Re: G4 for FACSCalibur

From: Karim Vermaelen (Karim.Vermaelen@rug.ac.be)
Date: Mon Oct 02 2000 - 07:41:27 EST


"J. Paul Robinson" wrote:

> So Mario, when are you going to talk about software that runs on computers that
> the other 90% plus that people have.......in terms of world computer
> sales, Apple is right up there at .....4%!!
> Paul
>

So if FlowJo was a PC-only application I would have to buy a Wintel machine on top
of my Mac, but that wouldn't matter because we're a minority anyway?
The 4% your talking about are indeed an estimate of worldwide Apple hardware
sales, but we're talking education/academic sector here, where the Mac market
share rises to 20% (and much higher in the flow cytometry area as we all know). I'
ve worked on both platforms and believe me I wouldn't even think about doing my
FACS work (or any other daily lab tasks) on an NT machine. This may not be very
obvious for someone with an engineering background, but I just like the fact that
I can easily do maintenance and troubleshooting on our Mac network without any
expert IT support.
So all things taken into consideration I think the choice of developing software
such as FlowJo on Macs is a very pragmatic one.

But the real shocking news now is this report by Keith Bahjat stating that
Cellquest is not only non-optimized for the G4's vector processing, but was never
optimized for the PowerPC architecture as well! What a shame and a waste of power:
a 68k-application on a G4!! I hope someone from BD is reading this and can comment
on this shortcoming in development.

Karim Vermaelen
Pulmonary Immunobiology
Ghent University Hospital
Ghent, Belgium


> Date sent:              Tue, 26 Sep 2000 16:16:30 -0400
> To: cyto-inbox
> From:                   Mario Roederer <Roederer@drmr.com>
> Subject:                Re: G4 for FACSCalibur
>
> >
> > I tested FlowJo for a number of different tasks (e.g.,generating
> > complex graphical layouts; generating large tabular outputs; even
> > analyzing 13-parameter, million-event data files) on G4's vs. G3.
> > The G4 was significantly faster (taking into account the clock
> > speed), even when the program was not G4-optimized.  I also tested a
> > G4-optimized version of FlowJo that takes advantage of the AltiVec
> > instructions (no, not yet commercially available!), and it's another
> > large step up.
> >
> > As more and more applications come out to be AltiVec-aware (and they
> > are!), everyone will realize the enormous speed benefit from this
> > technology.
> >
> > Note that the multi-processing nature of Altivec should not be
> > confused with the multi-processing capability of dual-processor
> > machines (like the new G4 cube).  There's a completely different set
> > of optimizations that programs need to fully take advantage of
> > multi-processor machines (and I'm sure that will be forthcoming in
> > the next year or two as well).
> >
> > Bottom line--G4's are faster, but G3's are a lot cheaper (right now). But
> > as more and more applications are Altivec-aware, then there won't be a
> > comparison.
> >
> > mr
> >
> > (PS--Intel-based machines, even the gigaherz processors, don't come
> > close to the G4 capabilities.)
>
> J.Paul Robinson, Ph.D.,
> Professor of Immunopharmacology
> Professor of Biomedical Engineering
> Director, Purdue University Cytometry Laboratories
> Hansen Hall, Roon B050
> Purdue University, West Lafayette, IN 47907-1515
> (Ph) (765) 494-0757  Fax (765) 494-0517
> jpr@flowcyt.cyto.purdue.edu
>
> http://www.cyto.purdue.edu
>
> http://www.bioscope.org



This archive was generated by hypermail 2b29 : Sat Mar 10 2001 - 19:31:33 EST