Re: self-extracting diskette image to hard disk? (or, IBM keeping inexpensive supercomputers away from the consumer).

On Sat, 22 Oct 2005 17:00:08 UTC, "Doug Bissett"
<doug.bissett!SPAM@xxxxxxxxxxxxx> wrote:

> On Fri, 21 Oct 2005 22:49:56 UTC, "Kevin K" <kevink4@xxxxxxxxx> wrote:
>
> ...snip...
> > Since you are the one who has claimed that all this could be done
> > economically with "existing" technology 5 or 10 years ago, I think it
> > is up to you to show references that this technology was really
> > available 5 years ago, just absent a will by IBM to make huge amounts
> > of money, or at least license this technology to others, which would
> > also have brought in huge amounts of money, and with even less risk.
>
> I know what IBM was doing, long before 5 years ago. 15 years ago, they
> could have done a 2 Ghz PC, with 10 GiB disks, and 1 GiB of memory.
> The cost would have been roughly $40,000 per machine, and many would
> argue that it was not a PC, because it would have been about 5 times
> the size of todays machines. That was also about the time that OS/2
> was really stating to take shape. perhaps, rather than saying it could
> have been a 2 Ghz machine, I should say that it could have been
> equivalent to 2 Ghz, compared to today's machines. There are ways to
> make a machine faster, other than with a faster clock.
>
> > Also, remember that you mentioned that the hard drives back then could
> > have been 1000GB back then. Years of hard drive expansion growth
> > since then still have not brought it up to that (though hard drive
> > capacity has grown faster than clock speed).
>
> How many platters do you want? Double the platters, and you double the
> capacity.
>
> > And what about your claim that, 5-10 years ago, you could also have
> > had 16 gig of RAM also included within this $3000 computer? Your
> > claiming not only that memory would be many times larger than existing
> > memory at this time frame than was currently available, but that it
> > would me many multiples faster, to keep up with the 30GHz CPU.
>
> Memory interleave was used in early PCs, to make slow memory work
> faster. IBM's mainframes had memory that was a LOT faster than
> anything ever used in PCs, and proper caching would also be required.
> Then again, memory, today, is very much slower than the processor
> anyway, so getting a faster processor, doesn't really make as much
> difference as many people seem to think.
>
> > Since companies, including IBM, have shown themselves willing to
> > invest billions of dollars in fabrication plants for CPUs, etc, your
> > argument that they just didn't want to spend the money on technology
> > that would have killed all their competition means that the technology
> > must have been VERY expensive. Just how expensive? 100s of
> > Billions? With that kind of money, they could not have made money
> > selling these super computers for $3000.
>
> Economies of scale, and the fact that IBM hplds the patents on a
> number of technologies to create low power, high speed, electronics,
> meant that they could have cut the cost to the point where they could
> have done it. Of course, no company wants to spend a lot of money on
> something that isn't going to pay off for a few years.
>
> > In 20+ years of following the computer industry, this is the first
> > time I've ever heard claims that computers so far past the state of
> > the art are not only possible (at any price), but at a very
> > inexpensive price.
> >
> > The ball is still in your court.
>
> It wasn't done, so it is all speculation. I know that IBM had the
> expertise, and the resources, to be able to do far more with the PC,
> than they (or you) ever imagined, and making large investments, to
> skip whole generations of computing technology, is not something that
> IBM had not done before. They missed the boat, and that is history.
>


--

It was your very specific statements about what IBM could have done 5
years ago at that price that I was contesting.

Today, ignoring price, you can get the equivalent of a 30GHz system by
putting 16 2GHz processors, assuming that the motherboard is
available. The disk capacity is easy, just put 4 250GB drives in the
case. And there are already computers with over 16GB of RAM.

I won't dispute that you could have generated the equivalent to that
computer, assuming a multiple CPU motherboard, 5 years ago. I'll even
accept the $40,000 pricetag, never pricing out computers like that
back then.

Of course, once you go massively multiprocessor you are either going
for overall system performance for many processes, like a server, or
specialty software. Your typical Windows or OS/2 PC, for example, on
the desktop, probably wouldn't fully use 16 or 32 CPUs. A web
browser use one or 2, mail 1 or 2, antivirus (on windows) one, if you
have a compile going on in the background 1 or 2 more. But most of
this won't peg a CPU.
.

Relevant Pages

  • Re: History--Bell toll routing machine cards--other applications? [telecom]
    ... Bell's CRT tubes seem to be similar to IBM's "Williams" tubes used as ... memory for IBM's earliest computers. ... As soon as core was perfected, IBM ...
    (comp.dcom.telecom)
  • Re: self-extracting diskette image to hard disk? (or, IBM keeping inexpensive supercomputers awa
    ... > of money, or at least license this technology to others, which would ... > also have brought in huge amounts of money, ... I know what IBM was doing, ... > claiming not only that memory would be many times larger than existing ...
    (comp.os.os2.misc)
  • Re: First-timer buying AS/400 and have questions
    ... 512mb of main memory works well. ... License keys are how IBM controls everything. ... you are at the mercy of IBM's list price. ... >> that eat up the CPU if you step over the line on interactive CPW. ...
    (comp.sys.ibm.as400.misc)
  • Re: can we break the wordwrap limit in kmail?
    ... at my knowledge autism is when people close on themselves ... At that time mass cellphone use was just starting and warnings ... more over Billy gives more than half of his money to charity. ... what is real is that 90% of computers run their OS. ...
    (comp.os.linux.misc)
  • Re: My Migrations: PC architecture
    ... unaddressable top 64K of memory - all the things that could be inserted into ... IBM doesn't learn from its mistakes. ... address space for memory mapped screens as well as the BIOS and that ... OS/2 was obviously written by MVS guys, because its internal structures used the same ...
    (comp.lang.cobol)