Update SoftFloat

[riscv-isa-sim.git] / softfloat / s_approxRecipSqrt32_1.c

diff --git a/softfloat/s_approxRecipSqrt32_1.c b/softfloat/s_approxRecipSqrt32_1.c
index a3186739cd4b2a929e7d5e4f291bc3d0c719865a..2ab71a25a611ef88d558b7b68504b738dbf3fe8a 100644 (file)
--- a/softfloat/s_approxRecipSqrt32_1.c
+++ b/softfloat/s_approxRecipSqrt32_1.c
@@ -2,9 +2,9 @@
 /*============================================================================
 
 This C source file is part of the SoftFloat IEEE Floating-Point Arithmetic
 /*============================================================================
 
 This C source file is part of the SoftFloat IEEE Floating-Point Arithmetic

-Package, Release 3a, by John R. Hauser.
+Package, Release 3d, by John R. Hauser.

 
 

-Copyright 2011, 2012, 2013, 2014, 2015 The Regents of the University of
+Copyright 2011, 2012, 2013, 2014, 2015, 2016 The Regents of the University of

 California.  All rights reserved.
 
 Redistribution and use in source and binary forms, with or without
 California.  All rights reserved.
 
 Redistribution and use in source and binary forms, with or without


@@ -39,34 +39,30 @@ SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 #ifndef softfloat_approxRecipSqrt32_1
 
 
 #ifndef softfloat_approxRecipSqrt32_1
 

+extern const uint16_t softfloat_approxRecipSqrt_1k0s[];
+extern const uint16_t softfloat_approxRecipSqrt_1k1s[];
+

 uint32_t softfloat_approxRecipSqrt32_1( unsigned int oddExpA, uint32_t a )
 {
     int index;
 uint32_t softfloat_approxRecipSqrt32_1( unsigned int oddExpA, uint32_t a )
 {
     int index;

-    uint16_t eps;
-    static const uint16_t k0s[] = {
-        0xB4C9, 0xFFAB, 0xAA7D, 0xF11C, 0xA1C5, 0xE4C7, 0x9A43, 0xDA29,
-        0x93B5, 0xD0E5, 0x8DED, 0xC8B7, 0x88C6, 0xC16D, 0x8424, 0xBAE1
-    };
-    static const uint16_t k1s[] = {
-        0xA5A5, 0xEA42, 0x8C21, 0xC62D, 0x788F, 0xAA7F, 0x6928, 0x94B6,
-        0x5CC7, 0x8335, 0x52A6, 0x74E2, 0x4A3E, 0x68FE, 0x432B, 0x5EFD
-    };
-    uint16_t r0;
+    uint16_t eps, r0;

     uint_fast32_t ESqrR0;
     uint_fast32_t ESqrR0;

-    uint32_t delta0;
+    uint32_t sigma0;

     uint_fast32_t r;
     uint_fast32_t r;

-    uint32_t sqrDelta0;
+    uint32_t sqrSigma0;

 
     index = (a>>27 & 0xE) + oddExpA;
     eps = (uint16_t) (a>>12);
 
     index = (a>>27 & 0xE) + oddExpA;
     eps = (uint16_t) (a>>12);

-    r0 = k0s[index] - ((k1s[index] * (uint_fast32_t) eps)>>20);
+    r0 = softfloat_approxRecipSqrt_1k0s[index]
+             - ((softfloat_approxRecipSqrt_1k1s[index] * (uint_fast32_t) eps)
+                    >>20);

     ESqrR0 = (uint_fast32_t) r0 * r0;
     if ( ! oddExpA ) ESqrR0 <<= 1;
     ESqrR0 = (uint_fast32_t) r0 * r0;
     if ( ! oddExpA ) ESqrR0 <<= 1;

-    delta0 = ~(uint_fast32_t) (((uint32_t) ESqrR0 * (uint_fast64_t) a)>>23);
-    r = ((uint_fast32_t) r0<<16) + ((r0 * (uint_fast64_t) delta0)>>25);
-    sqrDelta0 = ((uint_fast64_t) delta0 * delta0)>>32;
+    sigma0 = ~(uint_fast32_t) (((uint32_t) ESqrR0 * (uint_fast64_t) a)>>23);
+    r = ((uint_fast32_t) r0<<16) + ((r0 * (uint_fast64_t) sigma0)>>25);
+    sqrSigma0 = ((uint_fast64_t) sigma0 * sigma0)>>32;

     r += ((uint32_t) ((r>>1) + (r>>3) - ((uint_fast32_t) r0<<14))
     r += ((uint32_t) ((r>>1) + (r>>3) - ((uint_fast32_t) r0<<14))

-              * (uint_fast64_t) sqrDelta0)
+              * (uint_fast64_t) sqrSigma0)

              >>48;
     if ( ! (r & 0x80000000) ) r = 0x80000000;
     return r;
              >>48;
     if ( ! (r & 0x80000000) ) r = 0x80000000;
     return r;