4
// Copyright (c) 2001 - 2003, Intel Corporation
5
// All rights reserved.
7
// Contributed 2001 by the Intel Numerics Group, Intel Corporation
9
// Redistribution and use in source and binary forms, with or without
10
// modification, are permitted provided that the following conditions are
13
// * Redistributions of source code must retain the above copyright
14
// notice, this list of conditions and the following disclaimer.
16
// * Redistributions in binary form must reproduce the above copyright
17
// notice, this list of conditions and the following disclaimer in the
18
// documentation and/or other materials provided with the distribution.
20
// * The name of Intel Corporation may not be used to endorse or promote
21
// products derived from this software without specific prior written
24
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
25
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
26
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
27
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL OR ITS
28
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
29
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
30
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
31
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
32
// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY OR TORT (INCLUDING
33
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
34
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
36
// Intel Corporation is the author of this code, and requests that all
37
// problem reports or change requests be submitted to it directly at
38
// http://www.intel.com/software/products/opensource/libraries/num.htm.
41
//==============================================================
42
// 06/08/01 Initial version
43
// 08/23/01 Corrected error tag number
44
// 05/20/02 Cleaned up namespace and sf0 syntax
45
// 01/28/03 Improved performance
48
//==============================================================
49
// double fdim( double x, double y );
50
// input floating point f8, f9
51
// output floating point f8
54
// Overview of operation
55
//==============================================================
56
// fdim determines the positive difference between the arguments
57
// Result = x - y if x > y
60
// Error support is called if x-y overflows for x > y
64
//==============================================================
65
// General purpose registers: r14, r32 - r39
69
// r36-39 parameters for libm_error_support
76
GR_Parameter_RESULT = r38
77
GR_Parameter_TAG = r39
79
// Floating-point registers: f8 - f12
85
// Predicate registers: p6 - p10
89
GLOBAL_LIBM_ENTRY(fdim)
92
mov rExpBig = 0x103ff // Exponent to indicate overflow
93
fcmp.le.s1 p6,p7 = f8, f9 // Is x <= y?
98
fnorm.s1 fNormX = f8 // Save x
104
setf.exp fBig = rExpBig // Constant to test for overflow
105
fcmp.eq.s0 p8,p0 = f8, f9 // Dummy op to set Denormal or Invalid
112
fclass.m p9,p10 = f8, 0x1e3 // Test for x natval, nan, inf
119
(p6) fmerge.s f8 = f0, f0 // Result is +0 if x <= y
124
(p7) fms.d.s0 f8 = f8, f1, f9 // Result is x - y if x > y
131
(p10) fclass.m p9,p10 = f9, 0x1e3 // Test for y natval, nan, inf
138
(p10) fcmp.ge.s1 p8,p0 = f8, fBig // Test result for overflow
144
(p9) cmp.ne p8,p0 = r0,r0 // Clear p8 if x or y natval,nan,inf
145
(p8) br.cond.spnt FDIM_OVERFLOW // Branch if result overflows
146
br.ret.sptk b0 // Normal return
151
// Here if result will overflow
154
alloc r32=ar.pfs,2,2,4,0
155
fms.d.s0 f_tmp_result = f8,f1,f9 // Normalize result force overflow
159
mov GR_Parameter_TAG = 196 // Error code
161
br.cond.sptk __libm_error_region // Branch to error code
165
GLOBAL_LIBM_END(fdim)
168
LOCAL_LIBM_ENTRY(__libm_error_region)
169
// Call error support to report possible range error
173
add GR_Parameter_Y=-32,sp // Parameter 2 value
175
.save ar.pfs,GR_SAVE_PFS
176
mov GR_SAVE_PFS=ar.pfs // Save ar.pfs
180
add sp=-64,sp // Create new stack
182
mov GR_SAVE_GP=gp // Save gp
186
stfd [GR_Parameter_Y] = f9,16 // STORE Parameter 2 on stack
187
add GR_Parameter_X = 16,sp // Parameter 1 address
189
mov GR_SAVE_B0=b0 // Save b0
194
stfd [GR_Parameter_X] = fNormX // STORE Parameter 1 on stack
195
add GR_Parameter_RESULT = 0,GR_Parameter_Y // Parameter 3 address
199
stfd [GR_Parameter_Y] = f_tmp_result // STORE Parameter 3 on stack
200
add GR_Parameter_Y = -16,GR_Parameter_Y
201
br.call.sptk b0=__libm_error_support# // Call error handling function
205
add GR_Parameter_RESULT = 48,sp
211
ldfd f8 = [GR_Parameter_RESULT] // Get return result off stack
213
add sp = 64,sp // Restore stack pointer
214
mov b0 = GR_SAVE_B0 // Restore return address
218
mov gp = GR_SAVE_GP // Restore gp
219
mov ar.pfs = GR_SAVE_PFS // Restore ar.pfs
220
br.ret.sptk b0 // Return
223
LOCAL_LIBM_END(__libm_error_region)
226
.type __libm_error_support#,@function
227
.global __libm_error_support#