Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!mit-eddie!uw-beaver!tektronix!orca!tekecs!frip!andrew
From: andrew@frip.gwd.tek.com (Andrew Klossner)
Newsgroups: comp.arch
Subject: 88k speed on FP add/neg/mov (was Re: RISC machines and scoreboarding)
Message-ID: <10136@tekecs.TEK.COM>
Date: 5 Jul 88 20:42:10 GMT
References: <1082@nud.UUCP> <2438@winchester.mips.COM> <1098@nud.UUCP> <2459@gumby.mips.COM> <1110@nud.UUCP> <2483@gumby.mips.COM>
Sender: andrew@tekecs.TEK.COM
Organization: Tektronix, Wilsonville, Oregon
Lines: 20

In article <2483@gumby.mips.COM>, earl@mips.COM (Earl Killian) presents
a table of floating point times for the 88k and R3010, with this entry:

> 		88100	R3010	cycle
> Operation	cycles  cycles	ratio
> ======================================
> abs/neg/mov	  5?	  1

To clear up the question mark: on the 88k, floating point values are
kept in general registers, so "abs" is done by ANDing to 0 the sign
bit, "neg" is done by XORing the sign bit, and "mov" is done with
conventional register-to-register move instructions.  Thus:

  abs		  1
  neg		  1
  sp mov	  1
  dp mov	  2

  -=- Andrew Klossner   (decvax!tektronix!tekecs!andrew)       [UUCP]
                        (andrew%tekecs.tek.com@relay.cs.net)   [ARPA]