Return to README CVS log

Up to [local] / OpenXM_contrib / gmp / mpn / mips64

Annotation of OpenXM_contrib/gmp/mpn/mips64/README, Revision 1.1.1.1

1.1       ohara       1: Copyright 1996 Free Software Foundation, Inc.
                      2:
                      3: This file is part of the GNU MP Library.
                      4:
                      5: The GNU MP Library is free software; you can redistribute it and/or modify
                      6: it under the terms of the GNU Lesser General Public License as published by
                      7: the Free Software Foundation; either version 2.1 of the License, or (at your
                      8: option) any later version.
                      9:
                     10: The GNU MP Library is distributed in the hope that it will be useful, but
                     11: WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
                     12: or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
                     13: License for more details.
                     14:
                     15: You should have received a copy of the GNU Lesser General Public License
                     16: along with the GNU MP Library; see the file COPYING.LIB.  If not, write to
                     17: the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
                     18: 02111-1307, USA.
                     19:
                     20:
                     21:
                     22:
                     23:
                     24: This directory contains mpn functions optimized for MIPS3.  Example of
                     25: processors that implement MIPS3 are R4000, R4400, R4600, R4700, and R8000.
                     26:
                     27: RELEVANT OPTIMIZATION ISSUES
                     28:
                     29: 1. On the R4000 and R4400, branches, both the plain and the "likely" ones,
                     30:    take 3 cycles to execute.  (The fastest possible loop will take 4 cycles,
                     31:    because of the delay insn.)
                     32:
                     33:    On the R4600, branches takes a single cycle
                     34:
                     35:    On the R8000, branches often take no noticable cycles, as they are
                     36:    executed in a separate function unit..
                     37:
                     38: 2. The R4000 and R4400 have a load latency of 4 cycles.
                     39:
                     40: 3. On the R4000 and R4400, multiplies take a data-dependent number of
                     41:    cycles, contrary to the SGI documentation.  There seem to be 3 or 4
                     42:    possible latencies.
                     43:
                     44: STATUS
                     45:
                     46: Good...