New Hardware Company Dedicated to Squeak

David P. Harris dpharris at telus.net
Sat Apr 21 20:48:58 UTC 2001


Hi-
Any guestimate on a price for this unit (with or without the LCD)?
David

Tom Poliquin wrote:

> Thanks for the responses ..
>
> Let me give you a little more data ..
>
> We are running Squeak 2.2 and ran #benchmark ..
>
> 835,000 bytecodes / sec ..
>
> We are changing to Hyperstone's new chip which
> runs at 160Mhz (and still costs about $5-6 in qty)
> so that should double that result.
>
> The Hyperstone has a single core that does both
> the RISC operations and DSP operations (capable
> of partial overlap).
>
> We want to implement Squeak primitives which
> utilize the DSP .. (ahh.. not enough hours in
> the day).
>
> It's not a morphic machine (we tried it and it's
> very slow) but is quite smappy in MVC when developing
> on it.
>
> We're trying to get a cheap platform .. all the logic
> to control the LCD, Smart Media, Keyboard, and Mouse
> is in cheap CPLDs .. for Squeak based Internet Appliances.
> We actually develop on the platform too .. a side
> benefit.
>
> Using Squeak (aside from loving Smalltalk) is great because
> we can change anything from the lowliest latch in the CPLD hardware
> thru drivers, primitives, and high level code .. to increase
> performance.
>
> We're adding a modem (a soft one will come later) and
> a touchscreen option ..
>
> Thanks for all the comments ..
>
> Pictures from the show are on the bottom of the
> main page at
>
>    http://www.softcomp.com
>
> Tom
>
> Alan Kay wrote:
> >
> > Tim --
> >
> > At 11:26 AM -0700 4/20/01, Tim Rowledge wrote:
> > >Hans-Martin Mosner <hmm at heeg.de> is widely believed to have written:
> > >
> > >>  do you have any performance data for the Squeak VM on that RISC core?
> > >It's not going to be terribly fast at only 80MHz (ARM can do 400MHz for
> > >ARM1020FE, ARM/XScale can do 1GHz ). And the mips/watt is quite good but
> > >not exceptional (XSCALE claims 1.2gips/watt at 800MHz, 3+gips/watt at
> > >400MHz, ARM1020 is around 1.5gips/watt including an array float
> > >processor) . And variable length instructions doesn't really fit with
> > >the claims of being 'strictly RISC' in the blurb.
> > >
> > >Nonetheless, it's a good bit of work to do all the code to support it
> > >and get Squeak running - congratulations. Variety is the spice of life -
> > but try explaining that to your spouse :-)
> >
> > She'll understand if you make the analogy to new clothes ....
> >
> > Cheers,
> >
> > Alan





More information about the Squeak-dev mailing list