P4C-E Deluxe: Northwood or Prescott

[Z Man]

I am looking to upgrade my P4 1.6GHz CPU. I was going to upgrade to a 3.4C
(Northwood), but now that the Prescott is available, it is an option I
should consider. I am running 4GB PC2100 DDR, which I will continue to use.

Here's the newegg.con pricing ("E" is Prescott):

2.8E OEM $185

2.8C OEM $179

2.8C Retail $188

3.2C Retail $290

Will my motherboard support the Prescott? If so, which option is best?

 

[Paul]

I am looking to upgrade my P4 1.6GHz CPU. I was going to upgrade to a 3.4C
(Northwood), but now that the Prescott is available, it is an option I
should consider. I am running 4GB PC2100 DDR, which I will continue to use.

Here's the newegg.con pricing ("E" is Prescott):

2.8E OEM $185

2.8C OEM $179

2.8C Retail $188

3.2C Retail $290

Will my motherboard support the Prescott? If so, which option is best?

Here is the conclusion page from the Tomshardware review:

http://www.tomshardware.com/cpu/20040201/prescott-30.html

"In our opinion, the debut of a processor based on a cutting-edge
process technology and architecture that only offers the same
performance level as the predecessor is unanticipated. Despite 1 MB
L2 cache and some optimizations, Prescott is slower than Northwood
in roughly a third of our benchmarks. Software, like many 3D shooters
and more serious applications like Lame, MS Movie Maker 2,
Mathematica, Cinema 4D or even 3DStudio perform worse than
before."

The claim is, that the Prescott has more pipeline stages. That means,
any time an unexpected branch occurs, more work is thrown away. It
seems almost reminiscent of when P4 came out, compared to P3. At the
same clock speed, the P3 stomped the P4. In the end, the P4 won, by
clocking to much higher speeds than the P3.

The Prescott will have a decisive advantage when it is clocked higher.
At that time, a new socket will be introduced (that can presumably
handle the heat and current flow). I would find a Northwood with as
much cache on it as you can afford, for the time being. (Being
careful, of course, to make sure the OS is actually activating and
using the cache. OS are tricky that way - sometimes they ignore
the extra cache. Find a benchmark with an ever expanding memory
footprint, to chart the cache "breakpoints" versus memory block size,
to be sure everything is working.)
PCMarks FPS Norm Price Cache
P4 EE 3.4GHz PCMark 2004 Overall = 5502 Quake3 = 256.6 129% $999 (2M L3)
P4 3.4E $417 (1M L2)
P4 3.4Ghz 5255 = 231.4 117% $417 (.5 L2)
P4 EE 3.2GHz 5111 = 244.2 123% $925 (2M L3)
P4 3.2E Prescott 5018 = 215.5 109% $278 (1M L2)
P4 3.2/800 Northwood 4978 = 223.3 113% $278 (.5 L2)
P4 3.0/800 Northwood 4732 = 212.7 107% $218 (.5 L2)
P4 3.06/533 Northwood 4717 = 198.4 100% $218 (.5 L2)
P4 3.0E Prescott 4704 = 203.8 103% $218 (1M L2)

Chart is sorted by highest PCMarks. FPS shown for gaming comparison.
PCmark http://www20.tomshardware.com/cpu/20040201/prescott-19.html
Quake3FPS http://www20.tomshardware.com/cpu/20040201/prescott-10.html#opengl
Normalized column makes the 3.06/533 Northwood FPS equal to 100%.
Prices are wholesale http://www.intel.com/intel/finance/pricelist/

That is a pretty expensive performance curve.

Overclocking a cheap processor to 3.6Ghz is also an option.

Some people have managed to get a 3.0C to run at
3.8 on air. See http://www.cpudatabase.com/CPUdb ,
which still doesn't have any entries for the high
end stuff.

Paul

 

[daytripper]

On Thu, 12 Feb 2004 21:26:07 -0500, (e-mail address removed) (Paul) wrote:
[another brilliant essay from Tom's - flushed to where it belongs]

Being careful, of course, to make sure the OS is actually activating and
using the cache. OS are tricky that way - sometimes they ignore
the extra cache.

You'd think when people read something like the above from a self-proclaimed
tech site they'd realize the "quality" of the authors - and stay the hell
away....

/daytripper

 

[Z Man]

On Thu, 12 Feb 2004 21:26:07 -0500, (e-mail address removed) (Paul) wrote:
[another brilliant essay from Tom's - flushed to where it belongs]

Being careful, of course, to make sure the OS is actually activating and
using the cache. OS are tricky that way - sometimes they ignore
the extra cache.

You'd think when people read something like the above from a self-proclaimed
tech site they'd realize the "quality" of the authors - and stay the hell
away....

The current issue of PC Magazine has a front page article on the Prescott,
and they also mention somewhat degraded performance resulting from
additional pipeline stages. My initial feeling, not supported by objective
data or benchmarks, is that I am better off buying the newest Intel
technology unless it can be shown to be markedly worse-performing that
earlier technology. Having said that, I am still uncertain as to what course
to take. The 3.2C has been around for a while. If there were any bugs, they
probably would have come to our attention by now. The "E" is so new that we
don't know what might show up, and it may not be worth taking the risk if
there is no demonstrable benefit. Bottom line, having waited this long with
my aged 1.6GHz CPU, I am going to wait it out just a tad longer to see what
develops. The price of the 3.2C has dropped from around $125, from around
$425 to under $300 in about the past sixty days (newegg.com pricing), so I
have already derived a benefit.