More Conroe Details Emerge.....

[Sentential]

I specifically remember seeing that almost all of the ESs handed out by intel based off of Conroe were at 1.8ghz and did indeed have a ramping issue similar to what Yonah was shown to have (in preliminary tests).

However I read on the INQ today that Intel is starting to send out 3.33ghz XE chips to their prefered people. Curious... I investigated further and found this:

http://img323.imageshack.us/img323/8651/inteldcroadmap3sn.gif

Naturally I cannot validate this but *clearly* they did something, and recentally to get them to ramp like this...

With a 14 stage pipeline and a rumored 25% IPC improvement over Dothan this really could be a monster of a CPU of which we havent seen since the original Northwood days.

Granted AM2 will have the edge in terms of bandwith with their FSB (which ought go go past 500 from all indications I have seen). Intel might wind up with the upper hand on CPU power.

Time will tell but my interest in going Conroe needless to say, has significantally increased. If Intel's consistancy is that of Presler in terms of yields you are looking at roughly or atleast 3.5-3.7ghz range for Conroe if these specs are correct, possibly even 4.

(Assuming yields are as current and not similar to AMD's style)

(Intel 921?) (10x266) ~ 3700 / 10 = 370mhz FSB w/t possible $250 price tag?
**Max on i955X/i975X assuming its flashable is roughly 350-380**

[d94]

looks good
hmm i thought conroe was single core. guess i was WRONG
and just when i thought i wouldnt upgrade my cpu more than once this year :)

[hUMANbEATbOX]

hollly cow. i didn't think they would debut any where near 3.4. i'm hoping that is accurate, it could mean great things for us!

its gonna be a new mhz race in 2006/2007 now that the am2's and conroes will be closer in clock speed. looks like we will have a nice head start too :)

[thegreek]

conroe XE is going to need i985 which makes current MBs worthless?

[Andor]

Hrm I was going to buy a new CPU...but now I'll wait. Christmas in July for me :santa:

[Sentential]

I dont know what to say... something doesnt feel right about this. Everyone who owns a 65nm chip will admit to you that the yeilds arent so hot, but how in the world did they effectively double their speed in a matter of months?

I know intel techs are good but they arent magicians. Either the pipeline is wrong or the clocks are wrong. I just dont see how both can be right.

[Andor]

I dont know what to say... something doesnt feel right about this. Everyone who owns a 65nm chip will admit to you that the yeilds arent so hot, but how in the world did they effectively double their speed in a matter of months?

I know intel techs are good but they arent magicians. Either the pipeline is wrong or the clocks are wrong. I just dont see how both can be right.

One word: Aliens

[thegreek]

I dont know what to say... something doesnt feel right about this. Everyone who owns a 65nm chip will admit to you that the yeilds arent so hot, but how in the world did they effectively double their speed in a matter of months?

I know intel techs are good but they arent magicians. Either the pipeline is wrong or the clocks are wrong. I just dont see how both can be right.
Isn't Conroe a mix of Dothans architecture and netburst which would increase performance AND also a higher clock rate?

[DanIdentity]

I would question the authenticity of that chart. I don't know where it came from, but it's been in a thread on XS for awhile now.

[hUMANbEATbOX]

I dont know what to say... something doesnt feel right about this. Everyone who owns a 65nm chip will admit to you that the yeilds arent so hot, but how in the world did they effectively double their speed in a matter of months?

I know intel techs are good but they arent magicians. Either the pipeline is wrong or the clocks are wrong. I just dont see how both can be right.


it could be much like the first opterons that were sent out. weren't they around 800mhz, just to demo the new tech (like ondie memcontroller)?

now that they are ramping up production, yields have got to significantly improve, no?

[Who]

The Presler info on that chart is wrong, so it's either old or just made up. Either way, I wouldn't trust it.

[Npvk_x]

:) conroe will rock... hard...

[Sneaky]

whats wrong with the presler info on the chart? seems correct to me...


and yes... conroe is looking mighty sexy - and will definitely become part of Rev. D1 of my ongoing project (see sig)

[Sentential]

whats wrong with the presler info on the chart? seems correct to me...
No he's right. The cache listed is incorrect which is further reason to treat this with skepticism

[Who]

whats wrong with the presler info on the chart? seems correct to me...


and yes... conroe is looking mighty sexy - and will definitely become part of Rev. D1 of my ongoing project (see sig)
The L2 cache amount is incorrect.

[DDR-PIII]

conroe XE is going to need i985 which makes current MBs worthless?

maybe not, the presler will run on an i865pe chip set, so you never know.

[Hamm3r]

Can't wait for it to be released :drool:
Just enough time for me to save some money for the new platform.

[ronaldo]

I see that info today at http://www.theinquirer.net and i just say to myself, THAT CPU WILL BE IMPOSIBLE TO BEAT.
Now i just hope AMD will release something competitive, but i think they will cant till 2007-2008 at least... (I wannt INTEL/AMD to be competive, its better for us all)

Interesting in that graph that Sen posted, Conroe cache is only 2mb or 4mb and not 2x2mb and 2x4mb. (as these are dual-core cpus)

[Sneaky]

oh! i see - yea its supposed to be 2x2mb for all of them, not just the XE

wasn't paying close enough attention

[DanIdentity]

This inquirer article from yesterday says Conroe XE's will be 3.33 GHz.
http://www.theinquirer.net/?article=29031

Maybe there is truth to it?

[markodude]

In Conroe I think the cache is referred to as 4Mb as its not two seperate cores with seperate caches, its one cache shared between the 2 cores, AFAIK.

[DanIdentity]

In Conroe I think the cache is referred to as 4Mb as its not two seperate cores with seperate caches, its one cache shared between the 2 cores, AFAIK.

That's correct. Conroe will have shared cache.

[hUMANbEATbOX]

^^ i've heard that as well.

[Vrykyl]

well conroes based on yonah...and yonah has shared cache too

[markodude]

This inquirer article from yesterday says Conroe XE's will be 3.33 GHz.

Sounds feasible, bearing in mind we know a T2700 is on the way and thats designed for notebooks, it makes sense the 65nm Conroe will scale pretty well...but if that is the case we could well be back to the race for 4ghz again, thats slightly quicker than I imagined from non-netburst CPUs.

[>HyperlogiK<]

I thought that Conroe was due in Q3 of this year

[markodude]

^^^^it is!

[Ross]

Nice find Sen!!!

Everyone that gave me crap for saying there would be 1333FSB Conroe can now apologize...PM is to the left :)

From what everyone says, I'd expect them sometime in Q2 rather that Q3...fingers are so crossed. The 3.2GHz will be nuts, but the XE will be unstoppable, especially if they can top out at 4GHz! If it's less than $1000, I'll be all over the XE.

I guess no Conroe will work on the 975X at all? Yet another reason they're not looking like much of an upgrade over 955Xs :(

[Vrykyl]

im hoping 975x will support conroe, hopefully with a bios update... who knows?

[Ross]

I was too. In fact, I was almost sure they would, but at least they list the 955 as an option for Presler (975 for XE)...the 975 isn't even an option for the standard Conroe.

I guess we'll see, but can easily think Conroes are sufficiently different enough to require a new chipset. They may very well work on the 975 when the time comes, but I'll bet there will be a huge benefit going with a 965/985 chipset that's optimized for them.

I think unless 975X boards are significantly better than 955X on the Preslers/CMs, I am just sticking with my Prem until a 985 board is available ;)

[markodude]

Since they can get Yonah to work on 975x, I imagine Conroe std will work on it.
I would think Conroe XE will need the 985 chipset, much like Presler XE needs the 975 just now.

[Ross]

That would be great if it does. Like I said, I was almost positive that it would work on a 975X board until I saw that roadmap. I guess it wouldn't hurt to upgrade to a 975X board now and then a 965/985 in 5-8 months if needed, but I won't do that until I see just how much better or not the 975X is for Presler/CM....I have too many other things I want to upgrade too and $250-$300 might be much better spent somewhere else right now :)

[Vrykyl]

remember tho that its entirely possible that the roadmap in question is wrong....

[Ross]

Oh, I know that. I am just saying, that anyone buying a 975 board now "expecting" it to work on with Conroe, that might not be the case and another board will be in order.

[3DFlyer]

I was too. In fact, I was almost sure they would, but at least they list the 955 as an option for Presler (975 for XE)...the 975 isn't even an option for the standard Conroe.

I guess we'll see, but can easily think Conroes are sufficiently different enough to require a new chipset. They may very well work on the 975 when the time comes, but I'll bet there will be a huge benefit going with a 965/985 chipset that's optimized for them.

I think unless 975X boards are significantly better than 955X on the Preslers/CMs, I am just sticking with my Prem until a 985 board is available ;)

:cool:

[thegreek]

I'm going to start saving now for that Conroe XE

[greenmaji]

This news is making Mobo selction tough..
A good 955X board might end up being a better option temperarly??
:( bummed

[hUMANbEATbOX]

why be bummed? if you always wait for the next big thing, you'll never get anything.

[Ross]

green, just wait a week or two and I promise some good, probably even pre-modded P5WD2 Premiums will in the classifieds or Ebay ;) *IF* something makes me opt for a 975X board now, that's what I will be doing with mine :D

If I was in the market for a board now regardless and wanted new only, I see no reason to consider anything besides a 975X mobo if that helps your decision ;)

[User-X86]

Are the conroe XE a 64-bit chip?

[>HyperlogiK<]

all conroe are 64 bit

[mearle73]

I am so pleased,I just managed to sell my P5nd2-sli,you would not belive the trouble Ive had with that,now ordered up the gigabyte 975x board & 920,I dont mind in 6 months if it wont run conroe,Im sure if I can off load a p5nd2 getting rid of a 975 board will be no trouble.

[markodude]

I am so pleased,I just managed to sell my P5nd2-sli


What did you end up getting for it (if you dont mind me asking)? I may sell mines and the GTs when I get the P5WDG2....unless Nvidia have a change of heart with the driver support on 975x...then I will keep the GTs and just sell the board.....its still a good board and has served me well for 5 processors (570J, 3.46EE, 3.73EE, 840 and now 950), it clocks to 300FSB with EEs, but it has a number of issues with overclocking, particularly with Dual Cores, and could certainly be a lot better.
I will miss SLI though :Sniff:

[mearle73]

What did you end up getting for it

I got 70 pounds for it,to be honest,I would have taken less just to get rid of it

[>HyperlogiK<]

Just a thought, how well is the conroe xe likely to overclock, obviously we can't know without getting our hands on one, but 3.33Ghz sounds like it might be a bit of an FX-53 type chip (i.e. pushing the current process to the limit in terms of heat and clockspeed). 3.33Ghz seems awfully fast, and I could see the 2.66 model being an opty 165 to the FX-60 of the extreme edition, I hope so at least.

[proth]

Just a thought, how well is the conroe xe likely to overclock, obviously we can't know without getting our hands on one, but 3.33Ghz sounds like it might be a bit of an FX-53 type chip (i.e. pushing the current process to the limit in terms of heat and clockspeed). 3.33Ghz seems awfully fast, and I could see the 2.66 model being an opty 165 to the FX-60 of the extreme edition, I hope so at least.

There is a good chance that Conroe overclocking will be limited in comparison of what we are used to seeing due to the reduced internal pipeline that allows many levels of bufferring between a screaming core and the FSB. The reduced pipeline has big benefits in the real world by reducing internal cache flushes due to the unpredictable branching and the wait for memory fetches as data moves in and out of the core.

Conroe will surely outperform Prescott, Cedar Mill and Presler in the benches, but the days of high overclocks will be over.

This is only what I have read in articles and papers

[Illyest]

As much as I want to hope Conroe (XE) will be compatible with at least 955X/975X i'm not under any delusions. Plain jain Conroe procs are a slim maybe but I really doubt it. You gotta look at current history and the direction Intel is heading to be more platform-oriented.

Not to mention Conroe is an all new chip design. I think of it in terms of running a 386 cpu on a 875 chipset...

Heck all the info Intel PR's have said point clearly to needing the new chipsets to run Conroe proc's.