Home | History | Annotate | Download | only in arm-neon-tests
      1 /*
      2 
      3 Copyright (c) 2009, 2010, 2011, 2013 STMicroelectronics
      4 Written by Christophe Lyon
      5 
      6 Permission is hereby granted, free of charge, to any person obtaining a copy
      7 of this software and associated documentation files (the "Software"), to deal
      8 in the Software without restriction, including without limitation the rights
      9 to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
     10 copies of the Software, and to permit persons to whom the Software is
     11 furnished to do so, subject to the following conditions:
     12 
     13 The above copyright notice and this permission notice shall be included in
     14 all copies or substantial portions of the Software.
     15 
     16 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
     17 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
     18 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
     19 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
     20 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
     21 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
     22 THE SOFTWARE.
     23 
     24 */
     25 
     26 #if defined(__arm__) || defined(__aarch64__)
     27 #include <arm_neon.h>
     28 #else
     29 #include "stm-arm-neon.h"
     30 #endif
     31 
     32 #include "stm-arm-neon-ref.h"
     33 #include <math.h>
     34 
     35 
     36 #define FNNAME1(NAME) void exec_ ## NAME (void)
     37 #define FNNAME(NAME) FNNAME1(NAME)
     38 
     39 FNNAME (INSN_NAME)
     40 {
     41   int i;
     42 
     43   /* Basic test: y=vcomp(x,x), then store the result.  */
     44 #define TEST_VCOMP1(INSN, Q, T1, T2, T3, W, N)				\
     45   VECT_VAR(vector_res, T3, W, N) =					\
     46     INSN##Q##_##T2##W(VECT_VAR(vector, T1, W, N),			\
     47 		      VECT_VAR(vector2, T1, W, N));			\
     48   vst1##Q##_u##W(VECT_VAR(result, T3, W, N), VECT_VAR(vector_res, T3, W, N))
     49 
     50 #define TEST_VCOMP(INSN, Q, T1, T2, T3, W, N)				\
     51   TEST_VCOMP1(INSN, Q, T1, T2, T3, W, N)
     52 
     53   /* With ARM RVCT, we need to declare variables before any executable
     54      statement */
     55   /* No need for 64 bits elements */
     56   DECL_VARIABLE(vector, int, 8, 8);
     57   DECL_VARIABLE(vector, int, 16, 4);
     58   DECL_VARIABLE(vector, int, 32, 2);
     59   DECL_VARIABLE(vector, uint, 8, 8);
     60   DECL_VARIABLE(vector, uint, 16, 4);
     61   DECL_VARIABLE(vector, uint, 32, 2);
     62   DECL_VARIABLE(vector, float, 32, 2);
     63   DECL_VARIABLE(vector, int, 8, 16);
     64   DECL_VARIABLE(vector, int, 16, 8);
     65   DECL_VARIABLE(vector, int, 32, 4);
     66   DECL_VARIABLE(vector, uint, 8, 16);
     67   DECL_VARIABLE(vector, uint, 16, 8);
     68   DECL_VARIABLE(vector, uint, 32, 4);
     69   DECL_VARIABLE(vector, float, 32, 4);
     70 
     71   DECL_VARIABLE(vector2, int, 8, 8);
     72   DECL_VARIABLE(vector2, int, 16, 4);
     73   DECL_VARIABLE(vector2, int, 32, 2);
     74   DECL_VARIABLE(vector2, uint, 8, 8);
     75   DECL_VARIABLE(vector2, uint, 16, 4);
     76   DECL_VARIABLE(vector2, uint, 32, 2);
     77   DECL_VARIABLE(vector2, float, 32, 2);
     78   DECL_VARIABLE(vector2, int, 8, 16);
     79   DECL_VARIABLE(vector2, int, 16, 8);
     80   DECL_VARIABLE(vector2, int, 32, 4);
     81   DECL_VARIABLE(vector2, uint, 8, 16);
     82   DECL_VARIABLE(vector2, uint, 16, 8);
     83   DECL_VARIABLE(vector2, uint, 32, 4);
     84   DECL_VARIABLE(vector2, float, 32, 4);
     85 
     86   DECL_VARIABLE(vector_res, uint, 8, 8);
     87   DECL_VARIABLE(vector_res, uint, 16, 4);
     88   DECL_VARIABLE(vector_res, uint, 32, 2);
     89   DECL_VARIABLE(vector_res, uint, 8, 16);
     90   DECL_VARIABLE(vector_res, uint, 16, 8);
     91   DECL_VARIABLE(vector_res, uint, 32, 4);
     92 
     93   clean_results ();
     94 
     95   /* There is no 64 bits variant, we can't use the generic initializer  */
     96   VLOAD(vector, buffer, , int, s, 8, 8);
     97   VLOAD(vector, buffer, , int, s, 16, 4);
     98   VLOAD(vector, buffer, , int, s, 32, 2);
     99   VLOAD(vector, buffer, , uint, u, 8, 8);
    100   VLOAD(vector, buffer, , uint, u, 16, 4);
    101   VLOAD(vector, buffer, , uint, u, 32, 2);
    102   VLOAD(vector, buffer, , float, f, 32, 2);
    103 
    104   VLOAD(vector, buffer, q, int, s, 8, 16);
    105   VLOAD(vector, buffer, q, int, s, 16, 8);
    106   VLOAD(vector, buffer, q, int, s, 32, 4);
    107   VLOAD(vector, buffer, q, uint, u, 8, 16);
    108   VLOAD(vector, buffer, q, uint, u, 16, 8);
    109   VLOAD(vector, buffer, q, uint, u, 32, 4);
    110   VLOAD(vector, buffer, q, float, f, 32, 4);
    111 
    112   /* Choose init value arbitrarily, will be used for vector
    113      comparison */
    114   VDUP(vector2, , int, s, 8, 8, -10);
    115   VDUP(vector2, , int, s, 16, 4, -14);
    116   VDUP(vector2, , int, s, 32, 2, -16);
    117   VDUP(vector2, , uint, u, 8, 8, 0xF3);
    118   VDUP(vector2, , uint, u, 16, 4, 0xFFF2);
    119   VDUP(vector2, , uint, u, 32, 2, 0xFFFFFFF1);
    120   VDUP(vector2, , float, f, 32, 2, -15.0f);
    121 
    122   VDUP(vector2, q, int, s, 8, 16, -4);
    123   VDUP(vector2, q, int, s, 16, 8, -10);
    124   VDUP(vector2, q, int, s, 32, 4, -14);
    125   VDUP(vector2, q, uint, u, 8, 16, 0xF4);
    126   VDUP(vector2, q, uint, u, 16, 8, 0xFFF6);
    127   VDUP(vector2, q, uint, u, 32, 4, 0xFFFFFFF2);
    128   VDUP(vector2, q, float, f, 32, 4, -14.0f);
    129 
    130   /* The same result buffers are used multiple times, so output them
    131      before overwriting them  */
    132   fprintf(ref_file, "\n%s output:\n", TEST_MSG);
    133   fprintf(gcc_tests_file, "\n%s output:\n", TEST_MSG);
    134   TEST_VCOMP(INSN_NAME, , int, s, uint, 8, 8);
    135   TEST_VCOMP(INSN_NAME, , int, s, uint, 16, 4);
    136   TEST_VCOMP(INSN_NAME, , int, s, uint, 32, 2);
    137   DUMP(TEST_MSG, uint, 8, 8, PRIx8);
    138   DUMP(TEST_MSG, uint, 16, 4, PRIx16);
    139   DUMP(TEST_MSG, uint, 32, 2, PRIx32);
    140 
    141   TEST_VCOMP(INSN_NAME, , uint, u, uint, 8, 8);
    142   TEST_VCOMP(INSN_NAME, , uint, u, uint, 16, 4);
    143   TEST_VCOMP(INSN_NAME, , uint, u, uint, 32, 2);
    144   DUMP(TEST_MSG, uint, 8, 8, PRIx8);
    145   DUMP(TEST_MSG, uint, 16, 4, PRIx16);
    146   DUMP(TEST_MSG, uint, 32, 2, PRIx32);
    147 
    148   TEST_VCOMP(INSN_NAME, q, int, s, uint, 8, 16);
    149   TEST_VCOMP(INSN_NAME, q, int, s, uint, 16, 8);
    150   TEST_VCOMP(INSN_NAME, q, int, s, uint, 32, 4);
    151   DUMP(TEST_MSG, uint, 8, 16, PRIx8);
    152   DUMP(TEST_MSG, uint, 16, 8, PRIx16);
    153   DUMP(TEST_MSG, uint, 32, 4, PRIx32);
    154 
    155   TEST_VCOMP(INSN_NAME, q, uint, u, uint, 8, 16);
    156   TEST_VCOMP(INSN_NAME, q, uint, u, uint, 16, 8);
    157   TEST_VCOMP(INSN_NAME, q, uint, u, uint, 32, 4);
    158   DUMP(TEST_MSG, uint, 8, 16, PRIx8);
    159   DUMP(TEST_MSG, uint, 16, 8, PRIx16);
    160   DUMP(TEST_MSG, uint, 32, 4, PRIx32);
    161 
    162   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    163   DUMP(TEST_MSG, uint, 32, 2, PRIx32);
    164 
    165   TEST_VCOMP(INSN_NAME, q, float, f, uint, 32, 4);
    166   DUMP(TEST_MSG, uint, 32, 4, PRIx32);
    167 
    168   /* Extra tests to have 100% coverage on all the variants */
    169   VDUP(vector2, , uint, u, 32, 2, 0xFFFFFFF0);
    170   TEST_VCOMP(INSN_NAME, , uint, u, uint, 32, 2);
    171   DUMP(TEST_MSG, uint, 32, 2, PRIx32);
    172 
    173   VDUP(vector2, , int, s, 32, 2, -15);
    174   TEST_VCOMP(INSN_NAME, , int, s, uint, 32, 2);
    175   DUMP(TEST_MSG, uint, 32, 2, PRIx32);
    176 
    177   VDUP(vector2, , float, f, 32, 2, -16.0f);
    178   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    179   DUMP(TEST_MSG, uint, 32, 2, PRIx32);
    180 
    181 
    182   /* Extra FP tests with special values (NaN, ....) */
    183   VDUP(vector, , float, f, 32, 2, 1.0);
    184   VDUP(vector2, , float, f, 32, 2, NAN);
    185   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    186   DUMP(TEST_MSG " FP special (NaN)", uint, 32, 2, PRIx32);
    187 
    188   VDUP(vector, , float, f, 32, 2, 1.0);
    189   VDUP(vector2, , float, f, 32, 2, -NAN);
    190   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    191   DUMP(TEST_MSG " FP special (-NaN)", uint, 32, 2, PRIx32);
    192 
    193   VDUP(vector, , float, f, 32, 2, NAN);
    194   VDUP(vector2, , float, f, 32, 2, 1.0);
    195   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    196   DUMP(TEST_MSG " FP special (NaN)", uint, 32, 2, PRIx32);
    197 
    198   VDUP(vector, , float, f, 32, 2, 1.0);
    199   VDUP(vector2, , float, f, 32, 2, HUGE_VALF);
    200   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    201   DUMP(TEST_MSG " FP special (inf)", uint, 32, 2, PRIx32);
    202 
    203   VDUP(vector, , float, f, 32, 2, 1.0);
    204   VDUP(vector2, , float, f, 32, 2, -HUGE_VALF);
    205   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    206   DUMP(TEST_MSG " FP special (-inf)", uint, 32, 2, PRIx32);
    207 
    208   VDUP(vector, , float, f, 32, 2, HUGE_VALF);
    209   VDUP(vector2, , float, f, 32, 2, 1.0);
    210   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    211   DUMP(TEST_MSG " FP special (inf)", uint, 32, 2, PRIx32);
    212 
    213   VDUP(vector, , float, f, 32, 2, -0.0);
    214   VDUP(vector2, , float, f, 32, 2, 0.0);
    215   TEST_VCOMP(INSN_NAME, , float, f, uint, 32, 2);
    216   DUMP(TEST_MSG " FP special (-0.0)", uint, 32, 2, PRIx32);
    217 
    218 #ifdef EXTRA_TESTS
    219   EXTRA_TESTS();
    220 #endif
    221 }
    222