The G4 revolution

Andrew Rambaut andrew.rambaut at zoo.ox.ac.uk
Mon Sep 13 05:46:50 EST 1999


In article <7rarb6$e5a at net.bio.net>, Hans Sluiman
<h.sluiman at rbge.org.uk> wrote:

> My obvious dilemma is this: would it be more sensible at this point in
> time to invest in a high speed G3 Mac (tried and tested technology
> available at reasonable prices), or to pay a premium price for a top of
> the range superfast G4 whose improved performance may well be disappointing?

Actually the Macintosh compiler that everyone uses already has AltiVec
support in it so significant improvements can be expected with just
a recompile. For optimum speed, some rewriting (of a few core bits)
would be required. To answer your last question, PAUP and such programs
use floating point when doing analyses like maximum likelihood but not
with most parsimony analyses. The floating point processor (that is
seperate from the AltiVec) is much faster on the G4 than the G3. With
the exception of PAUP all the other programs you mentioned are free and
have source code so you can expect G4 versions as soon as someone with
an interest in them buys a new machine. I expect PAUP will have G4
support in a reasonable time frame. 

But remember these applications will run faster on a G4 of the same 
clock speed even without recompiling (faster processor, faster floating
point unit). Expect a 25% improvement (i.e. with a 500MHz G4 that would
be about 38% faster than a 450MHz G3).

Hope this helps,
Andrew




More information about the Bio-soft mailing list