Official Conroe information thread

[h20-ski]

Well Conroe is the desktop model that all consumers will buy. Woodcrest is the server model, so it allows for multiple cpu's to be installed. And Merom is the version that will go into notebooks. And ya Intel_User will give you more info on the differences.

[Sandog]

Originally Posted by h20-ski

Well Conroe is the desktop model that all consumers will buy. Woodcrest is the server model, so it allows for multiple cpu's to be installed. And Merom is the version that will go into notebooks. And ya Intel_User will give you more info on the differences.

Your probably right.

[h20-ski]

Haha, now that I read that it sounds funny. I guess what I meant was most of the general public. Although I know I will not be in the most category. It looks like I have gone dualie and won't go back.

[Intel_User]

Haha. You have summed it up so well, I don't know what to say. I can go a bit on the technical details, but it depends on your wishes .

[clubsport]

Thanks mate, well said

[h20-ski]

I don't mean to steal any of your fire Intel_User, this is a wonderful thread with lots of info.

[Intel_User]

They will need it for K8L. But anyway, its going REALLY good.

Link: http://www.theinquirer.net/?article=31789

Woodcrest is now 65W.

[Sandog]

[Intel_User]

The good news here means that it may apply to all Merom-derived cores and the quad-core variants like Kentsfield. Maybe we'll see 50W Conroe and 29W Merom(rather than 35W). The EE's could probably be 75W rather than 90W.

The article also says about faster running. Maybe per clock increase?? Almost bug-free?? The Pentium 4 Northwood B's got a small fix on the later steppings which gave 2% performance increase due to fixing data prefetcher error.

[h20-ski]

That is good news when I build my newer dualie. Considering that the TDP of 2 Woodcrest matches the TDP of a single 3.2 Xeon oc'ed to 3.6.

[Intel_User]

That is good news when I build my newer dualie. Considering that the TDP of 2 Woodcrest matches the TDP of a single 3.2 Xeon oc'ed to 3.6.

Remember, the differences between TDP and real world power usage for the Netburst based cores and Pentium M/Core Duo(and likely Core 2 Duo cores).

When Netburst based CPUs are said to be 130W TDP, actual power consumption figures sometimes put it over 170W.

When Pentium M is said to be 27W TDP, actual power consumption is 25-26W. When looking at that, Core 2 Duo looks even better.

[Intel_User]

Currently, I don't know all the differences between Merom/Conroe/Woodcrest cores. The main difference between them is power management, but there is good chance they are similar. Woodcrest has one clear difference and that's SMP support.

Even though the dynamic power management on current Intel CPUs are all called EIST, the implementations are actually quite different.

Simply put, the EIST on the Xeons(Netburst based) and the Pentium 4/D's are bad, while EIST on the Pentium M/Core Duo is very good. It could be though the Intel engineers put EIST on the Prescott cores by haste. Even TM2 on the Pentium M/Core Duo is superior than the ones on Xeons(Netburst based) and Pentium 4/D's. It even makes me think the desktop CPU designers attempted to put EIST on the Pentium 4/D's, when its not their specialty.


Obvious advantage of Merom over Conroe will be power consumption in general. The cores will be binned/tweaked to have low power, but they do sacrifice clock speed to do that.

[Intel_User]

A link I thought was interesting and shows advantages of EIST and TM2 implementation advantage of Pentium M/Core Duo over Pentium 4/D: http://www.digit-life.com/articles2/...atures-pm.html


Summary(CPU fan off)

Pentium 4/D EIST
Varying from 2.8GHz to max clock speed, say 3.6GHz, also changing voltages

TM2
It eventually goes to its lowest clock speed, which is 2.8GHz, and stays there, the temperature gradually increases few degrees from initial point.

Pentium M/Core Duo EIST
Varying from 800MHz/1GHz, to max clock speed, say 2.0GHz, also changing voltages

TM2
It goes to lowest clock speeds right away as fan is stopped. Unlike the Pentium 4/D's EIST, it also attemps to increase performance(I guess during moments when it doesn't run as hot), and the temperature doesn't increase from the initial reading.

[h20-ski]

It seems that you are onto something Intel_User, a lot of the implementations first were slated for P-M based processors, then implemented on the Netburst to help control the outrageous power consumption. IMO 2.8GHz limit on Prescott for the desktop seems way to low. But after thinking about it the lowest multiplier for the 130 nanometer process was 12 and for Prescott, 14.

[Intel_User]

It seems that you are onto something Intel_User, a lot of the implementations first were slated for P-M based processors, then implemented on the Netburst to help control the outrageous power consumption. IMO 2.8GHz limit on Prescott for the desktop seems way to low.

Thanks, I somewhat appreciate it.

But after thinking about it the lowest multiplier for the 130 nanometer process was 12 and for Prescott, 14.
Yesterday 11:55 PM

Hmm, the differences may be no more than to save money, or even purely being lazy. No reason to have kept same low speeds of 600MHz on the Core Duo's as being the lowest frequency mode for example.


BTW, I love the new logos. Pentium 4, and its logos, actually Pentiums in general is getting old.