The Official Prescott discussion Thread

Originally posted by chrisf6969
I think the strained (stretched) silicon is letting current leak. The layers are so thin to start with, then they stretch them out.... to allow the electricity to flow faster.
Click to expand...

I am hearing that it is leaking more than it is using....
 
Everybody LOVES coffee.

AMD hasn't shown any problems from the numbers they've put out but since we havent actually used real chips yet, we cant make any judgments.
 
Originally posted by p0opyfac3
TED WHERE ARE YOU!?
Click to expand...

I like to say i've been benching 3 prescotts all day that's why i havnt been on here. But no.

I called spartantech.com last night Monday, They said the got them in and will ship mine Tuesday. Im seeing no tracking number. So i shall call them in the morning. People have been getting backorder emails on other forums i've been at.

I was told that they got hit hard and cant fill all the orders.

But i'll see if they have a tracking number for me in the morning, I paid for Next day AM. If not off to resellerratings.com i go.
 
What makes me laugh is hearing "3.6Ghz overclocked the Prescott kicks ass". How could it be any faster than a Northwood at 3.6Ghz? It can't. That statement by Anand meant that once the Prescott speeds ramped up to 3.6Ghz, seeing as how the final Northwood is 3.4Ghz, it would be a better performer.

Am I right?
 
Originally posted by batotman
What makes me laugh is hearing "3.6Ghz overclocked the Prescott kicks ass". How could it be any faster than a Northwood at 3.6Ghz? It can't. That statement by Anand meant that once the Prescott speeds ramped up to 3.6Ghz, seeing as how the final Northwood is 3.4Ghz, it would be a better performer.

Am I right?
Click to expand...
On the next to last page Anandtech shows the Prescott vs Northwood performance differences in percentages over 2.8GHz to 3.4GHz (or 3.2...) the faster the prescott got the closer it got to the northwood.
I don't think the prescott is going to suck, but I do think Intel recreated the Williamette launch. Raise the clockspeed a bit and we're good, but they released the Prescott at too low a clockspeed to show any improvements. If they had stuck with 3.2GHz as the lowest Prescott, it would have tied and won more benchmarks rather than doing a full architectural comparison between the NW and PS.
 
Man I can get prescotts realy easily. All our suppliers hit us with the Big Intel marketing scam and all abou 90 nm. They were willing to give us realy good deals but seeing the recent problems and compatibility issues I think we are not going to use prescott till lga775 comes out. Until Then Northwood is best choice.
 
Originally posted by USMC2Hard4U
So is the next core gonna be the Pentium 5?
Click to expand...

Up next is Tejas, another version of the Netburst core. After that is Nehalem, a brand new next gen core. Who knows what Intel is planning on calling them.

I'm thinking whenever Intel activates whatever is hiding in those extra transistors they will give it a new name (Pentium 64?? :D ). That may happen with the move to LGA 775.

Edit: fixed NetBUST typo.
 
Originally posted by batotman
What makes me laugh is hearing "3.6Ghz overclocked the Prescott kicks ass". How could it be any faster than a Northwood at 3.6Ghz? It can't. That statement by Anand meant that once the Prescott speeds ramped up to 3.6Ghz, seeing as how the final Northwood is 3.4Ghz, it would be a better performer.

Am I right?
Click to expand...

I dont think thats totally correct.

I think in the long run you'll see the performance curve on the Prescott is not a straight line, I believe it will come about that as it increases in mhz, it's performance curves higher and higher, faster and faster.

Why?

Dont ask me... but from what I've read here and there, this is I believe the correct assumption.
 
Originally posted by Tedinde
I like to say i've been benching 3 prescotts all day that's why i havnt been on here. But no.

I called spartantech.com last night Monday, They said the got them in and will ship mine Tuesday. Im seeing no tracking number. So i shall call them in the morning. People have been getting backorder emails on other forums i've been at.

I was told that they got hit hard and cant fill all the orders.

But i'll see if they have a tracking number for me in the morning, I paid for Next day AM. If not off to resellerratings.com i go.
Click to expand...
Ted, I'm sorry for pointing you to them :)

You are a nice man to still be waiting until morning. I would have had that "Very Dissatisfied" post made Monday afternoon if I didn't get any tracking number ;)
 
Spartantech.com, ordered 3 prescott's

First they told me mine would ship, they got the order in monday.Tuesday nothign showed. I had to work all day today. Got home no 8 boxes on my doorstep, but not the one i wanted.

I called 10 minutes ago. And they said they wont have them till Friday. Then i asked why they told me my order would be filled Monday on the phone. He had no clue, Same customer service guy because i asked for his Ext. Last time i talked to them.

He apologized an promised they would get them. I asked for Their order tracking number!!! For a laugh.

What the hell i'll see what happens Friday. I cant post any reseller ratings fairly until i either get my order or cancel my order.

But i will be posting something Friday.


Hell just tell me you dont have them and that you have no clue when!!! and i wont even hit resellers and trash them. But dont blow smoke up my ass!!!
 
Originally posted by Tedinde
Spartantech.com, ordered 3 prescott's

First they told me mine would ship, they got the order in monday.Tuesday nothign showed. I had to work all day today. Got home no 8 boxes on my doorstep, but not the one i wanted.

I called 10 minutes ago. And they said they wont have them till Friday. Then i asked why they told me my order would be filled Monday on the phone. He had no clue, Same customer service guy because i asked for his Ext. Last time i talked to them.

He apologized an promised they would get them. I asked for Their order tracking number!!! For a laugh.

What the hell i'll see what happens Friday. I cant post any reseller ratings fairly until i either get my order or cancel my order.

But i will be posting something Friday.


Hell just tell me you dont have them and that you have no clue when!!! and i wont even hit resellers and trash them. But dont blow smoke up my ass!!!
Click to expand...


oops sorry *moves smoke machine away
 
Was reading through this post and thought I would drop a line to say we will be shipping some Prescott 2.8E OEM chips out tomorrow (2/5) and we still have some left out of the first batch.

Larry
AZZO Computers
www.azzo.com
 
Originally posted by VVVVVVVVVVVVVVVI
Up next is Tejas, another version of the Netburst core. After that is Nehalem, a brand new next gen core. Who knows what Intel is planning on calling them.

I'm thinking whenever Intel activates whatever is hiding in those extra transistors they will give it a new name (Pentium 64?? :D ). That may happen with the move to LGA 775.

Edit: fixed NetBUST typo.
Click to expand...


im not understanding this, what extra transistors?
 
Originally posted by rayman2k2
im not understanding this, what extra transistors?
Click to expand...
Williamette: 42M transistors, ~27M for the core (exclusive of all cache)
Northwood: 55M transistors, ~27M for the core
Prescott: 125M transistors, ~67M for the core

A simple die shrink of Northwood with doubled L1/L2 caches would have been 80-85M transistors.

So even accounting for the large pipeline changes and other enhancements, rumored 64-bit extensions there are still several tens of millions of transistors not accounted for.

Edit to add in rough estimate of number of transistors dedicated to the core only
 
Originally posted by SLee
Williamette: 42M transistors
Northwood: 55M transistors
Prescott: 125M transistors

A simple die shrink of Northwood with doubled L1/L2 caches would have been 80-85M transistors.

So even accounting for the large pipeline changes and other enhancements, rumored 64-bit extensions there are still several tens of millions of transistors not accounted for.
Click to expand...


oooOOOooo
 
There is ofcourse updated Hyperthreading as well more cache, and SSE3 extensions.
 
Kyle,
Obviously the Prescott does have heat issues, I'm just wondering why you didn't hook up that Koolance water cooling system your so damn proud of!! :p
I mean every other chance you've had in past reviews we've seen results using your water system.
Did Intel ask the reviewers NOT to do this? Just wondering as everyone and their brother gave actually pretty vague "air" overclocks if they gave any at all.
Several ES's apparently did reach 4GHz relatively easily. I also noticed that the main review sites used Intel supplied motherboards for their reviews as well.
Enlighten us oh [H]ardware God! ;)
 
he already said in the review he didn't want to test overclocking on a non retail chip.
 
I think we've been over-speculating the breadth of Intel's massive transistor count, thinking its all for some special AMD64 extension. Chip-Architect did a nice set of electron microscope images of the Prescott awhile back. I think therein lies a hint: Intel is venturing into Chip Multi-Processing (CMP).

Notice the center of the Prescott is dominated with what looks like twin execution units. If our friends at Chip-Architect are correct then it shows why the transistor count increased so much, its a twin-execution unit core that leans towards hidden thread-level parrallel (TLP) execution! The extra transistors is mostly put into its branch-prediction and prefetch logic and secondly into its caches. Intel must be waiting to enable the second execution units in their new-generation Xeon family, which makes sense considering its something you'd use more in a server chip than a consumer design.

The left hand side of each execution unit there appears to be separate D-caches. So the twin units share a common trace cache but each has an independent D-cache. There is but one large FPU, which is consistent with a TLP design according to Aceshardware 's article on the subject. The advantage of this design is that Intel can swap between execution units as well as use HyperThreading to pull out new levels of performance from their current line of CPU's. Anandtech hinted back in 2002 that Intel was working on two-core designs that had cores actually running at their own speeds, one fast and one slow. I can only imagine they'll label a TLP Pentium4 at 2.4GHz as a 4.8GHz to smash AMD's hopes at using PR numbers to compare directly to their products.

Hardware prefetch, cache size, and overall memory bandwidth are much more important to a twin-core design than is latency. Notice how Intel allowed the latency to increase in the Prescott whereas before they were so overty concerned with every cycle of latency they could squeeze out of the P4. The downside with the idea of TLP in Prescott is that Intel's 90nm process is probably preventing the uncorking of this design for the time being. Damn them for being so aggressive and falling short!

Lets not forget that AMD64 and IA64 code can be run on any x86-compatible processor using an emulator. Sure there will be some performance issues, but all in all this is the most sane way for Intel to compete with AMD64 at the moment. Call me crazy, but I think this is still part of the master plan for Intel. Intel is pushing 32-bit performance right over the edge with this new TLP design and then will meet the 64-bit crowd with some nice emulation package, but IA64 will run better on it than AMD64.

Windows Longhorn won't care what architecture it runs on being that its moving away from API's and towards XML for inter-program communication. This allows just about any platform to be Windows-compatible with the least amount of cost overhead for Microsoft. API's are too hardware dependent whereas XML will allow the compilers of each program to eek out their performance margins. Like I said, just call me mad(rat)...
 
Originally posted by chrisf6969
The should have just produced basically a Prescott core (all improvements EXCEPT longer pipeline) on the MATURE .13 process.

It would probably run faster & cooler, until they can get the .09 process straight. I know the die size would hurt them, temporariliy, but it cant be any worse than Williamettes were.

1Mb L2, improved HT, SSE3, etc.. various improvements on .13 would be the ticket at least temporarily. [/B]
Click to expand...

Would you also want a functional warp drive and cloaking field attached to it, while we're at it?

There are limitations to technology and design.
 
Mad Rat seems to have stumbled on to something here.

As you said thye fell a little short. If they would just of a waited a while instead of giving us a wilmate deja-vu.
 
Originally posted by SilverBack
Kyle,
Obviously the Prescott does have heat issues, I'm just wondering why you didn't hook up that Koolance water cooling system your so damn proud of!! :p
I mean every other chance you've had in past reviews we've seen results using your water system.
Did Intel ask the reviewers NOT to do this? Just wondering as everyone and their brother gave actually pretty vague "air" overclocks if they gave any at all.
Several ES's apparently did reach 4GHz relatively easily. I also noticed that the main review sites used Intel supplied motherboards for their reviews as well.
Enlighten us oh [H]ardware God! ;)
Click to expand...


http://www.hardocp.com/article.html?art=NTgzLDI=

he did hook it up??? just no benchies OC'd.
 
If he's got an ES then wouldn't it be more prudent to measure each speed grade using stock FSB's than just overclocking it? I'd rather know what a stock 4GHz does than know what an overclocked 4GHz does at the moment.
 
You must log in or register to reply here.
Back
Top