSimdAndnotR)
 137 {
 138     GMX_EXPECT_SIMD_REAL_EQ(rSimd_Bits4, gmx_simd_andnot_r(rSimd_Bits1, rSimd_Bits2)); // (~Bits1) & Bits2 = Bits3
 139 }
 140
 141 TEST_F(SimdFloatingpointTest, gmxSimdOrR)
 142 {
 143     GMX_EXPECT_SIMD_REAL_EQ(rSimd_Bits5, gmx_simd_or_r(rSimd_Bits1, rSimd_Bits2)); // Bits1 | Bits2 = Bits3
 144 }
 145
 146 TEST_F(SimdFloatingpointTest, gmxSimdXorR)
 147 {
 148     GMX_EXPECT_SIMD_REAL_EQ(rSimd_Bits6, gmx_simd_xor_r(rSimd_Bits1, rSimd_Bits2)); // Bits1 ^ Bits2 = Bits3
 149 }
 150 #endif
 151
 152 TEST_F(SimdFloatingpointTest, gmxSimdMaxR)
 153 {
 154     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(3, 2, 4), gmx_simd_max_r(rSimd_1_2_3, rSimd_3_1_4));
 155     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(3, 2, 4), gmx_simd_max_r(rSimd_3_1_4, rSimd_1_2_3));
 156     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(-1, -1, -3), gmx_simd_max_r(rSimd_m1_m2_m3, rSimd_m3_m1_m4));
 157     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(-1, -1, -3), gmx_simd_max_r(rSimd_m3_m1_m4, rSimd_m1_m2_m3));
 158 }
 159
 160 TEST_F(SimdFloatingpointTest, gmxSimdMinR)
 161 {
 162     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(1, 1, 3), gmx_simd_min_r(rSimd_1_2_3, rSimd_3_1_4));
 163     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(1, 1, 3), gmx_simd_min_r(rSimd_3_1_4, rSimd_1_2_3));
 164     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(-3, -2, -4), gmx_simd_min_r(rSimd_m1_m2_m3, rSimd_m3_m1_m4));
 165     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(-3, -2, -4), gmx_simd_min_r(rSimd_m3_m1_m4, rSimd_m1_m2_m3));
 166 }
 167
 168 TEST_F(SimdFloatingpointTest, gmxSimdRoundR)
 169 {
 170     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(2), gmx_simd_round_r(gmx_simd_set1_r(2.25)));
 171     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(4), gmx_simd_round_r(gmx_simd_set1_r(3.75)));
 172     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(-2), gmx_simd_round_r(gmx_simd_set1_r(-2.25)));
 173     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(-4), gmx_simd_round_r(gmx_simd_set1_r(-3.75)));
 174 }
 175
 176 TEST_F(SimdFloatingpointTest, gmxSimdTruncR)
 177 {
 178     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(2), gmx_simd_trunc_r(rSimd_2p25));
 179     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(3), gmx_simd_trunc_r(rSimd_3p75));
 180     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(-2), gmx_simd_trunc_r(rSimd_m2p25));
 181     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(-3), gmx_simd_trunc_r(rSimd_m3p75));
 182 }
 183
 184 TEST_F(SimdFloatingpointTest, gmxSimdFractionR)
 185 {
 186     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(0.25), gmx_simd_fraction_r(rSimd_2p25));   // fract(2.25)=0.25
 187     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(0.75), gmx_simd_fraction_r(rSimd_3p75));   // fract(3.75)=0.75
 188     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(-0.25), gmx_simd_fraction_r(rSimd_m2p25)); // fract(-2.25)=-0.25
 189     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom1R(-0.75), gmx_simd_fraction_r(rSimd_m3p75)); // fract(-3.75)=-0.75
 190 }
 191
 192 // We explicitly test the exponent/mantissa routines with double precision data,
 193 // since these usually rely on direct manipulation and shift of the SIMD registers,
 194 // where it is easy to make mistakes with single vs double precision.
 195
 196 TEST_F(SimdFloatingpointTest, gmxSimdGetExponentR)
 197 {
 198     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(60.0, -41.0, 54.0), gmx_simd_get_exponent_r(rSimd_Exp));
 199 #if (defined GMX_SIMD_HAVE_DOUBLE) && (defined GMX_DOUBLE)
 200     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(587.0, -462.0, 672.0), gmx_simd_get_exponent_r(rSimd_ExpDouble));
 201 #endif
 202 }
 203
 204 TEST_F(SimdFloatingpointTest, gmxSimdGetMantissaR)
 205 {
 206     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(1.219097320577810839026256,
 207                                               1.166738027848349235071623,
 208                                               1.168904015004464724825084), gmx_simd_get_mantissa_r(rSimd_Exp));
 209 #if (defined GMX_SIMD_HAVE_DOUBLE) && (defined GMX_DOUBLE)
 210     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(1.241261238952345623563251,
 211                                               1.047294723759123852359232,
 212                                               1.856066204750275957395734), gmx_simd_get_mantissa_r(rSimd_ExpDouble));
 213 #endif
 214 }
 215
 216 TEST_F(SimdFloatingpointTest, gmxSimdSetExponentR)
 217 {
 218     gmx_simd_real_t x0 = setSimdRealFrom3R(0.5, 11.5, 99.5);
 219     gmx_simd_real_t x1 = setSimdRealFrom3R(-0.5, -11.5, -99.5);
 220
 221     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(pow(2.0, 60.0), pow(2.0, -41.0), pow(2.0, 54.0)),
 222                             gmx_simd_set_exponent_r(setSimdRealFrom3R(60.0, -41.0, 54.0)));
 223 #if (defined GMX_SIMD_HAVE_DOUBLE) && (defined GMX_DOUBLE)
 224     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(pow(2.0, 587.0), pow(2.0, -462.0), pow(2.0, 672.0)),
 225                             gmx_simd_set_exponent_r(setSimdRealFrom3R(587.0, -462.0, 672.0)));
 226 #endif
 227     /* Rounding mode in gmx_simd_set_exponent_r() must be consistent with gmx_simd_round_r() */
 228     GMX_EXPECT_SIMD_REAL_EQ(gmx_simd_set_exponent_r(gmx_simd_round_r(x0)), gmx_simd_set_exponent_r(x0));
 229     GMX_EXPECT_SIMD_REAL_EQ(gmx_simd_set_exponent_r(gmx_simd_round_r(x1)), gmx_simd_set_exponent_r(x1));
 230 }
 231
 232 /*
 233  * We do extensive 1/sqrt(x) and 1/x accuracy testing in the math module, so
 234  * we just make sure the lookup instructions appear to work here
 235  */
 236
 237 TEST_F(SimdFloatingpointTest, gmxSimdRsqrtR)
 238 {
 239     gmx_simd_real_t x      = setSimdRealFrom3R(4.0, M_PI, 1234567890.0);
 240     gmx_simd_real_t ref    = setSimdRealFrom3R(0.5, 1.0/sqrt(M_PI), 1.0/sqrt(1234567890.0));
 241
 242     /* Set the allowed ulp error as 2 to the power of the number of bits in
 243      * the mantissa that do not have to be correct after the table lookup.
 244      */
 245     setUlpTol(1LL << (std::numeric_limits<real>::digits-GMX_SIMD_RSQRT_BITS));
 246
 247     GMX_EXPECT_SIMD_REAL_NEAR(ref, gmx_simd_rsqrt_r(x));
 248 }
 249
 250 TEST_F(SimdFloatingpointTest, gmxSimdRcpR)
 251 {
 252     gmx_simd_real_t x      = setSimdRealFrom3R(4.0, M_PI, 1234567890.0);
 253     gmx_simd_real_t ref    = setSimdRealFrom3R(0.25, 1.0/M_PI, 1.0/1234567890.0);
 254
 255     /* Set the allowed ulp error as 2 to the power of the number of bits in
 256      * the mantissa that do not have to be correct after the table lookup.
 257      */
 258     setUlpTol(1LL << (std::numeric_limits<real>::digits-GMX_SIMD_RCP_BITS));
 259
 260     GMX_EXPECT_SIMD_REAL_NEAR(ref, gmx_simd_rcp_r(x));
 261 }
 262
 263 TEST_F(SimdFloatingpointTest, gmxSimdBoolCmpEqAndBlendZeroR)
 264 {
 265     gmx_simd_bool_t eq   = gmx_simd_cmpeq_r(rSimd_5_7_9, rSimd_7_8_9);
 266     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(0, 0, 3), gmx_simd_blendzero_r(rSimd_1_2_3, eq));
 267 }
 268
 269 TEST_F(SimdFloatingpointTest, gmxSimdBlendNotZeroR)
 270 {
 271     gmx_simd_bool_t eq   = gmx_simd_cmpeq_r(rSimd_5_7_9, rSimd_7_8_9);
 272     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(1, 2, 0), gmx_simd_blendnotzero_r(rSimd_1_2_3, eq));
 273 }
 274
 275 TEST_F(SimdFloatingpointTest, gmxSimdBoolCmpLER)
 276 {
 277     gmx_simd_bool_t le   = gmx_simd_cmple_r(rSimd_5_7_9, rSimd_7_8_9);
 278     GMX_EXPECT_SIMD_REAL_EQ(rSimd_1_2_3, gmx_simd_blendzero_r(rSimd_1_2_3, le));
 279 }
 280
 281 TEST_F(SimdFloatingpointTest, gmxSimdBoolCmpLTR)
 282 {
 283     gmx_simd_bool_t lt   = gmx_simd_cmplt_r(rSimd_5_7_9, rSimd_7_8_9);
 284     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(1, 2, 0), gmx_simd_blendzero_r(rSimd_1_2_3, lt));
 285 }
 286
 287 TEST_F(SimdFloatingpointTest, gmxSimdBoolAndB)
 288 {
 289     gmx_simd_bool_t eq   = gmx_simd_cmpeq_r(rSimd_5_7_9, rSimd_7_8_9);
 290     gmx_simd_bool_t le   = gmx_simd_cmple_r(rSimd_5_7_9, rSimd_7_8_9);
 291     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(0, 0, 3), gmx_simd_blendzero_r(rSimd_1_2_3, gmx_simd_and_b(eq, le)));
 292 }
 293
 294 TEST_F(SimdFloatingpointTest, gmxSimdBoolOrB)
 295 {
 296     gmx_simd_bool_t eq   = gmx_simd_cmpeq_r(rSimd_5_7_9, rSimd_7_8_9);
 297     gmx_simd_bool_t lt   = gmx_simd_cmplt_r(rSimd_5_7_9, rSimd_7_8_9);
 298     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(1, 2, 3), gmx_simd_blendzero_r(rSimd_1_2_3, gmx_simd_or_b(eq, lt)));
 299 }
 300
 301 TEST_F(SimdFloatingpointTest, gmxSimdAnytrueB)
 302 {
 303     gmx_simd_bool_t eq;
 304
 305     /* this test is a bit tricky since we don't know the simd width.
 306      * We cannot check for truth values for "any" element beyond the first,
 307      * since that part of the data will not be used if simd width is 1.
 308      */
 309     eq = gmx_simd_cmpeq_r(rSimd_5_7_9, setSimdRealFrom3R(5, 0, 0));
 310     EXPECT_NE(0, gmx_simd_anytrue_b(eq));
 311
 312     eq = gmx_simd_cmpeq_r(rSimd_1_2_3, rSimd_4_5_6);
 313     EXPECT_EQ(0, gmx_simd_anytrue_b(eq));
 314 }
 315
 316 TEST_F(SimdFloatingpointTest, gmxSimdBlendvR)
 317 {
 318     gmx_simd_bool_t lt   = gmx_simd_cmplt_r(rSimd_5_7_9, rSimd_7_8_9);
 319     GMX_EXPECT_SIMD_REAL_EQ(setSimdRealFrom3R(4, 5, 3), gmx_simd_blendv_r(rSimd_1_2_3, rSimd_4_5_6, lt));
 320 }
 321
 322 TEST_F(SimdFloatingpointTest, gmxSimdReduceR)
 323 {
 324     // The horizontal sum of the SIMD variable depends on the width, so
 325     // simply store it an extra time and calculate what the sum should be
 326     std::vector<real> v   = simdReal2Vector(rSimd_4_5_6);
 327     real              sum = 0.0;
 328
 329     for (int i = 0; i < GMX_SIMD_REAL_WIDTH; i++)
 330     {
 331         sum += v[i];
 332     }
 333
 334     EXPECT_EQ(sum, gmx_simd_reduce_r(rSimd_4_5_6));
 335 }
 336
 337 #endif      // GMX_SIMD_HAVE_REAL
 338
 339 /*! \} */
 340 /*! \endcond */
 341
 342 }      // namespace
 343 }      // namespace
 344 }      // namespace