http://tinyurl.com/mepn4j


it has started already."

Um... wasn't parallel processing supposed to be "the next big thing" 20 
years ago?

The long and short of it is, it's much harder to write non-sequential 
software. (At least, using current techniques.) And then there's the 
"minor problem" of memory bandwidth...

Post a reply to this message

Invisible <voi### [at] devnull> wrote:
> "Parallel programming is the next big thing for the world of computing ??? 
> it has started already."

> Um... wasn't parallel processing supposed to be "the next big thing" 20 
> years ago?

  You should understand marketing speech. Everything is new when the
market decides it's new.

  I laughed some years ago when they introduced a really revolutionary and
innovative new concept: Dumb terminals. Of course the marketing speech didn't
say that dumb terminals have existed for something like 30 years (and fallen
mostly out of use by about the mid-90's).

-- 
                                                          - Warp

Post a reply to this message

Warp wrote:

>   I laughed some years ago when they introduced a really revolutionary and
> innovative new concept: Dumb terminals. Of course the marketing speech didn't
> say that dumb terminals have existed for something like 30 years (and fallen
> mostly out of use by about the mid-90's).

Ah yes, and that's the other thing. Computer technology seems to 
progress in cycles.

Once upon a time, everybody used dumb terminals connected to a socking 
great mainframe in the center. This was considered the ultimate way forward.

Then people decided that it was better to have powerful computers on 
individual people's desks, so that the cost of the hardware could be 
spread out as PCs are purchased one at a time, rather than having to 
spend on one huge mainframe, all at once.

And then everybody started getting excited about "thin clients". Have 
one central server that runs all the software; you only have to install 
and configure it once. Then you give everybody cheap-arse PCs and let 
them access the software remotely using VNC / HTTP / whatever.

In other words, dumb terminals.

I suspect what it comes down to is that both thin client and thick 
client approaches have advantages, so the market will forever oscilate 
between the two, continually deciding that whichever approach is 
currently unpopular is the "new" and "revolutionary" one that everybody 
should adopt, seeing its advantages and forgetting its disadvantages. 
Until the cycle repeats.

Also... that was fairly poor sentence structure. :-/

Post a reply to this message

On Thu, 27 Aug 2009 11:30:18 +0100, Invisible wrote:

> Um... wasn't parallel processing supposed to be "the next big thing" 20
> years ago?

Welcome to the true world of technology.  Innovation rarely happens, most 
of the things hailed as innovation tend to be reimplementations of older 
ideas on faster hardware.

Thin client or fat client?  Look at the history, and you'll see that 
we've just gone in circles.  The new thing is "cloud computing", which to 
me seems to be a reimplementation of previous thin client technologies, 
but instead of connecting to one system, you're connecting to several and 
an application could be on any of them (the load distribution stuff is 
something that looks pretty new to me, so there may be an actual 
innovation there).

Jim

Post a reply to this message

Warp wrote:
>   You should understand marketing speech. Everything is new when the
> market decides it's new.

Well, parallel processing *did* take off 20 years ago. Except that each 
process served one transaction. :-) (Think Google, Amazon, etc.)

>   I laughed some years ago when they introduced a really revolutionary and
> innovative new concept: Dumb terminals. Of course the marketing speech didn't
> say that dumb terminals have existed for something like 30 years  

30? What's a card punch other than a dumb terminal? :-) Dumb terminals have 
been around *way* longer than 30 years. Dumb *graphics* terminals 30 years 
ago, perhaps. :-)

-- 
   Darren New, San Diego CA, USA (PST)
   Understanding the structure of the universe
    via religion is like understanding the
     structure of computers via Tron.

Post a reply to this message

Jim Henderson wrote:

> Welcome to the true world of technology.  Innovation rarely happens, most 
> of the things hailed as innovation tend to be reimplementations of older 
> ideas on faster hardware.
> 
> Thin client or fat client?  Look at the history, and you'll see that 
> we've just gone in circles.

Functional programming? That'd be based on the lambda calculus then, 
which predates computer hardware just like the Turing machine that most 
current programming languages are based on. The FP guys are getting all 
excited because FP makes it inherantly easiER to write multithreaded 
code, but... FP languages have been around for decades, and never gone 
anywhere. (Hell, when Prolog was invented, people thought that HAL was 
just around the corner!)

Oh, and self-programming computers? Unlikely ever to happen. :-P

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

On Thu, 27 Aug 2009 20:54:30 +0100, Orchid XP v8 wrote:

> Jim Henderson wrote:
> 
>> Welcome to the true world of technology.  Innovation rarely happens,
>> most of the things hailed as innovation tend to be reimplementations of
>> older ideas on faster hardware.
>> 
>> Thin client or fat client?  Look at the history, and you'll see that
>> we've just gone in circles.
> 
> Functional programming? That'd be based on the lambda calculus then,
> which predates computer hardware just like the Turing machine that most
> current programming languages are based on. The FP guys are getting all
> excited because FP makes it inherantly easiER to write multithreaded
> code, but... FP languages have been around for decades, and never gone
> anywhere. (Hell, when Prolog was invented, people thought that HAL was
> just around the corner!)
> 
> Oh, and self-programming computers? Unlikely ever to happen. :-P

You'll note that I said "rarely", not "never".  :-P

Jim

Post a reply to this message

Darren New escreveu:
> Warp wrote:
>>   I laughed some years ago when they introduced a really revolutionary 
>> and
>> innovative new concept: Dumb terminals. Of course the marketing speech 
>> didn't
>> say that dumb terminals have existed for something like 30 years  
> 
> 30? What's a card punch other than a dumb terminal? :-) Dumb terminals 
> have been around *way* longer than 30 years. Dumb *graphics* terminals 
> 30 years ago, perhaps. :-)

BTW, there are some crazy gaming companies around pushing for 
centralized rendering for games:  you render the games in your servers 
and then push the graphics and sounds onto users' screens, this way 
actually subdisizing the hardware and making it irrelevant so that users 
can play in any event.  I wonder about the latency and fail whales...

-- 
a game sig: http://tinyurl.com/d3rxz9

Post a reply to this message

nemesis wrote:
> BTW, there are some crazy gaming companies around pushing for 
> centralized rendering for games:

I saw that. It was likely a scam.

Now, maybe in an arcade setting or something, where you can run fiber to a 
bunch of machines, but then why not just have multiple VDUs on one graphics 
card?

-- 
   Darren New, San Diego CA, USA (PST)
   Understanding the structure of the universe
    via religion is like understanding the
     structure of computers via Tron.

Post a reply to this message

Orchid XP v8 escreveu:
> FP languages have been around for decades, and never gone 
> anywhere.

Never had a chance in a heavily imperative mindset built around 
sequential Turing machines.

But now multicore processors got true parallel processing into an 
everyday reality for millions of desktop PCs.  An oportunity is open for 
FPLs to show the value of abstracting away and reducing the use of 
side-effects in order to achieve easy parallelism through out-of-order 
evaluation as well as doing away with locks in STM...

-- 
a game sig: http://tinyurl.com/d3rxz9

Post a reply to this message