PCs to change radically in 2004! (The Enquirer)

Richard Hopkins already posted that he ran a Prescott in an IC7...

--

Thomas Geery
Network+ certified

ftp://geerynet.d2g.com
ftp://68.98.180.8 Abit Mirror ----- Cable modem IP
This IP is dynamic so it *could* change!...
over 120,000 FTP users served!
^^^^^^^




"John Lewis" wrote in message
...
On Sun, 1 Feb 2004 17:02:44 +0000 (UTC), "Wayne Youngman"
wrote:


Hi Ed,
I'm in the same boat as you (except PIII550), just sold my 4 month old
AMD
Thunder-clap sniff lol but I am soooo interested in the INTEL stuff. I
was gonna build this INTEL box 4 months ago but opted to devote the time
to
catching up on the AMD platform AMD (T-Breds, Bartons, nForce2) as they
represented stunning performance for a great price (and I was on a
*Tight*
budget*). Now The Northwood/Springdale/Canterwood line has dropped in
price
allot, the time is ripe for me, a keen PC enthusiast to get hands on.

WAIT for the CPU and motherboard !

The Official Intel Price Llist issued on 25 January has ZERO price
change from the 26 October 2003 issue. This is a very unusual event.
Intel is obviously delaying price changes until the formal
announcement of Prescott. The Northwood prices should tumble then.
And of course you will then find out whether or not any of the current
478-pin Motherboards are compatible with Prescott without voltage
regulator hacks, or worse.............

John Lewis


So
now I am about 3-6months behind everyone else in the INTEL loop, but I
have
£400/£600 or so to spend on some new kit. The AMD side of things (AMD64,
Opteron, FX:51) is all over the place right now, with new platforms and
sockets coming up, so I thought I would catch up on the now *mature*
INTEL
platform. . .

I'm not so sure about buying a brand new ATX tower, and ATX PSU, I did
have
a CoolerMaster Praetorian/ANTEC True-Blue 480w lined up (nearly £200 for
both), but now I may opt for an all-in-one ANTEC Case/PSU combo

SX835II - Performance Series II Mid Tower
inc 350w Smartpower PSU (£71.66)
http://tinyurl.com/ywpqg

Also with DDR-II nearly here I'm not certain I want to spend too much on
current high-speed memory, but we will see. . .
--
Wayne ][
new specs coming soon!

On Sun, 1 Feb 2004 13:03:10 -0000, "Dorothy Bradbury"
wrote:

A problem is that CPU wattages are going to ramp...
o P4 Prescott is expected to hit 150W later on

Graphics card wattages aren't standing still
o Frankly 75W is minimalist by 2005

Overall, ATX wouldn't work very well by about late 2006.

Several solutions for chip wattage are underway:
o Solving the chip level heat problem
---- cooligy heatpipe-from-the-silicon is gaining attention
---- dual-core & multi-core to reduce avg heat per area

Perfectly valid... rememember the Pentiums with the
separate core and cache-ram ? Will need some very
clever substrate-and interconnect design to minimise
latencies.

------- P4 heat output is because of stalls = low avg heat per area
------- Prescott removes stalls = high avg heat per area
---- distribute CPU around the cache re watts/area
------- cache in place of clock scaling is one next mkting area

Yes, more on-chip parallel-processing, but it will need more
cache.

------- has always attracted big margins (re dbase app benefit)
---- limit is 150W/cm^3, so balance die-shrink re size
o Solving the system level heat problem
---- BTX will come in here
------- graphics + RAM + CPU will quickly top 250W

Agreed, and within the next year. Don't forget the power
required by the video ram supporting the GPU; parallel-
processing makes for very hot devices.

------- that's into 50cfm now and probably more later
---- BTX will eventually suffer the same limits as ATX
------- heatsinks may get a lot more expensive

Graphics cards in particular are suffering high wattage growth.

The fastest power-growth of the lot over the past 3 years.

As always it will not be a single solution, but combinations.
BTX is helping the box-shifters come up with a solution more
like Dell/Apple - without having the engineering costs of such.

Watts have outperformed die-shrink & voltage-drop.
Between 2007 & 2012 we may see a 500W CPU

--- at the high-performance "desktop" end of course.

The far more modest laptop will generally see increased
functionality at the same power, with a great number
of user-settable power-saving options, and more advanced
battery technology.

It may be
the WinZP-64-bit voice driver :-) but it will probably come.

Storage is hitting tougher barriers than CPU/Graphics, we are
down to a few crystals per bit - within few yrs we need atoms.
--

Demand for true system ram is slowing down. However,
cache size and performance is becoming more and more
an overall performance bottleneck.

John Lewis

Dorothy Bradbury

"John Lewis" wrote
WAIT for the CPU and motherboard !

The Official Intel Price Llist issued on 25 January has ZERO price
change from the 26 October 2003 issue. This is a very unusual event.
Intel is obviously delaying price changes until the formal
announcement of Prescott. The Northwood prices should tumble then.
And of course you will then find out whether or not any of the current
478-pin Motherboards are compatible with Prescott without voltage
regulator hacks, or worse.............


Hi,
hehe yes I am waiting, been a week now! but I just have to order something
this week its killing me, wanna get hands on and experienced with the
reasonably new Springdale and Canterwoods with a Ht enabled P4 Northwood. .
.. . just in time for a whole new bunch of stuff for me to get curious about.
The Intel CPU prices are good right now, I feel I should buy before

a) They go out of stock
b) The price goes back up ;P
--
Wayne ][
new specs coming soon!

I have seen some new Heatsink materials out there on the market that
would
make the SLK HSF's look like they dont even work..
When it's all said and done then I dont think we'll see thermal outputs
of
150/200/250w coming from these chips.. My guess is that we'll start
seeing
IBM/AMD bring in new processing methods that companys like Transmeta
uses.

Today's joke, no doubt. Transmeta is going nowhere but down.


Yea.. Transmeta is probably going to be out of the race before long (many
years).. They did get some USGov contracts however and I think tax dollars
will help keep them floating for a while. Their Idea is somewhat wrong..
Save processing power by doing more processing in other places.. But in
reality this is a good solution to keep power levels in an area where they
can be dissipated easily. Sure if a 500w processor is out then
Heatsink/Heatpipe/ThermalPumping(organic&non) will be able to keep
everything cool enough to work, but look at the weight behind it all..

I do back the underdogs.. so I would love to see Lenslet, VIA, and Transmeta
make some serious marks in the market before they die (or drop out in the
case of VIA).. They all have what's needed for the "next" generation of
computing I think..

IF, and by what I read I cant help but say it's a BIG IF Lenslet ever gets
their kit working correctly (and small enough) then Intel and AMD will
suddently be the #2 and #3 processor mfg'rs of the world..

VIA and Transmetta are able to supply low power processors to handheld
devices out there and That's probably where their processors would shine (if
VIA would go that path)..

I cant help but think that part of this move to a new form factor has to
also do with the way devices are moving into one central product..
Xbox/Playstation/Sound Processing/Video Recording/Email/Phone..
All of the devices are starting to turn into one product.. the
PC(Everything)
We'll see how it all works out really .. but I do hope that the underdogs
show something to the world before they go ..

I mean why not.. Something as fast as a duron 500 that runs at 6w is
what
we see out NOW...

Not without process shrinks and lower core-voltages.

Some of the VIA chips run about that fast.. and with that amount of power
used.. Transmetta information is difficult to find so I cant quote
benchmarks from them..


If that same technology grows then in 2 or 3 years time we'll have
something
similar that could still be passive cooled, and be more powerful than
todays
64 3600+'s..

The idea of having everything being fed by a central fan will def work IF
the power consumption doesnt go verry high

My hopes is that the power consumptions will stay around where they are
now,
100w for the CPU and 50-100w for the video/sound processing.. Anything
more
is just overkill..

Agreed, but the performance demands are outstripping the process
development, so we will certainly go through a cycle of very high
power, just as when the first toaster-oven Pentiums were released,
only to be replaced a year later with parts clocking faster and
dissipating 1/2 of the power.

John Lewis


Yea.. The Performance demands SHOULD slow down eventually..
I mean think about it.. When billy's computer can handle a real time voice
conversation.. Can give a Virtual Image that's damn near real quality.. and
other good stuff that I cant even guess aobout.. then people will start to
say.. Hey who needs more..
I dont know how long we are from such a thing but I would guess it's not
going to be THAT long.. You can only get TO realism and not go beyond it or
I would feel otherwise..

I give PC's another 10/12 years of this "faster is better" stuff.. Then It
will all go to making "Quality" more of a priority

course
I could be wrong
and intel might be planning a 18Thz processor for 2020

Good points Wayne but I must move up now! Waiting for my box to boot is
like watching grass grow :-) It'll probably take a month to get all the
parts together so I'll hold off on the mobo and CPU until the end. Right
now I just can't seem to find the perfect full tower case (must have 6
external 5.25s) + mobo tray + other stuff. Good luck on your choices and
let us know what you finally wind up with.
--
Tally Ho!
Ed
"Wayne Youngman" wrote in message
...

"Ed Forsythe" wrote
I'm gonna build my new box anyway. New Lian Li case, PS, 2.8C CPU, DDR,
etc.
I can't wait for the BTX form factor to appear since I *must* upgrade my
old
Coppermine 600E. I'm betting/hoping that Intel won't make the same
mistake
they did with the Microchannel architecture or Rambus. I paid $5000 for
the
first (as far as I know) PS2 computer sold in Washington, DC. Then, a
short
time later, I paid $350 for an ATI Microchannel graphics card. That was
*huge* money in those days! All of the MC stuff was extremely expensive.
I'm hoping that the BTX mobos will fit in an ATX case with a new
backplane
and that they will have an AGP slot as well as a PCI Express. I can
deal
with having to buy new memory. I'm fairly confidant that my PCP&C PS
will
work (so is Anand). Flipping form factors too rapidly could lead to the
big
guns expanding their AMD stuff?


Hi Ed,
I'm in the same boat as you (except PIII550), just sold my 4 month old AMD
Thunder-clap sniff lol but I am soooo interested in the INTEL stuff. I
was gonna build this INTEL box 4 months ago but opted to devote the time
to
catching up on the AMD platform AMD (T-Breds, Bartons, nForce2) as they
represented stunning performance for a great price (and I was on a *Tight*
budget*). Now The Northwood/Springdale/Canterwood line has dropped in
price
allot, the time is ripe for me, a keen PC enthusiast to get hands on. So
now I am about 3-6months behind everyone else in the INTEL loop, but I
have
£400/£600 or so to spend on some new kit. The AMD side of things (AMD64,
Opteron, FX:51) is all over the place right now, with new platforms and
sockets coming up, so I thought I would catch up on the now *mature* INTEL
platform. . .

I'm not so sure about buying a brand new ATX tower, and ATX PSU, I did
have
a CoolerMaster Praetorian/ANTEC True-Blue 480w lined up (nearly £200 for
both), but now I may opt for an all-in-one ANTEC Case/PSU combo

SX835II - Performance Series II Mid Tower
inc 350w Smartpower PSU (£71.66)
http://tinyurl.com/ywpqg

Also with DDR-II nearly here I'm not certain I want to spend too much on
current high-speed memory, but we will see. . .
--
Wayne ][
new specs coming soon!

"Ed Forsythe" wrote
Good points Wayne but I must move up now! Waiting for my box to boot is
like watching grass grow :-) It'll probably take a month to get all the
parts together so I'll hold off on the mobo and CPU until the end. Right
now I just can't seem to find the perfect full tower case (must have 6
external 5.25s) + mobo tray + other stuff. Good luck on your choices and
let us know what you finally wind up with.


Hi,
yes I can understand your need for speed. I went through that *wow* factor
about 3 months ago when I built an AMD machine (2500+ @ 3200+, SATA RAID 0,
Radeon 9800 etc). The booting time is very fast on a new system, you will
love it.

The only nice cases I saw that had *six* 5.25" bays were,

Lian-Li PC70
http://www.lian-li.com/product.php?a...ewPD&prdid=374

Lian-Li PC71 (Same as PC70 but black)
http://www.lian-li.com/product.php?a...ewPD&prdid=375

Lian-Li PC75 (Same as PC70 but side-panel)
http://www.lian-li.com/product.php?a...ewPD&prdid=376

what you think of them?
--
Wayne ][
new specs coming soon!

rstlne wrote:
[...]
IF, and by what I read I cant help but say it's a BIG IF Lenslet ever
gets their kit working correctly (and small enough) then Intel and
AMD will suddently be the #2 and #3 processor mfg'rs of the world..

Lenslet makes DSPs. Intel and AMD make general CPUs. The two are at opposite
ends of the computing spectrum and will almost certainly never compete
against each other.

[...]
--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more
Add michael@ to emboss.co.nz - My inbox is always open

There is also an issue with software design. The software designers have
not been able to provide saleable solutions for new hardware technology in a
timely manner. From the 8086 to 80486 software was not a big issue due to
the continuation of ISA slots. Then there was the PCI which require
software changes. Then APG, which for a long time still used PCI
programming. Now we have USB2.0 and firewire. There is still no 64 bit main
stream OS or programs available to the public at reasonable pricing,
although AMD and Intel are releasing 64 bit processors. Now we have a new
SATA, BTX, or PCI Express architecture with no software that will optimize
its features. SATA has been out for some time and we are just now seeing
good driver releases. The hardware manufactures need to come up with a
better way to inform software engineers as to future developments. The
fastest hardware will only be as good as the software it interfaces with.

"David Maynard" wrote in message
...
- HAL9000 wrote:
Thanks for the link. Since everyone is weighing in...

The technology change from a 4040 to a 8086 was a major change in cpu
architecture.

The technology change from a 8086 to a 80286 was a major change in
cpu architecture.

The technology change from a 80286 to a 80386 was a major change in
cpu architecture.

The technology change from a 80386 to a 80486 was a major change in
cpu architecture.

The technology change from a 80486 to a Pentium was a major change
in cpu technology. What was it, about 1994 when the Pentium 66 came
out? About ten years ago?

Since the inception of the Pentium, cpu architecture hasn't changed
much. Mainly the transistors are now just getting smaller and this
enables progressively higher and higher speeds. Testament to this
fact is the Athlon. It's architecture has been around a long, long
time. The architecture of the Athlon 64 isn't changed much either.
As far as I know, the P2, P3, and P4 have all been marginal design
improvements. Hardly worth the money for the new architecture, just
the speed improvements have had any real worth - just like in the
Athlon.

Having said this, there isn't a performance reason to update your
system in the next few months. The performance improvement is going
to be marginal *at best* if you do. Almost all of the stuff
mentioned in the article are just new connectors. Who needs new
connectors ?? You don't need SATA, BTX, or PCI Express. These are
primarily just connector changes with no improvement in the "weakest
link" in the performance chain.

Manufactures change things to force you to get a whole new system
instead of making a minor upgrade to the system you have now. They
want money from you, hehehe, and they think up ways to get your
money.

Everybody loves a good conspiracy theory but, as an
engineer/designer/product manager, I can tell you that designers change
things because they think it's a 'better solution' and it's the finance,
manufacturing, and marketing departments who restrain them from doing so
with their normal predilection of every time a new thought pops into
their head, meaning almost hourly, and who place reverse compatibility
requirements on them.

Finance doesn't like new designs because it costs money. Manufacturing
doesn't like new designs because it means retooling, retraining, and
obsolete stock problems. And while you might think Marketing does they
like to sell upgrades and 'making the user buy the whole thing new'
doesn't work very well as it tends to cause consumers to delay
purchasing, because they have to 'buy the whole thing new' (people hate
throwing away things that still work). 'The whole thing new' also risks
customers changing to the competition since, if they have to buy it ALL
'new', they no longer have a reusable investment in existing product,
stock, training, et al, to restrain the switch.



Now for what you might actually need in the next few years :-) You
should expect to need 3 or 4 times more memory then you have now.
Forget everything else, its just superficial. Plan your next system
around a memory upgrade. I would recommend looking at DDR 2 memory
where you can get at least a times two bandwidth improvement over
what you have now. If you can't get a times two improvement - don't
get it. The speed improvement will be too marginal to be worth it.

Forrest

Motherboard Help By HAL web site: http://home.comcast.net/~hal-9000/



On Fri, 30 Jan 2004 15:50:01 +0000 (UTC), "Wayne Youngman"
wrote:


Hi,

I'm sure most of you know about these changes, but it seems 2004 is
*INTELS* year to go WHOOOOSH! Makes me question spending £200.00
on a new case and PSU (sniff bye bye coolermaster/lian-li ATX case
and ANTEC TrueBlue 480w)

www.theinquirer.net/?article=13425

Rusty Smith wrote:
[...]
There is also an issue with software design. The software designers
have not been able to provide saleable solutions for new hardware
technology in a timely manner.
[...]
There is still no 64 bit main stream OS or programs available to the
public at reasonable pricing, although AMD and Intel are releasing 64
bit processors.
[...]
The hardware manufactures need to come up with a better
way to inform software engineers as to future developments. The
fastest hardware will only be as good as the software it interfaces
with.

OS's than Windows, you know It's no secret the MS is lagging behind badly
when it comes to AMD64 support, mainly because it's a typical big company,
and it takes a while to get several thousand feet marching all together
again when something changes. Linux was stable on AMD64 by the time the
Athlon64 was released (though not really stable at the Opteron release),
thanks to AMD providing complete documentation, a CPU simulator and
pre-release test hardware. There's not too much more that you could ask for
.... Likewise for PNI. Linux already has support for the hyperthreading
enhancements that come with Prescott (had it 4 months ago actually) but it
looks as though we're going to have to wait for Longhorn to get the support
in Windows.

Note that I'm not saying that Linux is mainstream (which depends on whether
you're looking at servers or desktops, and also your definition of
mainstream) but it's definately unfair to put the blame on AMD for the delay
in getting an AMD64 version of Windows out the door. Especially as they
already have a version (for the Itanium) that is fully 64-bit.

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more
Add michael@ to emboss.co.nz - My inbox is always open

Don't forget the power required by the video
ram supporting the GPU

I had, and it's quite significant - all adds up.

Yes cache ram will become a tool for differentiation;
cheapish re R&D/cost to add to CPUs vs other changes.
Additionally system ram's profit margin is minimal, and at
the same time 512 benefit is small for most.

I recall paying ==800$US for 16MB of RAM 8yrs ago.

Storage is hitting the atoms problem first (disk, not tape which is
miles off), then CPUs, and then getting the heat avg down per CPU.

Re 64-bit support, lag will be about 6yrs which puts us at 2009 for
mass-adoption phase. That may change, since I suspect adoption is
getting quicker at the consumer end, at the business end it's getting
slower at the replacement end unless it gives cost/revenue benefit.


The big move at consumer end will be heat/portable; and already
the P4-M is putting a lot of power - with less heat - in laptops. The
generation after will make laptop-as-powerful even more acceptable.

Home *.net appliances are an interesting area - do they stay as the
mini-heat/power/size/noise, or do they become the grunt for laptops
to thin-client over faster wireless remote-desktops for more power?
Faster wireless will tend to pull the latter, faster laptops tend to keep
the *.net appliance as silent 24/7 low-power.
--
Dorothy Bradbury