/* +--------------------------------------------------------------------------+ | CHStone : a suite of benchmark programs for C-based High-Level Synthesis | | ======================================================================== | | | | * Collected and Modified : Y. Hara, H. Tomiyama, S. Honda, | | H. Takada and K. Ishii | | Nagoya University, Japan | | | | * Remark : | | 1. This source code is modified to unify the formats of the benchmark | | programs in CHStone. | | 2. Test vectors are added for CHStone. | | 3. If "main_result" is 0 at the end of the program, the program is | | correctly executed. | | 4. Please follow the copyright of each benchmark program. | +--------------------------------------------------------------------------+ */ /*============================================================================ This C source fragment is part of the SoftFloat IEC/IEEE Floating-point Arithmetic Package, Release 2b. Written by John R. Hauser. This work was made possible in part by the International Computer Science Institute, located at Suite 600, 1947 Center Street, Berkeley, California 94704. Funding was partially provided by the National Science Foundation under grant MIP-9311980. The original version of this code was written as part of a project to build a fixed-point vector processor in collaboration with the University of California at Berkeley, overseen by Profs. Nelson Morgan and John Wawrzynek. More information is available through the Web page `http://www.cs.berkeley.edu/~jhauser/ arithmetic/SoftFloat.html'. THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE. Although reasonable effort has been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT TIMES RESULT IN INCORRECT BEHAVIOR. USE OF THIS SOFTWARE IS RESTRICTED TO PERSONS AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ALL LOSSES, COSTS, OR OTHER PROBLEMS THEY INCUR DUE TO THE SOFTWARE, AND WHO FURTHERMORE EFFECTIVELY INDEMNIFY JOHN HAUSER AND THE INTERNATIONAL COMPUTER SCIENCE INSTITUTE (possibly via similar legal notice) AGAINST ALL LOSSES, COSTS, OR OTHER PROBLEMS INCURRED BY THEIR CUSTOMERS AND CLIENTS DUE TO THE SOFTWARE. Derivative works are acceptable, even for commercial purposes, so long as (1) the source code for the derivative work includes prominent notice that the work is derivative, and (2) the source code includes prominent notice with these four paragraphs for those parts of this code that are retained. =============================================================================*/ /*---------------------------------------------------------------------------- | Shifts `a' right by the number of bits given in `count'. If any nonzero | bits are shifted off, they are ``jammed'' into the least significant bit of | the result by setting the least significant bit to 1. The value of `count' | can be arbitrarily large; in particular, if `count' is greater than 64, the | result will be either 0 or 1, depending on whether `a' is zero or nonzero. | The result is stored in the location pointed to by `zPtr'. *----------------------------------------------------------------------------*/ INLINE void shift64RightJamming (bits64 a, int16 count, bits64 * zPtr) { bits64 z; if (count == 0) { z = a; } else if (count < 64) { z = (a >> count) | ((a << ((-count) & 63)) != 0); } else { z = (a != 0); } *zPtr = z; } /*---------------------------------------------------------------------------- | Shifts the 128-bit value formed by concatenating `a0' and `a1' right by 64 | _plus_ the number of bits given in `count'. The shifted result is at most | 64 nonzero bits; this is stored at the location pointed to by `z0Ptr'. The | bits shifted off form a second 64-bit result as follows: The _last_ bit | shifted off is the most-significant bit of the extra result, and the other | 63 bits of the extra result are all zero if and only if _all_but_the_last_ | bits shifted off were all zero. This extra result is stored in the location | pointed to by `z1Ptr'. The value of `count' can be arbitrarily large. | (This routine makes more sense if `a0' and `a1' are considered to form | a fixed-point value with binary point between `a0' and `a1'. This fixed- | point value is shifted right by the number of bits given in `count', and | the integer part of the result is returned at the location pointed to by | `z0Ptr'. The fractional part of the result may be slightly corrupted as | described above, and is returned at the location pointed to by `z1Ptr'.) *----------------------------------------------------------------------------*/ INLINE void shift64ExtraRightJamming (bits64 a0, bits64 a1, int16 count, bits64 * z0Ptr, bits64 * z1Ptr) { bits64 z0, z1; int8 negCount; negCount = (-count) & 63; if (count == 0) { z1 = a1; z0 = a0; } else if (count < 64) { z1 = (a0 << negCount) | (a1 != 0); z0 = a0 >> count; } else { if (count == 64) { z1 = a0 | (a1 != 0); } else { z1 = ((a0 | a1) != 0); } z0 = 0; } *z1Ptr = z1; *z0Ptr = z0; } /*---------------------------------------------------------------------------- | Adds the 128-bit value formed by concatenating `a0' and `a1' to the 128-bit | value formed by concatenating `b0' and `b1'. Addition is modulo 2^128, so | any carry out is lost. The result is broken into two 64-bit pieces which | are stored at the locations pointed to by `z0Ptr' and `z1Ptr'. *----------------------------------------------------------------------------*/ INLINE void add128 (bits64 a0, bits64 a1, bits64 b0, bits64 b1, bits64 * z0Ptr, bits64 * z1Ptr) { bits64 z1; z1 = a1 + b1; *z1Ptr = z1; *z0Ptr = a0 + b0 + (z1 < a1); } /*---------------------------------------------------------------------------- | Subtracts the 128-bit value formed by concatenating `b0' and `b1' from the | 128-bit value formed by concatenating `a0' and `a1'. Subtraction is modulo | 2^128, so any borrow out (carry out) is lost. The result is broken into two | 64-bit pieces which are stored at the locations pointed to by `z0Ptr' and | `z1Ptr'. *----------------------------------------------------------------------------*/ INLINE void sub128 (bits64 a0, bits64 a1, bits64 b0, bits64 b1, bits64 * z0Ptr, bits64 * z1Ptr) { *z1Ptr = a1 - b1; *z0Ptr = a0 - b0 - (a1 < b1); } /*---------------------------------------------------------------------------- | Multiplies `a' by `b' to obtain a 128-bit product. The product is broken | into two 64-bit pieces which are stored at the locations pointed to by | `z0Ptr' and `z1Ptr'. *----------------------------------------------------------------------------*/ INLINE void mul64To128 (bits64 a, bits64 b, bits64 * z0Ptr, bits64 * z1Ptr) { bits32 aHigh, aLow, bHigh, bLow; bits64 z0, zMiddleA, zMiddleB, z1; aLow = a; aHigh = a >> 32; bLow = b; bHigh = b >> 32; z1 = ((bits64) aLow) * bLow; zMiddleA = ((bits64) aLow) * bHigh; zMiddleB = ((bits64) aHigh) * bLow; z0 = ((bits64) aHigh) * bHigh; zMiddleA += zMiddleB; z0 += (((bits64) (zMiddleA < zMiddleB)) << 32) + (zMiddleA >> 32); zMiddleA <<= 32; z1 += zMiddleA; z0 += (z1 < zMiddleA); *z1Ptr = z1; *z0Ptr = z0; } /*---------------------------------------------------------------------------- | Returns an approximation to the 64-bit integer quotient obtained by dividing | `b' into the 128-bit value formed by concatenating `a0' and `a1'. The | divisor `b' must be at least 2^63. If q is the exact quotient truncated | toward zero, the approximation returned lies between q and q + 2 inclusive. | If the exact quotient q is larger than 64 bits, the maximum positive 64-bit | unsigned integer is returned. *----------------------------------------------------------------------------*/ static bits64 estimateDiv128To64 (bits64 a0, bits64 a1, bits64 b) { bits64 b0, b1; bits64 rem0, rem1, term0, term1; bits64 z; if (b <= a0) return LIT64 (0xFFFFFFFFFFFFFFFF); b0 = b >> 32; z = (b0 << 32 <= a0) ? LIT64 (0xFFFFFFFF00000000) : (a0 / b0) << 32; mul64To128 (b, z, &term0, &term1); sub128 (a0, a1, term0, term1, &rem0, &rem1); while (((sbits64) rem0) < 0) { z -= LIT64 (0x100000000); b1 = b << 32; add128 (rem0, rem1, b0, b1, &rem0, &rem1); } rem0 = (rem0 << 32) | (rem1 >> 32); z |= (b0 << 32 <= rem0) ? 0xFFFFFFFF : rem0 / b0; return z; } /*---------------------------------------------------------------------------- | Returns the number of leading 0 bits before the most-significant 1 bit of | `a'. If `a' is zero, 32 is returned. *----------------------------------------------------------------------------*/ static int8 countLeadingZeros32 (bits32 a) { static const int8 countLeadingZerosHigh[256] = { 8, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; int8 shiftCount; shiftCount = 0; if (a < 0x10000) { shiftCount += 16; a <<= 16; } if (a < 0x1000000) { shiftCount += 8; a <<= 8; } shiftCount += countLeadingZerosHigh[a >> 24]; return shiftCount; } /*---------------------------------------------------------------------------- | Returns the number of leading 0 bits before the most-significant 1 bit of | `a'. If `a' is zero, 64 is returned. *----------------------------------------------------------------------------*/ static int8 countLeadingZeros64 (bits64 a) { int8 shiftCount; shiftCount = 0; if (a < ((bits64) 1) << 32) { shiftCount += 32; } else { a >>= 32; } shiftCount += countLeadingZeros32 (a); return shiftCount; }