From: Robert Myers on
Del Cecchi wrote:

> "Robert Myers" <rbmyersusa(a)gmail.com> wrote in message

> > Without that customer, Blue Gene
> > would not exist, or certainly not at the scale and with the visibility
> > it has now.
> >
> Your supposition about blue gene is incorrect. I was involved somewhat
> at the time Blue Gene/L was getting initially started and there was
> nothing about bombs or national labs involved. Actually it was
> computational chemistry, proteins, genes, drugs, all that sort of thing.
> It turned out to be interesting to the bomb guys also apparently. But
> it was proteins first.
>

I wrote that statement with considerable care, and it is correct as I
wrote it.

Robert.

From: David Kanter on

Greg Lindahl wrote:
> In article <CbadnSpeDeLqweHYnZ2dnUVZ_segnZ2d(a)comcast.com>,
> Joe Seigh <jseigh_01(a)xemaps.com> wrote:

> Note that Blue Gene's chip has 2 cpus which are similar yet don't
> share cache. That's the odder one.

That's certainly true.

> >I wonder how the old shared memory strategy will work out. Will
> >coherent cache shared memory scale up to 10's and 100's of processors
> >and stay competitive?
>
> Altix scales to 1000s, and for some things, it's killer. But not
> enough things; distributed memory machines are too good for too much
> of the market.

Distributed memory? The Altix is a DSM system. Perhaps you mean that
the price/performance for commodity clusters with cheap (or even
sometimes, nice) interconnects is just superior for certain classes of
problems.

DK

From: Del Cecchi on

"Robert Myers" <rbmyersusa(a)gmail.com> wrote in message
news:1165790341.056925.229560(a)j72g2000cwa.googlegroups.com...
> Del Cecchi wrote:
>
>> "Robert Myers" <rbmyersusa(a)gmail.com> wrote in message
>
>> > Without that customer, Blue Gene
>> > would not exist, or certainly not at the scale and with the
>> > visibility
>> > it has now.
>> >
>> Your supposition about blue gene is incorrect. I was involved
>> somewhat
>> at the time Blue Gene/L was getting initially started and there was
>> nothing about bombs or national labs involved. Actually it was
>> computational chemistry, proteins, genes, drugs, all that sort of
>> thing.
>> It turned out to be interesting to the bomb guys also apparently.
>> But
>> it was proteins first.
>>
>
> I wrote that statement with considerable care, and it is correct as I
> wrote it.
>
> Robert.
>
I missed the "or" in the middle. It would have existed, although perhaps
as a one off research thing.


From: Greg Lindahl on
In article <1165792897.676572.323980(a)79g2000cws.googlegroups.com>,
David Kanter <dkanter(a)gmail.com> wrote:

>> Altix scales to 1000s, and for some things, it's killer. But not
>> enough things; distributed memory machines are too good for too much
>> of the market.
>
>Distributed memory? The Altix is a DSM system.

Distributed memory (note the lack of "shared") is the term some people
use to refer to clusters running MPI.

> Perhaps you mean that
> the price/performance for commodity clusters with cheap (or even
> sometimes, nice) interconnects is just superior for certain classes of
> problems.

Yes, that's another way of saying what I said.

-- greg


From: Casper H.S. Dik on
"Robert Myers" <rbmyersusa(a)gmail.com> writes:

>Blue Gene is a perfect example. It was not driven by market economics,
>but by a specific need of a bomb lab. Without that customer, Blue Gene
>would not exist, or certainly not at the scale and with the visibility
>it has now. IBM has been reasonably shrewd in trying to leverage those
>nuclear weapons dollars into other business, but that doesn't change
>the reality of how the money flows.

And in those cases that out-of-US competitors threatened to outbid
local vendors with faster and cheaper systems, they were invariably
(and generally falsely) accussed of dumping and fined with an
import duty.

The US' understanding of free market is "free for the US to sell in
(full stop)"

Casper
--
Expressed in this posting are my opinions. They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5 6
Prev: Need SimpleScalar GCC Compiler
Next: Software vs Hardware