From 3590c846d4c2febbc05b4ad6b14a06edc549e453 Mon Sep 17 00:00:00 2001 From: "Manuel A. Fernandez Montecelo" Date: Fri, 27 May 2016 14:35:16 +0100 Subject: Imported Upstream version 0.9.6+really0.9.6 --- tests/test-float.c | 384 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 384 insertions(+) create mode 100644 tests/test-float.c (limited to 'tests/test-float.c') diff --git a/tests/test-float.c b/tests/test-float.c new file mode 100644 index 00000000..8b621336 --- /dev/null +++ b/tests/test-float.c @@ -0,0 +1,384 @@ +/* Test of substitute. + Copyright (C) 2011-2015 Free Software Foundation, Inc. + + 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 . */ + +/* Written by Bruno Haible , 2011. */ + +#include + +#include + +#include "fpucw.h" +#include "macros.h" + +/* Check that FLT_RADIX is a constant expression. */ +int a[] = { FLT_RADIX }; + +#if FLT_RADIX == 2 + +/* Return 2^n. */ +static float +pow2f (int n) +{ + int k = n; + volatile float x = 1; + volatile float y = 2; + /* Invariant: 2^n == x * y^k. */ + if (k < 0) + { + y = 0.5f; + k = - k; + } + while (k > 0) + { + if (k != 2 * (k / 2)) + { + x = x * y; + k = k - 1; + } + if (k == 0) + break; + y = y * y; + k = k / 2; + } + /* Now k == 0, hence x == 2^n. */ + return x; +} + +/* Return 2^n. */ +static double +pow2d (int n) +{ + int k = n; + volatile double x = 1; + volatile double y = 2; + /* Invariant: 2^n == x * y^k. */ + if (k < 0) + { + y = 0.5; + k = - k; + } + while (k > 0) + { + if (k != 2 * (k / 2)) + { + x = x * y; + k = k - 1; + } + if (k == 0) + break; + y = y * y; + k = k / 2; + } + /* Now k == 0, hence x == 2^n. */ + return x; +} + +/* Return 2^n. */ +static long double +pow2l (int n) +{ + int k = n; + volatile long double x = 1; + volatile long double y = 2; + /* Invariant: 2^n == x * y^k. */ + if (k < 0) + { + y = 0.5L; + k = - k; + } + while (k > 0) + { + if (k != 2 * (k / 2)) + { + x = x * y; + k = k - 1; + } + if (k == 0) + break; + y = y * y; + k = k / 2; + } + /* Now k == 0, hence x == 2^n. */ + return x; +} + +/* ----------------------- Check macros for 'float' ----------------------- */ + +/* Check that the FLT_* macros expand to constant expressions. */ +int fb[] = + { + FLT_MANT_DIG, FLT_MIN_EXP, FLT_MAX_EXP, + FLT_DIG, FLT_MIN_10_EXP, FLT_MAX_10_EXP + }; +float fc[] = { FLT_EPSILON, FLT_MIN, FLT_MAX }; + +static void +test_float (void) +{ + /* Check that the value of FLT_MIN_EXP is well parenthesized. */ + ASSERT ((FLT_MIN_EXP % 101111) == (FLT_MIN_EXP) % 101111); + + /* Check that the value of DBL_MIN_10_EXP is well parenthesized. */ + ASSERT ((FLT_MIN_10_EXP % 101111) == (FLT_MIN_10_EXP) % 101111); + + /* Check that 'float' is as specified in IEEE 754. */ + ASSERT (FLT_MANT_DIG == 24); + ASSERT (FLT_MIN_EXP == -125); + ASSERT (FLT_MAX_EXP == 128); + + /* Check the value of FLT_MIN_10_EXP. */ + ASSERT (FLT_MIN_10_EXP == - (int) (- (FLT_MIN_EXP - 1) * 0.30103)); + + /* Check the value of FLT_DIG. */ + ASSERT (FLT_DIG == (int) ((FLT_MANT_DIG - 1) * 0.30103)); + + /* Check the value of FLT_MIN_10_EXP. */ + ASSERT (FLT_MIN_10_EXP == - (int) (- (FLT_MIN_EXP - 1) * 0.30103)); + + /* Check the value of FLT_MAX_10_EXP. */ + ASSERT (FLT_MAX_10_EXP == (int) (FLT_MAX_EXP * 0.30103)); + + /* Check the value of FLT_MAX. */ + { + volatile float m = FLT_MAX; + int n; + + ASSERT (m + m > m); + for (n = 0; n <= 2 * FLT_MANT_DIG; n++) + { + volatile float pow2_n = pow2f (n); /* 2^n */ + volatile float x = m + (m / pow2_n); + if (x > m) + ASSERT (x + x == x); + else + ASSERT (!(x + x == x)); + } + } + + /* Check the value of FLT_MIN. */ + { + volatile float m = FLT_MIN; + volatile float x = pow2f (FLT_MIN_EXP - 1); + ASSERT (m == x); + } + + /* Check the value of FLT_EPSILON. */ + { + volatile float e = FLT_EPSILON; + volatile float me; + int n; + + me = 1.0f + e; + ASSERT (me > 1.0f); + ASSERT (me - 1.0f == e); + for (n = 0; n <= 2 * FLT_MANT_DIG; n++) + { + volatile float half_n = pow2f (- n); /* 2^-n */ + volatile float x = me - half_n; + if (x < me) + ASSERT (x <= 1.0f); + } + } +} + +/* ----------------------- Check macros for 'double' ----------------------- */ + +/* Check that the DBL_* macros expand to constant expressions. */ +int db[] = + { + DBL_MANT_DIG, DBL_MIN_EXP, DBL_MAX_EXP, + DBL_DIG, DBL_MIN_10_EXP, DBL_MAX_10_EXP + }; +double dc[] = { DBL_EPSILON, DBL_MIN, DBL_MAX }; + +static void +test_double (void) +{ + /* Check that the value of DBL_MIN_EXP is well parenthesized. */ + ASSERT ((DBL_MIN_EXP % 101111) == (DBL_MIN_EXP) % 101111); + + /* Check that the value of DBL_MIN_10_EXP is well parenthesized. */ + ASSERT ((DBL_MIN_10_EXP % 101111) == (DBL_MIN_10_EXP) % 101111); + + /* Check that 'double' is as specified in IEEE 754. */ + ASSERT (DBL_MANT_DIG == 53); + ASSERT (DBL_MIN_EXP == -1021); + ASSERT (DBL_MAX_EXP == 1024); + + /* Check the value of DBL_MIN_10_EXP. */ + ASSERT (DBL_MIN_10_EXP == - (int) (- (DBL_MIN_EXP - 1) * 0.30103)); + + /* Check the value of DBL_DIG. */ + ASSERT (DBL_DIG == (int) ((DBL_MANT_DIG - 1) * 0.30103)); + + /* Check the value of DBL_MIN_10_EXP. */ + ASSERT (DBL_MIN_10_EXP == - (int) (- (DBL_MIN_EXP - 1) * 0.30103)); + + /* Check the value of DBL_MAX_10_EXP. */ + ASSERT (DBL_MAX_10_EXP == (int) (DBL_MAX_EXP * 0.30103)); + + /* Check the value of DBL_MAX. */ + { + volatile double m = DBL_MAX; + int n; + + ASSERT (m + m > m); + for (n = 0; n <= 2 * DBL_MANT_DIG; n++) + { + volatile double pow2_n = pow2d (n); /* 2^n */ + volatile double x = m + (m / pow2_n); + if (x > m) + ASSERT (x + x == x); + else + ASSERT (!(x + x == x)); + } + } + + /* Check the value of DBL_MIN. */ + { + volatile double m = DBL_MIN; + volatile double x = pow2d (DBL_MIN_EXP - 1); + ASSERT (m == x); + } + + /* Check the value of DBL_EPSILON. */ + { + volatile double e = DBL_EPSILON; + volatile double me; + int n; + + me = 1.0 + e; + ASSERT (me > 1.0); + ASSERT (me - 1.0 == e); + for (n = 0; n <= 2 * DBL_MANT_DIG; n++) + { + volatile double half_n = pow2d (- n); /* 2^-n */ + volatile double x = me - half_n; + if (x < me) + ASSERT (x <= 1.0); + } + } +} + +/* -------------------- Check macros for 'long double' -------------------- */ + +/* Check that the LDBL_* macros expand to constant expressions. */ +int lb[] = + { + LDBL_MANT_DIG, LDBL_MIN_EXP, LDBL_MAX_EXP, + LDBL_DIG, LDBL_MIN_10_EXP, LDBL_MAX_10_EXP + }; +long double lc1 = LDBL_EPSILON; +long double lc2 = LDBL_MIN; +#if 0 /* LDBL_MAX is not a constant expression on some platforms. */ +long double lc3 = LDBL_MAX; +#endif + +static void +test_long_double (void) +{ + /* Check that the value of LDBL_MIN_EXP is well parenthesized. */ + ASSERT ((LDBL_MIN_EXP % 101111) == (LDBL_MIN_EXP) % 101111); + + /* Check that the value of LDBL_MIN_10_EXP is well parenthesized. */ + ASSERT ((LDBL_MIN_10_EXP % 101111) == (LDBL_MIN_10_EXP) % 101111); + + /* Check that 'long double' is at least as wide as 'double'. */ + ASSERT (LDBL_MANT_DIG >= DBL_MANT_DIG); + ASSERT (LDBL_MIN_EXP - LDBL_MANT_DIG <= DBL_MIN_EXP - DBL_MANT_DIG); + ASSERT (LDBL_MAX_EXP >= DBL_MAX_EXP); + + /* Check the value of LDBL_DIG. */ + ASSERT (LDBL_DIG == (int)((LDBL_MANT_DIG - 1) * 0.30103)); + + /* Check the value of LDBL_MIN_10_EXP. */ + ASSERT (LDBL_MIN_10_EXP == - (int) (- (LDBL_MIN_EXP - 1) * 0.30103)); + + /* Check the value of LDBL_MAX_10_EXP. */ + ASSERT (LDBL_MAX_10_EXP == (int) (LDBL_MAX_EXP * 0.30103)); + + /* Check the value of LDBL_MAX. */ + { + volatile long double m = LDBL_MAX; + int n; + + ASSERT (m + m > m); + for (n = 0; n <= 2 * LDBL_MANT_DIG; n++) + { + volatile long double pow2_n = pow2l (n); /* 2^n */ + volatile long double x = m + (m / pow2_n); + if (x > m) + ASSERT (x + x == x); + else + ASSERT (!(x + x == x)); + } + } + + /* Check the value of LDBL_MIN. */ + { + volatile long double m = LDBL_MIN; + volatile long double x = pow2l (LDBL_MIN_EXP - 1); + ASSERT (m == x); + } + + /* Check the value of LDBL_EPSILON. */ + { + volatile long double e = LDBL_EPSILON; + volatile long double me; + int n; + + me = 1.0L + e; + ASSERT (me > 1.0L); + ASSERT (me - 1.0L == e); + for (n = 0; n <= 2 * LDBL_MANT_DIG; n++) + { + volatile long double half_n = pow2l (- n); /* 2^-n */ + volatile long double x = me - half_n; + if (x < me) + ASSERT (x <= 1.0L); + } + } +} + +int +main () +{ + test_float (); + test_double (); + + { + DECL_LONG_DOUBLE_ROUNDING + + BEGIN_LONG_DOUBLE_ROUNDING (); + + test_long_double (); + + END_LONG_DOUBLE_ROUNDING (); + } + + return 0; +} + +#else + +int +main () +{ + fprintf (stderr, "Skipping test: FLT_RADIX is not 2.\n"); + return 77; +} + +#endif -- cgit v1.2.3