Lines Matching refs:and
4 | Normalizes denormalized numbers if necessary and updates the
6 | machine and the 040 completes the operation. This routine
11 | completed here; data is packed and moved to user memory.
85 | At this point, only opclass 0 and 2 possible
102 | At this point, fabs, fneg, fsmove, fdmove, ftst, fsqrt, fssqrt, and
104 | For cases fabs, fneg, fsmove, and fdmove goto spos (do not normalize)
105 | For cases fsqrt, fssqrt, and fdsqrt goto nrm_src (do normalize)
112 | destination precision and mode.
125 | Inst is ftst. Check the source operand and set the cc's accordingly.
179 | and bit 6 is set, round is forced to single. If both are clear,
194 | FPCR_MODE byte to get rounding on norms and zeros.
300 | The mantissa is zero from the denorm loop. Check sign and rmode
338 | Write the result to memory, setting the fpsr cc bits. NaN and Inf
376 | At this point, fabs, fneg, fsmove, fdmove, ftst, fsqrt, fssqrt, and
378 | For cases fabs, fneg, fsmove, and fdmove goto spos (do not normalize)
379 | For cases fsqrt, fssqrt, and fdsqrt goto nrm_src (do normalize)
414 | cu_dnrm handles all cu-only instructions (fmove, fabs, fneg, and
428 | Inst is ftst. Check the source operand and set the cc's accordingly.
475 | and bit 6 is set, round is forced to single. If both are clear,
477 | is double or single, the result is zero, and the mode is checked
497 | Check rmode for rp or rm and set lsb accordingly.
512 | Check for rp or rm and set lsb accordingly.
527 | Simply set unfl (not inex2 or aunfl) and write the result to
537 | cu_dpdr and cu_dndr set the double lsb.
569 | cu_dpdr and cu_dndr set the single lsb.
603 | unit and must be completed in software. Basically, this
604 | condition exists with a very large norm and a denorm. One
614 | The wrap-around condition occurs for add, sub, div, and cmp
619 | and for mul when
632 | use fpu to perform dest+new_src (user's rmode and X)
636 | call round with user's precision and mode
637 | move result to fpn and wbtemp
642 | use fpu to perform dest+new_src (user's rmode and X)
646 | call round with user's precision and mode
647 | move result to fpn and wbtemp
698 | and force the result.
802 | and complete the instruction.
825 | can be used to add the norm and 1.0 with the sign of the
826 | denorm and it will correctly generate the result in extended
827 | precision. We can then call round with no sticky and the result
828 | will be correct for the user's rounding mode and precision. If
830 | and the result will be correct for the user's rounding mode and
850 fmovel %d0,%fpcr |set up users rmode and X
855 orl %d1,USER_FPSR(%a6) |capture cc's and inex from fadd
878 fmovel %d0,%fpcr |set up users rmode and X
882 orl %d1,USER_FPSR(%a6) |capture cc's and inex from fadd
963 | and aovfl, and clr the mantissa (incorrectly set by the
977 | and complete the instruction.
1001 | denorm and it will correctly generate the result in extended
1002 | precision. We can then call round with no sticky and the result
1003 | will be correct for the user's rounding mode and precision. If
1005 | and the result will be correct for the user's rounding mode and
1025 fmovel %d0,%fpcr |set up users rmode and X
1029 orl %d1,USER_FPSR(%a6) |capture cc's and inex from fadd
1053 fmovel %d0,%fpcr |set up users rmode and X
1057 orl %d1,USER_FPSR(%a6) |capture cc's and inex from fadd
1144 | and aovfl, and clr the mantissa (incorrectly set by the
1158 | and complete the instruction.
1196 | and complete the instruction.
1329 | A check is made on enabled and signalled snan exceptions,
1330 | and the destination is not overwritten if this condition exists.
1335 btstb #snan_bit,FPSR_EXCEPT(%a6) |if snan is set, and
1337 btstb #snan_bit,FPCR_ENABLE(%a6) |and don't overwrite
1445 | set if the result is inex and unfl is signalled.
1503 | at this point, the answer is between the largest pos and neg values
1549 | at this point, the answer is between the largest pos and neg values
1594 | at this point, the answer is between the largest pos and neg values
1624 | operand and round down for negative answer. In the first case (result = 1)
1884 | This subroutine takes an extended normalized number and denormalizes
1974 | Notes on handling of special case (zero, inf, and nan) inputs: