about summary refs log tree commit diff
path: root/sysdeps/ia64/fpu/s_libm_ldexpl.S
blob: 72d45602cfbb0006cdc771ea6cdbe9a892fef028 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
.file "libm_ldexpl.s"


// Copyright (c) 2000 - 2003, Intel Corporation
// All rights reserved.
//
// Contributed 2000 by the Intel Numerics Group, Intel Corporation
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// * Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// * The name of Intel Corporation may not be used to endorse or promote
// products derived from this software without specific prior written
// permission.

// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL OR ITS 
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY 
// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 
// 
// Intel Corporation is the author of this code, and requests that all
// problem reports or change requests be submitted to it directly at 
// http://www.intel.com/software/products/opensource/libraries/num.htm.
//
// History
//==============================================================
// 02/02/00 Initial version
// 01/26/01 ldexpl completely reworked and now standalone version 
// 01/04/02 Added handling for int 32 or 64 bits
// 05/20/02 Cleaned up namespace and sf0 syntax
// 02/10/03 Reordered header: .section, .global, .proc, .align
//
// API
//==============================================================
// long double __libm_ldexpl  (long double x, int n, int int_type) 
// input  floating point f8 and int n (r34), int int_type (r35)
// output floating point f8
//
// int_type = 0 if int is 32 bits
// int_type = 1 if int is 64 bits
//
// Returns x* 2**n using an fma and detects overflow
// and underflow.   
//
//

FR_Big         = f6
FR_NBig        = f7
FR_Floating_X  = f8
FR_Result      = f8
FR_Result2     = f9
FR_Result3     = f11
FR_Norm_X      = f12
FR_Two_N       = f14
FR_Two_to_Big  = f15

GR_N_Biased    = r15
GR_Big         = r16
GR_NBig        = r17
GR_Scratch     = r18
GR_Scratch1    = r19
GR_Bias        = r20
GR_N_as_int    = r21

GR_SAVE_B0          = r32
GR_SAVE_GP          = r33
GR_SAVE_PFS         = r34
GR_Parameter_X      = r35
GR_Parameter_Y      = r36
GR_Parameter_RESULT = r37
GR_Tag              = r38

.section .text
GLOBAL_LIBM_ENTRY(__libm_ldexpl)

//
//   Is x NAN, INF, ZERO, +-?
//   Build the exponent Bias
//
{    .mfi
     alloc         r32=ar.pfs,3,0,4,0
     fclass.m      p7,p0 = FR_Floating_X, 0xe7 //@snan | @qnan | @inf | @zero
     addl          GR_Bias = 0x0FFFF,r0
}

//
//   Is N zero?
//   Normalize x
//   Is integer type 32 bits?
//
{    .mfi
     cmp.eq        p6,p0 = r34,r0  
     fnorm.s1      FR_Norm_X  =   FR_Floating_X 
     cmp.eq        p8,p9 = r35,r0  
}
;;

//   Sign extend N if int is 32 bits
{    .mfi
(p9) mov           GR_N_as_int = r34     // Copy N if int is 64 bits
     nop.f 0
(p8) sxt4          GR_N_as_int = r34     // Sign extend N if int is 32 bits
}
;;

//
//   Branch and return special values.
//   Create -35000
//   Create 35000
//
{    .mfi
     addl          GR_Big = 35000,r0
     nop.f         0
     add           GR_N_Biased = GR_Bias,GR_N_as_int
}
{    .mfb
     addl          GR_NBig = -35000,r0
(p7) fma.s0        FR_Result = FR_Floating_X,f1, f0 
(p7) br.ret.spnt   b0  
};;

//
//   Build the exponent Bias
//   Return x when N = 0
//
{    .mfi
     setf.exp      FR_Two_N = GR_N_Biased                   
     nop.f         0
     addl          GR_Scratch1  = 0x063BF,r0 
}
{    .mfb
     addl          GR_Scratch  = 0x019C3F,r0 
(p6) fma.s0        FR_Result = FR_Floating_X,f1, f0 
(p6) br.ret.spnt   b0  
};;

//
//   Create 2*big
//   Create 2**-big 
//   Is N > 35000     
//   Is N < -35000     
//   Raise Denormal operand flag with compare
//   Main path, create 2**N
//
{    .mfi
     setf.exp      FR_NBig = GR_Scratch1                  
     nop.f         0
     cmp.ge        p6, p0 = GR_N_as_int, GR_Big
}
{    .mfi
     setf.exp      FR_Big = GR_Scratch                  
     fcmp.ge.s0    p0,p11 = FR_Floating_X,f0
     cmp.le        p8, p0 = GR_N_as_int, GR_NBig
};;

//
//   Adjust 2**N if N was very small or very large
//
{    .mfi
     nop.m 0
(p6) fma.s1        FR_Two_N = FR_Big,f1,f0
     nop.i 0
}
{ .mlx
     nop.m 999
     movl          GR_Scratch = 0x0000000000033FFF 
};;


{    .mfi
     nop.m 0
(p8) fma.s1        FR_Two_N = FR_NBig,f1,f0
     nop.i 0
}
{    .mlx
     nop.m 999
     movl          GR_Scratch1= 0x0000000000013FFF 
};;

//   Set up necessary status fields 
//
//   S0 user supplied status
//   S2 user supplied status + WRE + TD  (Overflows)
//   S3 user supplied status + FZ + TD   (Underflows)
//
{    .mfi
     nop.m 999
     fsetc.s3      0x7F,0x41
     nop.i 999
}
{    .mfi
     nop.m 999
     fsetc.s2      0x7F,0x42
     nop.i 999
};;

//
//   Do final operation
//
{    .mfi
     setf.exp      FR_NBig = GR_Scratch
     fma.s0        FR_Result = FR_Two_N,FR_Norm_X,f0 
     nop.i         999
}
{    .mfi
     nop.m         999
     fma.s3        FR_Result3 = FR_Two_N,FR_Norm_X,f0 
     nop.i         999
};;
{    .mfi
     setf.exp      FR_Big = GR_Scratch1
     fma.s2        FR_Result2 = FR_Two_N,FR_Norm_X,f0 
     nop.i         999
};;

//   Check for overflow or underflow.
//   Restore s3
//   Restore s2
//
{    .mfi
     nop.m 0
     fsetc.s3      0x7F,0x40
     nop.i 999 
}
{    .mfi
     nop.m 0
     fsetc.s2      0x7F,0x40
     nop.i 999
};;

//
//   Is the result zero?
//
{    .mfi
     nop.m 999
     fclass.m      p6, p0 =  FR_Result3, 0x007
     nop.i 999 
} 
{    .mfi
     addl          GR_Tag = 144, r0
     fcmp.ge.s1    p7, p8 = FR_Result2 , FR_Big
     nop.i 0
};;

//
//   Detect masked underflow - Tiny + Inexact Only
//
{    .mfi
     nop.m 999
(p6) fcmp.neq.unc.s1 p6, p0 = FR_Result , FR_Result2
     nop.i 999 
};; 

//
//   Is result bigger the allowed range?
//   Branch out for underflow
//
{    .mfb
(p6) addl           GR_Tag = 145, r0
(p8) fcmp.le.unc.s1 p9, p10 = FR_Result2 , FR_NBig
(p6) br.cond.spnt   LDEXPL_UNDERFLOW 
};;

//
//   Branch out for overflow
//
{ .mbb
     nop.m 0
(p7) br.cond.spnt   LDEXPL_OVERFLOW 
(p9) br.cond.spnt   LDEXPL_OVERFLOW 
};;

//
//   Return from main path.
//
{    .mfb
     nop.m 999
     nop.f 0
     br.ret.sptk     b0;;                   
}

GLOBAL_LIBM_END(__libm_ldexpl)
__libm_error_region:

LDEXPL_OVERFLOW: 
LDEXPL_UNDERFLOW: 

//
// Get stack address of N
//
.prologue
{ .mfi
    add   GR_Parameter_Y=-32,sp         
    nop.f 0
.save   ar.pfs,GR_SAVE_PFS
    mov  GR_SAVE_PFS=ar.pfs              
}
//
// Adjust sp 
//
{ .mfi
.fframe 64
   add sp=-64,sp                         
   nop.f 0
   mov GR_SAVE_GP=gp       
};;

//
//  Store N on stack in correct position 
//  Locate the address of x on stack
//
{ .mmi
   st8 [GR_Parameter_Y] =  GR_N_as_int,16       
   add GR_Parameter_X = 16,sp          
.save   b0, GR_SAVE_B0
   mov GR_SAVE_B0=b0                  
};;

//
// Store x on the stack.
// Get address for result on stack.
//
.body
{ .mib
   stfe [GR_Parameter_X] = FR_Norm_X 
   add   GR_Parameter_RESULT = 0,GR_Parameter_Y   
   nop.b 0
}
{ .mib
   stfe [GR_Parameter_Y] = FR_Result                 
   add   GR_Parameter_Y = -16,GR_Parameter_Y
   br.call.sptk b0=__libm_error_support#   
};;

//
//  Get location of result on stack
//
{ .mmi
   nop.m 0
   nop.m 0
   add   GR_Parameter_RESULT = 48,sp    
};;

//
//  Get the new result 
//
{ .mmi
   ldfe  FR_Result = [GR_Parameter_RESULT]      
.restore sp
   add   sp = 64,sp                       
   mov   b0 = GR_SAVE_B0                  
};;

//
//  Restore gp, ar.pfs and return
//
{ .mib
   mov   gp = GR_SAVE_GP                  
   mov   ar.pfs = GR_SAVE_PFS             
   br.ret.sptk     b0                  
};;

LOCAL_LIBM_END(__libm_error_region)

.type   __libm_error_support#,@function
.global __libm_error_support#