Archived from groups: alt.comp.hardware.overclocking (
More info?)
"ElJerid" <s.vanderhaeghen.nospam@pandora.be> wrote in message
news:qgzyd.8169$VC.339084@phobos.telenet-ops.be...
>
> "Richard Hopkins" <richh@dsl.nospam.co.uk> wrote in message
> news:41c9bb85$0$16582$cc9e4d1f@news-text.dial.pipex.com...
> > "Rich" wrote in message...
> > > Not sure if I know my subject well enough to really be mentioning
> > > this
> >
> > Erm, no, not really. But you do get brownie points for trying. ;-)
> >
> > > but it appears that, after scrolling through messages in this group,
> there
> > > are many ppl who are concerned about the high heats being generated
> > > from their Prescotts, and problems associated with those high heats...
> > > mainly stability.
> >
> > Not sure where you get the idea that stability is an issue. To the best
of
> > my knowledge there has only been *one* solitary thread here where the
guy
> > couldn't get the processor stable at stock speeds, and I'm afraid one
> > specimen doesn't make a case, especially when we don't know the full
> detail
> > about it. It could simply be that that particular user had done
something
> > wrong or that one of the components in the system concerned was faulty.
> >
> > > Computer corporations and affiliates getting into legal binds is not
> new.
> >
> > Of course. However, to trip Intel up in this case, you'd have to prove
> > they'd done something wrong. Unfortunately selling a bloated,
> power-hungry,
> > inefficient CPU design is not in itself illegal provided said CPU
> satisfies
> > its published specifications, which Prescott does.
> >
> > > Fujitsu recently had a litigation against them for manufacturing and
> > > releasing defective drives.
> >
> > Your point being? Fujitsu got in trouble because they sold drives that
> > *didn't* satisfy their published (MTBF) specification. Prescott's heat
> > output isn't a defect, it's a design feature, albeit an undesirable one.
> The
> > thermal design power figures are published by Intel, so you can't claim
it
> > to be a fault.
> >
> > > Now it appears maybe Intel should be looked at.
> >
> > On what grounds? You've got no claim on product description or fitness
for
> > purpose grounds, so unless Prescotts suddenly start failing in large
> number,
> > you don't have a leg to stand on.
> >
> > > Judging from the temps we are seeing in here, although overclocked...
> >
> > Er, you can't form a legally grounded argument from anecdotal evidence
> > gained from people who run stuff outside the manufacturer's recommended
> > specification.
> >
> > > the readings are higher than they've ever been.
> >
> > Of course they are, but this is hardly something you can blame on the
> > Prescott, or indeed single out Intel for. Component counts and power
> inputs
> > have been on an upward curve ever since the 8086. Don't forget that
until
> > the 486 came out, X86 CPU's didn't usually need a heatsink at all, while
> AMD
> > have also had big problems with the .9u fab process.
> >
> > > These prescotts run extremely hot even at normal settings!
> >
> > Yes, and they're supplied with a heatsink designed to dissipate that
> amount
> > of power, along with a very specific set of installation
> > criteria/instructions, so your argument, in legal terms, is moot.
> >
> > > When does the heat start to become a fire risk,
> >
> > PMSL, are you serious? The fastest Prescotts consume 110W, are buried
> > underneath a large metal heatsink, inside a metal case, and are equipped
> > with a couple of layers of electronic overheat protection. There's less
> fire
> > risk there than there is in something like an incandescent lightbulb.
> >
> > > much less the thought of losing valuable data?
> >
> > Again you have no argument. At default speeds, Prescotts are stable. If
> > you're talking about overclocked CPU's, they all get unstable if you
push
> > them too far...
> >
> > > Just a thought.....
> >
> > Exactly.
> > --
> >
> >
> > Richard Hopkins
> > Cardiff, Wales, United Kingdom
> > (replace .nospam with .com in reply address)
> >
>
> Nevertheless, behind those thoughts, another fact starts to appear. With
> actual technology, the cpu manufacturers seem to have reached a speed
> barrier. There has been a time that every twelve months, the cpu frequency
> was +/- doubled, and there was no visible limit to this trend. Since a few
> months however we are hanging around 3 GHz. Sometimes a little higher, but
> in "unconfortable' circumstances. If you want to run the actual highest
> speeds, you will have to add some esoteric cooling systems and a lot of
> fans. Your PC has become a microwave. The total system power consumption
> will raise from an average of 60 W to more than 300 W. Imagine the waste
in
> an office with more than 200 or 300 systems. But what about the resulting
> floor productivity uncrease? The answer is simple: zero.
> Even if we have a look to benchmarks, it appers that the latest super PC's
> are only a few percents faster than the previous generation (I mean f.ex.
a
> P4 at +/- 2.8 Ghz with a i865PE chipset).
> This frequency barrier is apparently accepted by Intel and AMD. Intels
stops
> (for the moment) the Ghz race and announces to focus on parallel
processing
> for their next generation. AMD was the first to change the cpu names that
> included the frequency, and was followed by Intel.
> Technollogy evolves as a staircase. Today we are on the flat part.
Wondering
> what will be the next cpu technology that will give a new platform for
> future developments. Press articles reveal some trends, but what will it
be?
> Biocomputers? Quantum computers?
>
>
For the simple explanation on some stuff, just go to the bottom.
Quantum Physics is already in development.
Some of what it might be able to do is control gravity, and also be the
strongest
encryption ever to be developed.
Next Generation CPU's will feature something called "Tri~Gate Tech"
If I got that right, those should be able to go over the 5 GHz limit.
10 GHz is very possible.
Only thing I can think of is that Tri~Gate will be bused 6 Times.
or act like 3 or 4 CPU's ??
I'm not really sure @ all for what the Tri~Gate Tech will mean.
I have only guess's.
The first Warp engine is also in development.
Surf around
http://www.howthingswork.com to find more info on this.
Intel Does have a 3.6 GHz CPU by the way.
It's a 775 LGA (Prescott)
I saw some1 on this same thread claim that we're still hanging around
the 3 GHz range.
3.6 GHz is most definitely getting closer to the 4 GHz mark.
And as far as I know, there are already prototypes of the Prescott that go
even faster in GHz.
I have no idea for "Bandwidth"
I do figure they have the same number of transistors.
What r we up to? Something like 15 or 25 Billion?
that seems about right for some reason.
The P4 (Northwood) I think has 10 Billion
And the one I'm typing this on I think is about 6 or 8 Billion.
He he, I have a good idea on this stuff, but I cant fix a dishwasher.
That's @ least what them ppl say to me @ work when I ask them
cant any one of us fix the Dish Washer.
They act like I couldn't possibly fix the thing.
Like it would be to hard for me to understand or something.
There's probably not even 100 transistors in a dish washer.
And from what I can remember, there's either 2 or 3 relays.
I figure all it really is, is a simple timing chip, with relays &
regulators.
Back to it..
In simple, look @ everything available on the market,
then just figure on 200 - 400 MHz more for Prototypes.
That goes for everything including AMD.
Denny. ;-)