A Good, Inexpensive Single-Core cpu

Does Don Knuth know anything about computing?

From http://www.informit.com/articles/art...1193856&rll=1:
-------------------------------------------------------------------
...
Andrew: Vendors of multicore processors have expressed frustration
at the difficulty of moving developers to this model. As a former
professor, what thoughts do you have on this transition and how to
make it happen? Is it a question of proper tools, such as better
native support for concurrency in languages, or of execution
frameworks? Or are there other solutions?

Donald: I don’t want to duck your question entirely. I might as well
flame a bit about my personal unhappiness with the current trend
toward multicore architecture. To me, it looks more or less like the
hardware designers have run out of ideas, and that they’re trying to
pass the blame for the future demise of Moore’s Law to the software
writers by giving us machines that work faster only on a few key
benchmarks! I won’t be surprised at all if the whole multithreading
idea turns out to be a flop, worse than the "Itanium" approach that
was supposed to be so terrific—until it turned out that the wished-
for compilers were basically impossible to write.

Let me put it this way: During the past 50 years, I’ve written well
over a thousand programs, many of which have substantial size. I
can’t think of even five of those programs that would have been
enhanced noticeably by parallelism or multithreading
-------------------------------------------------------------------

I'm not in touch with the industry, I just build a little desktop
system every few years.

Does Intel (or AMD for that matter) make A Good, Inexpensive Single-
Core cpu that doesn't heat the entire county (like Prescott)?? :-)

Any/all info much appreciated.

Puddin'

" ... and the bees made honey in the lion's head."
- from "If I Had My Way", Blind Willie Johnson

You're taking the quote out of context by omitting his disclaimer; in the
preceding paragraph:

Knuth: "So I decided long ago to stick to what I know best. Other people
understand parallel machines much better than I do; programmers should
listen to them, not me, for guidance on how to deal with simultaneity."

Puddin' Man wrote:
> I'm not in touch with the industry, I just build a little desktop
> system every few years.
>
> Does Intel (or AMD for that matter) make A Good, Inexpensive Single-
> Core cpu that doesn't heat the entire county (like Prescott)?? :-)

Even a Q9450 will use less than 10W if you're not doing much with it.

Other major considerations: power supply efficiency, graphics card power,
boiling more water in a kettle than you actually need.

If you do build a new computer, what would you actually want to use it for?
Only by knowing that can we make sensible suggestions.

Andrew

On Tue, 29 Jul 2008 09:07:36 +0100, "Andrew Morton" <> wrote:

>You're taking the quote out of context by omitting his disclaimer; in the
>preceding paragraph:

I don't think so ...

>Knuth: "So I decided long ago to stick to what I know best. Other people
>understand parallel machines much better than I do; programmers should
>listen to them, not me, for guidance on how to deal with simultaneity."

For me, that part of the interview plays kinda like:

"Well, you've written tons on serial processing. Will you now write/
publish algorithms on parallelism?

"No. It would take forever, the way I write, and it changes too quickly.
There are others devoting their time to the subject: anyone who
thinks it will be relevant for them is free to look to the others for
guidance."

Then he goes on to explain how parallelism is largely irrelevant
for his own common computer usage.

>Puddin' Man wrote:
>> I'm not in touch with the industry, I just build a little desktop
>> system every few years.
>>
>> Does Intel (or AMD for that matter) make A Good, Inexpensive Single-
>> Core cpu that doesn't heat the entire county (like Prescott)?? :-)
>
>Even a Q9450 will use less than 10W if you're not doing much with it.

What part of "Inexpensive" is it that you don't understand? :-)

>Other major considerations: power supply efficiency, graphics card power,
>boiling more water in a kettle than you actually need.
>
>If you do build a new computer, what would you actually want to use it for?
>Only by knowing that can we make sensible suggestions.

It was kinda implied that my common usage would be very similar to Knuth's,
but ...

Desktop system, Windows, Email client, Usenet client, editor, multiple
instances of IE6, occasional programming.

Puddin'

" ... and the bees made honey in the lion's head."
- from "If I Had My Way", Blind Willie Johnson

Puddin' Man wrote:
>>> Does Intel (or AMD for that matter) make A Good, Inexpensive Single-
>>> Core cpu that doesn't heat the entire county (like Prescott)?? :-)
>>
>> Even a Q9450 will use less than 10W if you're not doing much with it.
>
> What part of "Inexpensive" is it that you don't understand? :-)

Sorry, I didn't emphasise the "even" part enough: what I was trying to imply
is that even the more powerful Intel CPUs now consume considerably less
power than the Prescott currently heating my office.

>> If you do build a new computer, what would you actually want to use
>> it for? Only by knowing that can we make sensible suggestions.
>
> It was kinda implied that my common usage would be very similar to
> Knuth's, but ...
>
> Desktop system, Windows, Email client, Usenet client, editor, multiple
> instances of IE6, occasional programming.

[We're not very good at this "implying" thing, are we :-)]

For light usage, I'd suggest any of the lower price Core2 Duo processors. If
you end up using Vista as the OS, experience tells me you should get at
least 2GB of RAM. Seeing as Windows (2000/XP/whatever) will make use of dual
cores and single-core Intel processors (from what I've read, it would be
painful to use an Atom processor) aren't readily available, you might as
well get a dual-core processor.

HTH

Andrew

On Mon, 28 Jul 2008 12:09:19 -0500, Puddin' Man
<> wrote:


>I'm not in touch with the industry, I just build a little desktop
>system every few years.
>
>Does Intel (or AMD for that matter) make A Good, Inexpensive Single-
>Core cpu that doesn't heat the entire county (like Prescott)?? :-)
>
>Any/all info much appreciated.

Intel Celeron 420/430/440:
New Low-End Vs. Old Middle-End
http://www.digit-life.com/articles2/...440-page1.html

Intel Celeron 440 vs. AMD Sempron 3600+
2-GHz Budget CPUs Battle It Out
http://www.legionhardware.com/document.php?id=663

>
> Puddin'
>
>" ... and the bees made honey in the lion's head."
> - from "If I Had My Way", Blind Willie Johnson

Puddin' Man wrote:
> I'm not in touch with the industry, I just build a little desktop
> system every few years.
>
> Does Intel (or AMD for that matter) make A Good, Inexpensive Single-
> Core cpu that doesn't heat the entire county (like Prescott)?? :-)
>
> Any/all info much appreciated.


There are still single-core processors available from the two major
processor companies, such as the AMD Sempron, or the Intel Celeron. Both
of those single-core versions are relegated to very low-end laptop
functions.

There's also single-core specialty processors such as AMD Geode, Intel
Atom, or VIA C7. They are basically meant for the embedded market.

But trust me when I say, a dual-core is really the smallest thing you
should be looking for. Even if individual programs aren't designed for
multi-core, there is rarely ever just a single program ever running on
modern operating systems, even on simple "Internet surfing and word
processing" machines. As soon as an OS is capable of multitasking, then
it can make use of multiple threads. There's always background processes
running, and a multitasking OS is aided greatly by having multiple cores
to send tasks off to.

Yousuf Khan

On Wed, 30 Jul 2008 09:41:07 -0400, Yousuf Khan <> wrote:

>Puddin' Man wrote:
>> I'm not in touch with the industry, I just build a little desktop
>> system every few years.
>>
>> Does Intel (or AMD for that matter) make A Good, Inexpensive Single-
>> Core cpu that doesn't heat the entire county (like Prescott)?? :-)
>>
>> Any/all info much appreciated.
>
>
>There are still single-core processors available from the two major
>processor companies, such as the AMD Sempron, or the Intel Celeron. Both
>of those single-core versions are relegated to very low-end laptop
>functions.

And not really suitable for a gen'l purpose (non-cheapo) desktop?

>There's also single-core specialty processors such as AMD Geode, Intel
>Atom, or VIA C7. They are basically meant for the embedded market.

Also not suitable?

>But trust me when I say, a dual-core is really the smallest thing you
>should be looking for. Even if individual programs aren't designed for
>multi-core, there is rarely ever just a single program ever running on
>modern operating systems, even on simple "Internet surfing and word
>processing" machines. As soon as an OS is capable of multitasking, then
>it can make use of multiple threads. There's always background processes
>running, and a multitasking OS is aided greatly by having multiple cores
>to send tasks off to.

In the context of the common nickel/dime usage we've been discussing,
this is simply not true.

I originally considered the issue in detail back when "dualies" hit
the market and begat a tiny culture.

Yes, there are numerous background processes running, and a
multitasking OS can conceivably allocate differing threads to
different cores. But the threads have to be demanding material cpu
cycles for this to have a non-trivial effect. In common usage,
they don't.

I now have Email client, Usenet client, editor, multiple instances
of IE6 and some utility windows open. I am an average user. As I
type in the editor, I run Task Mngr and find (no surprise) that I am
using under 10% of cpu cycles available from my old Celeron 3ghz.
Most of the "under 10%" is consumed by Spybot S/D Teatime. For this
stuff, a 2nd (3rd, 4th, ... 15th, 16th etc) core would be about
as useful as the tits on a boar hog.

What you say is true for someone running multiple cycle-intensive
apps. Another guy running multiple i/o intensive apps would be
better off with a good single-core and spending extra $ on
hi-performance i/o subsystem.

The only reason I'd spring for dual-core for the nickel/dime usage
would be because Intel/AMD have virtually abandoned the market for
good, inexpensive single-core cpu's (for desktop). I will assume
they have done so until I have evidence to the contrary.

They have been corrupted by the hi-tech glitz that their marketing
dept's have chosen to hustle (to the exclusion of certain common
computing needs). No?

P

" ... and the bees made honey in the lion's head."
- from "If I Had My Way", Blind Willie Johnson

On Jul 28, 1:09*pm, Puddin' Man <puddingDOT...@gmail.com> wrote:
> Does Don Knuth know anything about computing?
>
Donald Knuth knows quite a lot about computing. The question is
whether he is the source to go to as computing moves away from single
core processors, and the answer is that he is not. Even he says so:

"Other people understand parallel machines much better than I do;
programmers should listen to them, not me, for guidance on how to deal
with simultaneity."

His denigration of the usefulness of multiple cores has to be viewed
in that light. Knuth claims that those who will most benefit from
multiple cores are not typical users, but we're probably more typical
than he is. Most users will get some benefit from at least one more
core. I'd advise buying a single core platform only where size,
noise, or power consumption are critical.

Robert.

On Wed, 30 Jul 2008 10:08:57 -0700 (PDT), Robert Myers <> wrote:

>On Jul 28, 1:09*pm, Puddin' Man <puddingDOT...@gmail.com> wrote:
>> Does Don Knuth know anything about computing?
>>
>Donald Knuth knows quite a lot about computing. The question is
>whether he is the source to go to as computing moves away from single
>core processors, and the answer is that he is not. Even he says so:
>
>"Other people understand parallel machines much better than I do;
>programmers should listen to them, not me, for guidance on how to deal
>with simultaneity."

I've already discussed Knuth's position. You can read or not read
previous posts as you wish.

It's not enough that Knuth is great at algorithms. He is like
"a religion/culture of one" in that he looks at computing -only-
for computing's sake regardless of what the industry, etc is saying.
He is impervious to hype.

>His denigration of the usefulness of multiple cores has to be viewed
>in that light. Knuth claims that those who will most benefit from
>multiple cores are not typical users,

They're not. Unless you belong to a "multi-core religion", it's
indisputable (for common apps). Did you run a "dualie" years ago? :-)

>but we're probably more typical
>than he is.

Speak for yourself. Near as I can tell, Knuth's common usage is -very-
similar to mine.

>Most users will get some benefit from at least one more
>core.

That's not disputed. Becomes a question of degree. Do you imagine
you could measure it (again, for common usage as we've been discussing)?
I'm not talking about a seat-of-the-pants guess or an obviously biased
estimate.

>I'd advise buying a single core platform only where size,
>noise, or power consumption are critical.

There may well be no practical alternative to multi-core for desktop,
because it appears that m-c is all they want to produce/market.

P


" ... and the bees made honey in the lion's head."
- from "If I Had My Way", Blind Willie Johnson

On Jul 30, 5:50*pm, Puddin' Man <puddingDOT...@gmail.com> wrote:
> On Wed, 30 Jul 2008 10:08:57 -0700 (PDT), Robert Myers <rbmyers...@gmail.com> wrote:
> >On Jul 28, 1:09*pm, Puddin' Man <puddingDOT...@gmail.com> wrote:
> >> Does Don Knuth know anything about computing?
>
> >Donald Knuth knows quite a lot about computing. *The question is
> >whether he is the source to go to as computing moves away from single
> >core processors, and the answer is that he is not. *Even he says so:
>
> >"Other people understand parallel machines much better than I do;
> >programmers should listen to them, not me, for guidance on how to deal
> >with simultaneity."
>
> I've already discussed Knuth's position. You can read or not read
> previous posts as you wish.
>
I wasn't soliciting your wisdom about Knuth. I've already very tartly
discussed the appropriateness of constantly referring to him when the
world of hardware he imagined is rapidly vanishing. We need to pin a
medal on the man and stop referring to him. Time to get a move-on.
Time's a wasting and Knuth has become a distraction.

> It's not enough that Knuth is great at algorithms. He is like
> "a religion/culture of one" in that he looks at computing -only-
> for computing's sake regardless of what the industry, etc is saying.
> He is impervious to hype.
>
That's just plain silly. Mathematicians don't make microprocessors.
Manufacturers do. "I've got some great ideas, if only they'd abandon
backward compatibility." This man is not in touch with reality.

> >His denigration of the usefulness of multiple cores has to be viewed
> >in that light. *Knuth claims that those who will most benefit from
> >multiple cores are not typical users,
>
> They're not. Unless you belong to a "multi-core religion", it's
> indisputable (for common apps). Did you run a "dualie" years ago? :-)
>
> >but we're probably more typical
> >than he is. *
>
> Speak for yourself. Near as I can tell, Knuth's common usage is -very-
> similar to mine.
>
> >Most users will get some benefit from at least one more
> >core. *
>
> That's not disputed. Becomes a question of degree. Do you imagine
> you could measure it (again, for common usage as we've been discussing)?
> I'm not talking about a seat-of-the-pants guess or an obviously biased
> estimate.
>
The metric (did I just say that?) that matters most is perceived
responsiveness, and it's not such an easy thing to measure. I don't
get all the responsiveness I want all the time in spite of using a
modern dual core processor, and I've heard many others say the same.
The fact that your second core is idling most of the time doesn't
matter if it keeps your browser from becoming unresponsive at awkward
moments.

> >I'd advise buying a single core platform only where size,
> >noise, or power consumption are critical.
>
> There may well be no practical alternative to multi-core for desktop,
> because it appears that m-c is all they want to produce/market.
>
And since manufacturers have become very good at minimizing power
consumption by hardware that idles, what incentive is there for you to
hunt down that single core machine, anyway? Would you buy the single
core processor if it were more expensive, as it should be?

Robert.