[M3devel] hardware/software sales.. AlphaServer DS10L, Tru64

Thu Jun 30 21:00:41 CEST 2011

Hi all:
if you ask me why then would I share the question with you,
can you make this graphic in postcript easily in how many seconds can
you draw this document p13 in the screen in ps or eps:
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.47.7537

If
you can build such quicker than and more than that edit -> compile
-> view quickly then you're ready to go. Think in ULSI and VLSI
systems doing this for the work stuff (if anything us included) .

The Ps (already present) or JVM backends (see:
http://moscova.inria.fr/~leifer/articles/srcreport.html

)
are quick enough, but to what it makes that is the constraint solver,
which is indeed very advanced combining symbolic and numerical
solving using provided initial approximation. So at least less fpu
cycles to
help the later are welcome (e.g any hardware speed-up might solve this
better). I'm not really sure how much of this does Modula-3 as for I
know in i86pc boxes it doesn't really do that (i.e it doesn't
accelerate FPU in the intended way, e.g doesn't handle in same cycles
as you said normally), but SGI, DEC, perhaps PA-RISC, and PPC are other
history (this makes
again more important to keep this older machine ports, as for reference
points, as
I don't have such good machines to keep it alive with some cm3ide
server on).  Experimental platforms might be for reference as well:
http://www.ercim.eu/publication/Ercim_News/enw32/dobson.html

Anyway
to test i86pc or possible *DOS implementations will need something like
NTHydra or so UI client of MIPSr4000, Alpha, i86pc in NT, or so to do in-situ testing and others alike as well as Zeus sessions available (see man Juno.1 for more details).

Thanks in advance

--- El jue, 30/6/11, felipe valdez <felipevaldez at gmail.com> escribió:

De: felipe valdez <felipevaldez at gmail.com>
Asunto: Re: [M3devel] hardware/software sales.. AlphaServer DS10L, Tru64
Para: "Daniel Alejandro Benavides D." <dabenavidesd at yahoo.es>
Fecha: jueves, 30 de junio, 2011 10:55

2011/6/30 Daniel Alejandro Benavides D. <dabenavidesd at yahoo.es>

Hi all:

perhaps, you could check if this recording makes sense, i says at the beginning is an excerpt (So a good idea would get a copy of the original).

http://www.open-video.org/details.php?videoid=8026

this was actually  very intersting and educational!
it reminded me of "the mother of all demso", by douglas engelbert.

Anyway, the major critics of this is the supposed to be low speed of execution, I think if this run on a Win9* pc, I can just to conclude that:

1. The system is functional enough to run in a 199* system.

certainly, one could conclude that...
but just trowing more cycles at it, doesn't make it faster...

2. The critics are true if the system is stress-tested (see 1 for this info)

3. The idea behind the constraint solver is not enough compiler-enabled optimized for today's applied benchmarks

If the above is true then there might be some techniques to make that happen In example tune the compiler as they did in development.

in this case, I don't think performance is the responsability of the compiler writers, but of the sofwtare writer.

As it hasn't been too much updated besides a JVM-based backend there is a big question how much of this is good to pick, but JVM-based backend

http://moscova.inria.fr/~leifer/articles/srcreport.html

doesn't handle the constraint solver at all. So we might check what have been the main constraint solver techniques developed for that purpose (lately logic analyzers, but sold as a kit, perhaps doable on Modula-3, perhaps and ASIP but can we build one

to what end, would we build one exactly?other than to satisfy this  curiosity?
does this have a real-world application?
I fail to think of one...

  of such, perhaps yes with some library called m3tools):

http://members.tripod.com/srini_seetharam/thesis.pdf

Back indee in DEC-SRC they offered to license the toolkit in other platforms, perhaps somebody would asked about it, I'm not aware of that but it seems they could seek that in the future.

given that cm3 used to be paid, but now is free, I think the market would go for that which is cheaper.therefore, a commercial implementation, of a not-widely used language, seems to be only desirable, for those who

*already* have legacy systems based on such tecnology, and people would probably not "buy" into it, so that they could develop new software, given free (and open source) alternatives.

also, since the main use of m3 seems to be research, and researches are often trying to use free sofwtare, and getting away from paid tools, it seems plausible to say, with a certain level of certainty, that this (license tooklit) is very unlikely to happen, in my opinion

Thanks in advance

--- El mié, 29/6/11, Hendrik Boom <hendrik at topoi.pooq.com> escribió:

> De: Hendrik Boom <hendrik at topoi.pooq.com>

> Asunto: Re: [M3devel] hardware/software sales.. AlphaServer DS10L, Tru64

> Para: m3devel at elegosoft.com

> Fecha: miércoles, 29 de junio, 2011 20:35

> On Wed, Jun 29, 2011 at 09:55:34PM

> +0000, Jay K wrote:

> >

> >   > completely forgot the Win9*

> systems

> >

> >   That is probably best -- to just

> forget it.

>

> I still have a working windows 95 system.  I turn it

> on once in a blue moon.

> It dual-boots Windows 95 and linux.  If it would

> really be useful, I could try

> Moduls-3-ing on it, but it is probably not be worth the

> effort.

>

> >

> >

> >    > Finally a word about of Juno-2, if

> the most important of tools is that,

> > there is a way to do it, I run it on cygwinb20, it

> runs smoothly,

> > interesting tough to see it on a Chicago/PC!

>

> I've never figured out waht Juno is supposed to do.  I

> tried running it once, and

> it was most mysterious.

>

> -- hendrik

>

-- 
312-444-2124Skype: f3l.headhunterCasa: 8043901

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://m3lists.elegosoft.com/pipermail/m3devel/attachments/20110630/9d74454f/attachment-0002.html>