Lines Matching refs:to
11 | Double-extended value in memory location pointed to by address
21 | within 0.5001 ulp to 53 bits if the result is subsequently rounded
22 | to double precision. The result is provably monotonic in double
60 | 1.1 If |X| >= 2^(-65), go to Step 1.3.
61 | 1.2 Go to Step 7.
62 | 1.3 If |X| < 16380 log(2), go to Step 2.
63 | 1.4 Go to Step 8.
74 | to Step 2 guarantees |X| < 16380 log(2). There is no harm
75 | to have a small number of cases where |X| is less than,
76 | but close to, 16380 log(2) and the branch to Step 9 is
79 | Step 2. Calculate N = round-to-nearest-int( X * 64/log2 ).
81 | 2.2 N := round-to-nearest-integer( X * 64/log2 ).
89 | N := round-to-nearest-integer(Z)
101 | This error has to be considered later in Steps 3 and 4.
107 | the value -log2/64 to 88 bits of accuracy.
110 | c) The calculation X+N*L1 is also exact due to cancellation.
111 | Thus, R is practically X+N(L1+L2) to full 64 bits.
112 | d) It is important to estimate how large can |R| be after
115 | N = rnd-to-int( X*64/log2 (1+eps) ), |eps|<=2^(-24)
129 | Notes: a) In order to reduce memory access, the coefficients are
145 | 2^(J/64) to roughly 85 bits; T is in extended precision
147 | to 62 bits so that the last two bits of T are zero. The
154 | 6.1 If AdjFlag = 0, go to 6.3
169 | raised, to simulate that exception cost to much than the
178 | Note also that we use the FMOVEM instruction to move X
179 | in Step 7.1 to avoid unnecessary trapping. (Although
186 | 8.1 If |X| > 16480 log2, go to Step 9.
188 | 8.2 N := round-to-integer( X * 64/log2 )
193 | 8.7 Go to Step 3.
194 | Notes: Refer to notes for 2.2 - 2.6.
197 | 9.1 If X < 0, go to 9.3
198 | 9.2 ans := Huge, go to 9.4
222 | 1.1 If |X| >= 1/4, go to Step 1.3.
223 | 1.2 Go to Step 7.
224 | 1.3 If |X| < 70 log(2), go to Step 2.
225 | 1.4 Go to Step 10.
228 | because EXPM1 is intended to evaluate exp(X)-1 accurately
232 | Step 2. Calculate N = round-to-nearest-int( X * 64/log2 ).
233 | 2.1 N := round-to-nearest-integer( X * 64/log2 ).
249 | Notes: a) In order to reduce memory access, the coefficients are
265 | 2^(J/64) to roughly 85 bits; T is in extended precision
267 | to 62 bits so that the last two bits of T are zero. The
271 | bigger than 2^(-67.7) compared to the final result.
275 | 6.1 If M <= 63, go to Step 6.3.
276 | 6.2 ans := T + (p + (t + OnebySc)). Go to 6.6
277 | 6.3 If M >= -3, go to 6.5.
278 | 6.4 ans := (T + (p + t)) + OnebySc. Go to 6.6
286 | 7.1 If |X| >= 2^(-65), go to Step 9.
287 | 7.2 Go to Step 8.
295 | Notes: The idea is to return "X - tiny" under the user
303 | Notes: a) In order to reduce memory access, the coefficients are
304 | made as "short" as possible: B1 (which is 1/2), B9 to B12
305 | are single precision; B3 to B8 are double precision; and
321 | purposes. Therefore, go to Step 1 of setox.
489 fmovel %d0,%fp0 | ...convert to floating-format
503 |--a0 points to 2^(J/64), D0 is biased expo. of 2^(M)
590 fmovel %d0,%fp0 | ...convert to floating-format
607 bra EXPCONT1 | ...go back to Step 3
641 cmpil #0x4004C215,%d0 | ...70log2 rounded up to 16 bits
656 fmovel %d0,%fp0 | ...convert to floating-format
669 |--a0 points to 2^(J/64), D0 and a1 both contain M