Pentium 4 Extreme Edition - P4C 3.2/800 with 2MB L3!

[Sohcan]

Originally posted by: Pariah
The regular 3.06 Xeon with 512KB cache is $455, while the standard 3.06 P4 is $401. So you're not paying much for SMP capability. The price hikes are all in the cache. The regular 3.06 Xeon with 1MB L2 is $690, a $245 boost for an additional 512KB L2.

The price hikes are due their tarket markets, not due to increased manufacturing cost. With the 2MB Xeon MP, you're paying for 4-way SMP support. The Opteron 146 and 846 share the same die and clock speed (with 8-way MP support fused off on the former), though their respective prices are $669 and $3199.

There's no reason to think that this Pentium 4 will share the same price, or anywhere near it, as the 2.8 GHz Xeon MP.

 

[JetBlack69]

It only costs a few dollars to manufacture a cpu.

😉

 

[Czar]

sounds alot like the gainward card 😉

 

[Pariah]

Originally posted by: Sohcan

Originally posted by: Pariah
The regular 3.06 Xeon with 512KB cache is $455, while the standard 3.06 P4 is $401. So you're not paying much for SMP capability. The price hikes are all in the cache. The regular 3.06 Xeon with 1MB L2 is $690, a $245 boost for an additional 512KB L2.

The price hikes are due their tarket markets, not due to increased manufacturing cost. With the 2MB Xeon MP, you're paying for 4-way SMP support. The Opteron 146 and 846 share the same die and clock speed (with 8-way MP support fused off on the former), though their respective prices are $669 and $3199.

There's no reason to think that this Pentium 4 will share the same price, or anywhere near it, as the 2.8 GHz Xeon MP.

By the same token, if a 3.2GHz P4 with 2MB L3 cache is selling for $900, and a 2.8GHz Xeon is selling for $3692. Someone better really really need that SMP in a box capability since a cluster of the P4's would be much cheaper to build.

 

[Sohcan]

Originally posted by: Pariah


By the same token, if a 3.2GHz P4 with 2MB L3 cache is selling for $900, and a 2.8GHz Xeon is selling for $3692. Someone better really really need that SMP in a box capability since a cluster of the P4's would be much cheaper to build.

A Xeon MP cluster would offer higher performance per node as well as 2x to 4x higher density. Xeon MP rack mount systems typically come as 4-way for a 2U box, or 16-way for the 4U IBM x445.

 

[beatle]

Originally posted by: Sohcan

Originally posted by: Pariah


By the same token, if a 3.2GHz P4 with 2MB L3 cache is selling for $900, and a 2.8GHz Xeon is selling for $3692. Someone better really really need that SMP in a box capability since a cluster of the P4's would be much cheaper to build.

A Xeon MP cluster would offer higher performance per node as well as 2x to 4x higher density. Xeon MP rack mount systems typically come as 4-way for a 2U box, or 16-way for the 4U IBM x445.

4-way in a 2U? :Q:Q:Q:Q I had no idea that was typical. At work we have several Dell 2600s that are simple 2-way systems in a 5U chassis. A lot of that is to accomodate the 6 hds, but still... 4 in a 2U is crazy to me. 😛 Or did I miss something?

 

[Anand Lal Shimpi]

Just an update from the show - it actually has a 2MB on-die L3 cache, in addition to the 512KB on-die L2 cache. This is basically a faster version of the Gallatin Xeon MP. Here's the info.

Take care,
Anand

 

[AgaBoogaBoo]

Originally posted by: Anand Lal Shimpi
Just an update from the show - it actually has a 2MB on-die L3 cache, in addition to the 512KB on-die L2 cache. This is basically a faster version of the Gallatin Xeon MP. Here's the info.

Take care,
Anand

L3? I wonder how that will help...

 

[AgaBoogaBoo]

Originally posted by: Anand Lal Shimpi
Just an update from the show - it actually has a 2MB on-die L3 cache, in addition to the 512KB on-die L2 cache. This is basically a faster version of the Gallatin Xeon MP. Here's the info.

Take care,
Anand

Anand, did they mention any other changes made to the cpu other than the L3 cache? Such as things prescott may have?

 

[Excelsior]

Originally posted by: AgaBooga

Originally posted by: Anand Lal Shimpi
Just an update from the show - it actually has a 2MB on-die L3 cache, in addition to the 512KB on-die L2 cache. This is basically a faster version of the Gallatin Xeon MP. Here's the info.

Take care,
Anand

L3? I wonder how that will help...

Supposedly up to a 20% increase in speed.

 

[AgaBoogaBoo]

Originally posted by: Excelsior

Originally posted by: AgaBooga

Originally posted by: Anand Lal Shimpi
Just an update from the show - it actually has a 2MB on-die L3 cache, in addition to the 512KB on-die L2 cache. This is basically a faster version of the Gallatin Xeon MP. Here's the info.

Take care,
Anand

L3? I wonder how that will help...

Supposedly up to a 20% increase in speed.

Any links on that? If its just what Intel said, I expect it to be increases where memory latency really helps, in other cases, I'd say 10% or less. (Just a guesstimate)

 

[First]

Originally posted by: Excelsior

Originally posted by: AgaBooga

Originally posted by: Anand Lal Shimpi
Just an update from the show - it actually has a 2MB on-die L3 cache, in addition to the 512KB on-die L2 cache. This is basically a faster version of the Gallatin Xeon MP. Here's the info.

Take care,
Anand

L3? I wonder how that will help...

Supposedly up to a 20% increase in speed.

The average will probably be more like 7-8%.

 

[Excelsior]

Originally posted by: AgaBooga

Originally posted by: Excelsior

Originally posted by: AgaBooga

Originally posted by: Anand Lal Shimpi
Just an update from the show - it actually has a 2MB on-die L3 cache, in addition to the 512KB on-die L2 cache. This is basically a faster version of the Gallatin Xeon MP. Here's the info.

Take care,
Anand

L3? I wonder how that will help...

Supposedly up to a 20% increase in speed.

Any links on that? If its just what Intel said, I expect it to be increases where memory latency really helps, in other cases, I'd say 10% or less. (Just a guesstimate)

Well, from the trustworthy Inquirer article
"As a matter of fact, 2MB cache will help a lot those users whose apps (including games and such) have a lot of big cache-friendly pieces of code and data, but probably not the data-streaming intensive stuff. I do expect to see speedups anywhere from 2% to 20% depending on the application, maybe some more if using multithreading/multitasking (large cache can keep in code / date pieces from more threads)."

 

[Pariah]

Originally posted by: Sohcan

Originally posted by: Pariah


By the same token, if a 3.2GHz P4 with 2MB L3 cache is selling for $900, and a 2.8GHz Xeon is selling for $3692. Someone better really really need that SMP in a box capability since a cluster of the P4's would be much cheaper to build.

A Xeon MP cluster would offer higher performance per node as well as 2x to 4x higher density.

At 4x the cost per CPU, not taking into account the significantly higher costs of the SMP boards, and the $/performance ratio is heavily in the favorite of the clustered P4 setup. The density would be the only concern an only with significantly high # of CPU's required. Basically Intel would be eliminating the lower end market for Xeon MP's.

 

[AgaBoogaBoo]

Originally posted by: Pariah

Originally posted by: Sohcan

Originally posted by: Pariah


By the same token, if a 3.2GHz P4 with 2MB L3 cache is selling for $900, and a 2.8GHz Xeon is selling for $3692. Someone better really really need that SMP in a box capability since a cluster of the P4's would be much cheaper to build.

A Xeon MP cluster would offer higher performance per node as well as 2x to 4x higher density.

At 4x the cost per CPU, not taking into account the significantly higher costs of the SMP boards, and the $/performance ratio is heavily in the favorite of the clustered P4 setup. The density would be the only concern an only with significantly high # of CPU's required. Basically Intel would be eliminating the lower end market for Xeon MP's.

True, but a company would rather have a single computer to manage with four cpus than 4 computers with 1 cpu each.

 

[SexyK]

The 2MB on-die L3 cache takes the Northwood's 55 million transistors and balloons it to an incredible 108 million transistors, which is still less than the Prescott's 125 million transistors. What's important to note here is that although the Prescott has less than half of the cache of this new Pentium 4, it still has more transistors - giving you some insight into how much Intel enhanced the core. We will have some benchmarking time with the Extreme Edition very soon...

😀😀😀

Looks like Prescott is going to be a monster!

 

[Pariah]

Not necessarily. In recent years there has a been a definite shift towards using off the shelf desktop CPU's for large scale clusters in supercomputers as opposed to more exotic (and expensive) solutions. I'm not saying Intel is eliminating the market, but possibly taking a decent percentage of it away.

 

[kuk]

Looks like the we're reaching another consumer golden age in computer hardware.

ATI vs. nVidia vs. XGI
Athlon64 vs. P4 EE
Better LCDs
Cheaper and faster DVD-Writers
5+MP digital cameras

Feel the excitement? 😀

 

[AgaBoogaBoo]

Originally posted by: Pariah
Not necessarily. In recent years there has a been a definite ship towards using off the shelf desktop CPU's for large scale clusters in supercomputers as opposed to more exotic (and expensive) solutions. I'm not saying Intel is eliminating the market, but possibly taking a decent percentage of it away.

Ok, I understand what you're saying now. The last comment I have is that, Intel may be getting ready to release the 800mhz FSB Xeons, they've been mentioned a few times before and now would seem like a logical time in my mind because A64 is also coming out.

 

[digitalsm]

Originally posted by: Pariah

Originally posted by: Sohcan

Originally posted by: Pariah


By the same token, if a 3.2GHz P4 with 2MB L3 cache is selling for $900, and a 2.8GHz Xeon is selling for $3692. Someone better really really need that SMP in a box capability since a cluster of the P4's would be much cheaper to build.

A Xeon MP cluster would offer higher performance per node as well as 2x to 4x higher density.

At 4x the cost per CPU, not taking into account the significantly higher costs of the SMP boards, and the $/performance ratio is heavily in the favorite of the clustered P4 setup. The density would be the only concern an only with significantly high # of CPU's required. Basically Intel would be eliminating the lower end market for Xeon MP's.

Did you not read the article? Mass producing these P4 Extreme chips on a larger scale will also lower the costs of their Xeon MP chips.

 

[AgaBoogaBoo]

Originally posted by: digitalsm

Originally posted by: Pariah

Originally posted by: Sohcan

Originally posted by: Pariah


By the same token, if a 3.2GHz P4 with 2MB L3 cache is selling for $900, and a 2.8GHz Xeon is selling for $3692. Someone better really really need that SMP in a box capability since a cluster of the P4's would be much cheaper to build.

A Xeon MP cluster would offer higher performance per node as well as 2x to 4x higher density.

At 4x the cost per CPU, not taking into account the significantly higher costs of the SMP boards, and the $/performance ratio is heavily in the favorite of the clustered P4 setup. The density would be the only concern an only with significantly high # of CPU's required. Basically Intel would be eliminating the lower end market for Xeon MP's.

Did you not read the article? Mass producing these P4 Extreme chips on a larger scale will also lower the costs of their Xeon MP chips.

How will it lower the cost of producing Xeon chips? Or will they sell them at a lower price?

 

[digitalsm]

Originally posted by: AgaBooga

Originally posted by: Pariah
Not necessarily. In recent years there has a been a definite ship towards using off the shelf desktop CPU's for large scale clusters in supercomputers as opposed to more exotic (and expensive) solutions. I'm not saying Intel is eliminating the market, but possibly taking a decent percentage of it away.

Ok, I understand what you're saying now. The last comment I have is that, Intel may be getting ready to release the 800mhz FSB Xeons, they've been mentioned a few times before and now would seem like a logical time in my mind because A64 is also coming out.

800Mhz FSB Xeons are not in the present future. In the 1st half of next year they will be introducing 667mhz FSB Xeons.

 

[Lyfer]

Make it cheap and fast and the people shall rejoice!

 

[AgaBoogaBoo]

Originally posted by: digitalsm

Originally posted by: AgaBooga

Originally posted by: Pariah
Not necessarily. In recent years there has a been a definite ship towards using off the shelf desktop CPU's for large scale clusters in supercomputers as opposed to more exotic (and expensive) solutions. I'm not saying Intel is eliminating the market, but possibly taking a decent percentage of it away.

Ok, I understand what you're saying now. The last comment I have is that, Intel may be getting ready to release the 800mhz FSB Xeons, they've been mentioned a few times before and now would seem like a logical time in my mind because A64 is also coming out.

800Mhz FSB Xeons are not in the present future. In the 1st half of next year they will be introducing 667mhz FSB Xeons.

Hmm... ok.

 

[Pariah]

Originally posted by: digitalsm

Originally posted by: Pariah

Originally posted by: Sohcan

Originally posted by: Pariah


By the same token, if a 3.2GHz P4 with 2MB L3 cache is selling for $900, and a 2.8GHz Xeon is selling for $3692. Someone better really really need that SMP in a box capability since a cluster of the P4's would be much cheaper to build.

A Xeon MP cluster would offer higher performance per node as well as 2x to 4x higher density.

At 4x the cost per CPU, not taking into account the significantly higher costs of the SMP boards, and the $/performance ratio is heavily in the favorite of the clustered P4 setup. The density would be the only concern an only with significantly high # of CPU's required. Basically Intel would be eliminating the lower end market for Xeon MP's.

Did you not read the article? Mass producing these P4 Extreme chips on a larger scale will also lower the costs of their Xeon MP chips.

And in response I'll quote Sohcan:

"The price hikes are due their tarket markets, not due to increased manufacturing cost."

How exactly does higher production reduce the price hikes incurred by market targetting?