P4 Extreme Edition discontinued

[Hank Oredson]

Who would buy something that slow?

A collector of classic systems.

Falls between a typical 8080 (2 MHZ)
and the wonderfully fast 8088 (4.77 MHZ).

--

... Hank

http://horedson.home.att.net
http://w0rli.home.att.net

 

[Hank Oredson]

It's all in the IPC, MHz is just a myth. ;p

So given an issue width of around 1000 ...

--

... Hank

http://horedson.home.att.net
http://w0rli.home.att.net

 

[Grumble]

Chips with such a huge cache are very expensive to make. Production
of more complex chips probably results in a larger percentage of
chips that need to be discarded. Much fewer of the more complex chips
fit on a wafer since the die size is so large relative to chips with
a much smaller cache.

http://endian.net/details_compare.asp?ItemNo=3951&ItemNo=3153&ItemNo=3429&ItemNo=331

Gallatin
~24 KB L1, 512 KB L2, 2048 KB L3
178 million transistors (~120 million for L3)
237 mm^2 @ 130 nm

Gallatin is not much larger than the Opteron (193 mm^2 @ 130 nm).

Prescott
~32 KB L1, 1024 KB L2
125 million trnasistors
112 mm^2 @ 90 nm
1.116 Mtransistors / mm^2

Dothan
64 KB L1, 2048 KB L2
140 million transistors
83.6 mm^2 @ 90 nm
1.675 Mtransistors / mm^2

AFAICT, a 90 nm Gallatin would take 110-140 mm^2, in other words, it
would be cheaper to manufacture than an Opteron.

Why hasn't Intel shrunk the Gallatin core?

 

[George Macdonald]

Sorry, I can't help it, I've been brainwashed. ;p

What? You're not going to blame the AMD "gang of thugs" in this NG for
your err, bias?

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Ed]

What? You're not going to blame the AMD "gang of thugs" in this NG for
your err, bias?

I blame Intel.

Cheers,
Ed

 

[Judd]

Where did you see that?

It's been in the news for some time. Q4 announcement likely to come soon.
925XE is supposed to have the 1 GHz bus and probable support for DDR2-667.
The new EE chip has a 2 MB L2 cache kind of like the Dothan's and is on a
90nm part. Should be a screamer. It's the first of the 700 series of
Pentiums.

 

[chrisv]

I don't understand why the EE costs so much, how much more could it cost
them to make a EE, I thought Intel chips were supposed to get faster and
cheaper, 300mm wafers @ 90nm, what's the problem? Just pure greed?

They don't want them putting price pressure on their mainstream chips.
It's not like they really want to sell many EE's - they just want
something that looks good in the benchmarks, so it's less apparent
that they are getting their butts kicked by the A64.

 

[George Macdonald]

It's been in the news for some time. Q4 announcement likely to come soon.
925XE is supposed to have the 1 GHz bus and probable support for DDR2-667.
The new EE chip has a 2 MB L2 cache kind of like the Dothan's and is on a
90nm part. Should be a screamer. It's the first of the 700 series of
Pentiums.

"EE edition"... "XE edition"... sounds like car mfrs when they have a
turkey on their hands.... 'coming to you from the performance division of
Intel'?<gulp>. So is this thing going to have EM64T and will the 925XE
have the 36-bit data on FSB?

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Judd]

"EE edition"... "XE edition"... sounds like car mfrs when they have a
turkey on their hands.... 'coming to you from the performance division of
Intel'?<gulp>. So is this thing going to have EM64T and will the 925XE
have the 36-bit data on FSB?

Rgds, George Macdonald

I'm almost sure it will have EM64T as it seems Intel will finally release it
for Prescott. I presume it will have the kludgy 36-bit Xeon MP borrowed
data too... or is it 40-bit?

 

[Keith]

I blame Intel.

Good plan. Go after the deeep pockets! ;-)

 

[Grumble]

Where did you see that?

http://anandtech.com/cpuchipsets/showdoc.aspx?i=2117
http://anandtech.com/cpuchipsets/showdoc.aspx?i=2152

 

[George Macdonald]

I'm almost sure it will have EM64T as it seems Intel will finally release it
for Prescott. I presume it will have the kludgy 36-bit Xeon MP borrowed
data too... or is it 40-bit?

It's 36-bit and according to the data sheets, the 925X does not have the
extra 4 bits on FSB. Can they add them for 925XE?... I dunno but from my
POV the marketing people at Intel are fiddling with disaster.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

 

[Tony Hill]

I saw prices around $825 for the P4 3.2 EE. Still too close to the
$1,000 P4 3.4 EE.

The two prices I quoted were from www.newegg.com, a nice handy shop
that I often use for quoting prices. I'm sure other spots might have
one chip or the other for a bit cheaper, particularly if you check the
Pricewatch bottom-feeders.

Imo the Athlon FX-51 is being discontinued since
the Athlon 64 3800+ is probably less expensive to make than the
FX-51, and seems to perform at least as well in almost all benchmarks.

Yeah, that's part of it too. Same deal goes for Intel's P4EE vs. P4
line. The 3.6GHz P4 beats out the 3.2GHz P4EE in most benchmarks but
is cheaper to produce and to sell. Since presumably this chip is
going to enter wide-scale production and availability soon, it makes
sense to discontinue the 3.2GHz P4EE.

With the prices and market segments for these Extreme Edition and FX
chips it really only makes sense to have a single, top-of-the-line
chip at any given time. Anything else is just too expensive and no
one is going to buy them.

 

[Tony Hill]

I don't understand why the EE costs so much, how much more could it cost
them to make a EE, I thought Intel chips were supposed to get faster and
cheaper, 300mm wafers @ 90nm, what's the problem? Just pure greed?

Production price has VERY little to do with selling price, it almost
all comes down to market factors, product positioning and all those
other fun things that us CS, IT and engineering types like to avoid at
all costs.

I just ordered a nf3-250 mobo and 3200+ newcastle for under $300, I just
couldn't buy an Intel, not really sure why, must be because all my
friends have A64s, funny cause they wouldn't touch an AMD just a couple
years ago.

Peer pressure on CPU purchases? Eh, why not... It's worked for cars,
clothes, stereo equipment and any number of other things... :>

 

[Me]

This article states that the original P4EE is being discontinued.

Who cares? I think $100 is too much for a CPU. But $400+??! Ouch.
Indy cars are cool, but not very practical as grocery-getters. (I'm not a
3D gamer or anything like that though, but I think that segment may get
too much press rather than more practical computing).

AJ

 

[KR Williams]

Who cares? I think $100 is too much for a CPU. But $400+??! Ouch.

You aren't in the target market. Who cares?

Indy cars are cool, but not very practical as grocery-getters. (I'm not a
3D gamer or anything like that though, but I think that segment may get
too much press rather than more practical computing).

Perhaps you're right, though the g(l)amers are the ones who *do*
need performance. There are others in that category, but to hold
your nose up at those who do want to spend the bux is a tad lame,
don't you think? Yes, I bought an Opteron 144, pretty much
because I can. ;-) ...and no, I don't do games either.

 

[Alan Walpool]

KR> You aren't in the target market. Who cares?

KR> Perhaps you're right, though the g(l)amers are the ones who *do*
KR> need performance. There are others in that category, but to hold
KR> your nose up at those who do want to spend the bux is a tad lame,
KR> don't you think? Yes, I bought an Opteron 144, pretty much
KR> because I can. ;-) ...and no, I don't do games either.

Another reason to buy a faster CPU would be to increasing overall
system lifetime. I find it amazing these days that cases, and cpu fans
can cost more than memory or processors.

Later,

Alan