Copyright 1999, 2001, 2002, 2004 Free Software Foundation, Inc.

This file is part of the GNU MP Library.

The GNU MP Library is free software; you can redistribute it and/or modify
it under the terms of either:

  * the GNU Lesser General Public License as published by the Free
    Software Foundation; either version 3 of the License, or (at your
    option) any later version.

or

  * the GNU General Public License as published by the Free Software
    Foundation; either version 2 of the License, or (at your option) any
    later version.

or both in parallel, as here.

The GNU MP Library is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
for more details.

You should have received copies of the GNU General Public License and the
GNU Lesser General Public License along with the GNU MP Library.  If not,
see https://www.gnu.org/licenses/.




This directory contains mpn functions for 64-bit PA-RISC 2.0.

PIPELINE SUMMARY

The PA8x00 processors have an orthogonal 4-way out-of-order pipeline.  Each
cycle two ALU operations and two MEM operations can issue, but just one of the
MEM operations may be a store.  The two ALU operations can be almost any
combination of non-memory operations.  Unlike every other processor, integer
and fp operations are completely equal here; they both count as just ALU
operations.

Unfortunately, some operations cause hickups in the pipeline.  Combining
carry-consuming operations like ADD,DC with operations that does not set carry
like ADD,L cause long delays.  Skip operations also seem to cause hickups.  If
several ADD,DC are issued consecutively, or if plain carry-generating ADD feed
ADD,DC, stalling does not occur.  We can effectively issue two ADD,DC
operations/cycle.

Latency scheduling is not as important as making sure to have a mix of ALU and
MEM operations, but for full pipeline utilization, it is still a good idea to
do some amount of latency scheduling.

Like for all other processors, RAW memory scheduling is critically important.
Since integer multiplication takes place in the floating-point unit, the GMP
code needs to handle this problem frequently.

STATUS

* mpn_lshift and mpn_rshift run at 1.5 cycles/limb on PA8000 and at 1.0
  cycles/limb on PA8500.  With latency scheduling, the numbers could
  probably be improved to 1.0 cycles/limb for all PA8x00 chips.

* mpn_add_n and mpn_sub_n run at 2.0 cycles/limb on PA8000 and at about
  1.6875 cycles/limb on PA8500.  With latency scheduling, this could
  probably be improved to get close to 1.5 cycles/limb.  A problem is the
  stalling of carry-inputting instructions after instructions that do not
  write to carry.

* mpn_mul_1, mpn_addmul_1, and mpn_submul_1 run at between 5.625 and 6.375
  on PA8500 and later, and about a cycle/limb slower on older chips.  The
  code uses ADD,DC for adjacent limbs, and relies heavily on reordering.


REFERENCES

Hewlett Packard, "64-Bit Runtime Architecture for PA-RISC 2.0", version 3.3,
October 1997.