Intel quietly ships 64-bit Prescott

[Yousuf Khan]

Apparently the 64-bit Pentium 4's are now ready. And they're server parts!?

http://www.eetimes.com/semi/news/showArticle.jhtml?articleID=26805631

Yousuf Khan

 

[JK]

Server parts? So they must be quite expensive? If they aren't very expensive,
then they will probably put tremendous downward pressure on the prices of
Intel's 32 bit processors. I wonder about the power consumption and about
the performance.

 

[daytripper]

Apparently the 64-bit Pentium 4's are now ready. And they're server parts!?

http://www.eetimes.com/semi/news/showArticle.jhtml?articleID=26805631

Not all "server" parts are created equal - the Prescotts are more likely just
desktop parts that will be used in some blades and pizza boxen (read:
seriously cost conscious products) vs the Noconas (Xeons) that will go in less
cost-sensitive applications.

The somewhat interesting part of this is that Intel is going along with
desktop processors being used in servers. They've had a major malfunction when
we thought out loud about doing the same with earlier P4s...

/daytripper ("Nancona"? Another fine job by EE Times! ;-)

 

[Rob Stow]

Apparently the 64-bit Pentium 4's are now ready. And they're server parts!?

Huh ? You seem surprised ? Intel has been saying
for quite a while that only the P4-Xeons were going
to be getting the AMD64 compatibility.

 

[Keith]

Not all "server" parts are created equal - the Prescotts are more likely just
desktop parts that will be used in some blades and pizza boxen (read:
seriously cost conscious products) vs the Noconas (Xeons) that will go in less
cost-sensitive applications.

The somewhat interesting part of this is that Intel is going along with
desktop processors being used in servers. They've had a major malfunction when
we thought out loud about doing the same with earlier P4s...

Maybe they've listened to the market speak? Naw, gotta be something else.
They've never had that particular problem before.

 

[Tony Hill]

<grumble> stupid top-posting...

Server parts? So they must be quite expensive? If they aren't very expensive,
then they will probably put tremendous downward pressure on the prices of
Intel's 32 bit processors.

Intel is not selling the chips at retail, only as an OEM part to IBM
and probably Dell (HPaq will probably be left out in the cold on this
one due to their loud support of the Athlon64 and Opteron). You might
see a few of them showing up in the gray market, but probably not
enough to make a serious dent in things one way or the other.

I wonder about the power consumption and about
the performance.

Power consumption should be more or less unchanged from the 32-bit
part. Maybe a few more watts here or there, I highly doubt that it
would be even a 5% change.

As for performance though.. well now that's the real question isn't
it. Benchmarks comparing the Xeon vs. Opteron (or this new P4 vs.
Athlon64) in both 32 and 64-bit code are still basically non-existent.

 

[Yousuf Khan]

Huh ? You seem surprised ? Intel has been saying
for quite a while that only the P4-Xeons were going
to be getting the AMD64 compatibility.

The Xeons were the only ones officially announced at that point. The P4's
were always "eventually".

Yousuf Khan

 

[Alex Johnson]

Server parts? So they must be quite expensive? If they aren't very expensive,
then they will probably put tremendous downward pressure on the prices of
Intel's 32 bit processors. I wonder about the power consumption and about
the performance.

Not necessarily expensive. Intel's lowest Xeons have long been about
$10 more than the Pentium 4. Those are the 2-way only, no extra cache
versions. This is probably going to be priced the same + $15 or so.

Power consumption will be the same. It's the exact same chip. Since
exactly the same circuits are in use, the power will be 99.9999%
identical. Minute variation only due to greater activity factor on the
upper bits of pathways. But expect the difference to be under a watt or
two.

By performance, I assume you mean the performance on 64b code. I assume
that because there's been tons of 32b benchmarking of the 64b-capable
processors so far. I'm personally of the belief that performance of 64b
code vs Opteron will be comparable to 32b code on the same chips. Thus
if the performance of P4 vs A64 is X:Y in 32b, then it will be X:Y in
64b. Until someone can show benchmarks, that's a fair assumption.

Alex

 

[Yousuf Khan]

By performance, I assume you mean the performance on 64b code. I
assume that because there's been tons of 32b benchmarking of the
64b-capable processors so far. I'm personally of the belief that
performance of 64b code vs Opteron will be comparable to 32b code on
the same chips. Thus if the performance of P4 vs A64 is X:Y in 32b,
then it will be X:Y in 64b. Until someone can show benchmarks,
that's a fair assumption.

I'd tend to agree with that assumption, but there have been some benchmarks
done by C't magazine in Germany which suggests that 64-bit performance on
Nocona (and hence Prescott) isn't as good as its 32-bit performance.

http://www.theinquirer.net/?article=15149

And here's the original article:

http://www.heise.de/ct/04/08/020/

Translated:

http://makeashorterlink.com/?D509258F8

I think whatever performance problems that they are talking about here,
might have to do with the 4GB limit of DMA for this processor.

Yousuf Khan

 

[Johannes H Andersen]

[...]

I'd tend to agree with that assumption, but there have been some benchmarks
done by C't magazine in Germany which suggests that 64-bit performance on
Nocona (and hence Prescott) isn't as good as its 32-bit performance.

http://www.theinquirer.net/?article=15149

And here's the original article:

http://www.heise.de/ct/04/08/020/

Translated:

http://makeashorterlink.com/?D509258F8

I think whatever performance problems that they are talking about here,
might have to do with the 4GB limit of DMA for this processor.

Yousuf Khan

The machine translation of German is quite amusing. 'Processor whispers' is
of course 'Processor rumours', since whispering is sometimes the way rumours
are spread.

 

[Nate Edel]

Server parts? So they must be quite expensive? If they aren't very expensive,
then they will probably put tremendous downward pressure on the prices of
Intel's 32 bit processors. I wonder about the power consumption and about
the performance.

Since it's a Xeon, the higher cost of the motherboard and PSU will count
against it, even if the chip price itself if not much higher than a
comparable-speed desktop chip.

 

[daytripper]

Since it's a Xeon, the higher cost of the motherboard and PSU will count
against it, even if the chip price itself if not much higher than a
comparable-speed desktop chip.

Prescott is not a Xeon...

 

[Keith]

[...]

I'd tend to agree with that assumption, but there have been some benchmarks
done by C't magazine in Germany which suggests that 64-bit performance on
Nocona (and hence Prescott) isn't as good as its 32-bit performance.

http://www.theinquirer.net/?article=15149

And here's the original article:

http://www.heise.de/ct/04/08/020/

Translated:

http://makeashorterlink.com/?D509258F8

I think whatever performance problems that they are talking about here,
might have to do with the 4GB limit of DMA for this processor.

Yousuf Khan

The machine translation of German is quite amusing. 'Processor whispers' is
of course 'Processor rumours', since whispering is sometimes the way rumours
are spread.

Oh, PLease! A whisper is a physical layer issue, while a rumor is
a link layer. One is a transport mechanism and one is a message. Rather
like one doesn't scream sweet nothings into one's lover's ears.

;-)

 

[MyndPhlyp]

Oh, PLease! A whisper is a physical layer issue, while a rumor is
a link layer. One is a transport mechanism and one is a message. Rather
like one doesn't scream sweet nothings into one's lover's ears.

It all depends in the height of ecstasy. :-o

 

[Hank Oredson]

[...]

I'd tend to agree with that assumption, but there have been some
benchmarks
done by C't magazine in Germany which suggests that 64-bit performance
on
Nocona (and hence Prescott) isn't as good as its 32-bit performance.

http://www.theinquirer.net/?article=15149

And here's the original article:

http://www.heise.de/ct/04/08/020/

Translated:

http://makeashorterlink.com/?D509258F8

I think whatever performance problems that they are talking about here,
might have to do with the 4GB limit of DMA for this processor.

Yousuf Khan

The machine translation of German is quite amusing. 'Processor whispers'
is
of course 'Processor rumours', since whispering is sometimes the way
rumours
are spread.

Oh, PLease! A whisper is a physical layer issue, while a rumor is
a link layer. One is a transport mechanism and one is a message. Rather
like one doesn't scream sweet nothings into one's lover's ears.

Um ... not into S & M then?

--

... Hank

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

 

[Rick Jones]

Oh, PLease! A whisper is a physical layer issue, while a rumor is
a link layer. One is a transport mechanism and one is a message.

I think you have skipped a few layers and jumbled things a bit.
Whisper is indeed physical layer, but the data link layer is phonemes
(sp). Words form the network layer, sentences the transport, and the
message is application-layer specific, which is why translation at the
network layer isn't always perfect

rick jones

 

[Keith]

Yousuf Khan wrote:

[...]

I'd tend to agree with that assumption, but there have been some
benchmarks
done by C't magazine in Germany which suggests that 64-bit performance
on
Nocona (and hence Prescott) isn't as good as its 32-bit performance.

http://www.theinquirer.net/?article=15149

And here's the original article:

http://www.heise.de/ct/04/08/020/

Translated:

http://makeashorterlink.com/?D509258F8

I think whatever performance problems that they are talking about here,
might have to do with the 4GB limit of DMA for this processor.

Yousuf Khan

The machine translation of German is quite amusing. 'Processor whispers'
is
of course 'Processor rumours', since whispering is sometimes the way
rumours
are spread.

Oh, PLease! A whisper is a physical layer issue, while a rumor is
a link layer. One is a transport mechanism and one is a message. Rather
like one doesn't scream sweet nothings into one's lover's ears.

Um ... not into S & M then?

No, not on purpose. ...unless you count 33 years of marriage. ;-)

 

[Keith]

It all depends in the height of ecstasy. :-o

^^^^^^

I don't do viagra either. ;-)

 

[Hank Oredson]

On Wed, 04 Aug 2004 18:51:01 +0000, Johannes H Andersen wrote:



Yousuf Khan wrote:

[...]

I'd tend to agree with that assumption, but there have been some
benchmarks
done by C't magazine in Germany which suggests that 64-bit performance
on
Nocona (and hence Prescott) isn't as good as its 32-bit performance.

http://www.theinquirer.net/?article=15149

And here's the original article:

http://www.heise.de/ct/04/08/020/

Translated:

http://makeashorterlink.com/?D509258F8

I think whatever performance problems that they are talking about
here,
might have to do with the 4GB limit of DMA for this processor.

Yousuf Khan

The machine translation of German is quite amusing. 'Processor
whispers'
is
of course 'Processor rumours', since whispering is sometimes the way
rumours
are spread.

Oh, PLease! A whisper is a physical layer issue, while a rumor is
a link layer. One is a transport mechanism and one is a message.
Rather
like one doesn't scream sweet nothings into one's lover's ears.

Um ... not into S & M then?

No, not on purpose. ...unless you count 33 years of marriage. ;-)

ROTFL!

--

... Hank

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

 

[Walt]

At our age, you have to scream for your partner to even hear you.

Hearing is the second thing to go.