Intel changes LGA775 package for Kentsfield...

[freecableguy]

I have done my best to focus on each chip, CPUs are in the same position in each pic:

Left: Conroe E6700 ES
Right: Kentsfield "dual E6700" ES (quad 2.66GHz)

As you can see, the Kentsfield LGA775 package is visibly thicker. If my memory serves me right the package has been built on a maximum of 8 layers up until Kentsfield. Looks like an additional 2 layers have been added (additional signal layers for the 2 additional cores would make sense)....











-FCG

Edit: add third picture

[taemun]

lol man you have too much spare time
good info

edit: is it just me or does that make the IHS sit 2 layers higher as well? is that gonna be a concern with some HS mounting?

[ZX7891]

it will most likely just put a tiny bit more pressure on the IHS, not really a big deal i think, though it is an interesting observation

[Haltech]

Yea, its a tad taller. HSF pressure will probably increase 8 pounds of so. How hard is it to pull the IHS off these?

[aintz]

lol very good observations

[Azh_fx]

it does look thickers but if you level the top of the cpus (ihs) its looks the same height.

[hipogtr]

Now that's getting into the details. Very interesting stuff, FCG.

[MasterOfTheReal]

Hehehe did u spend an hour looking at there then just go oooooo! Thicker ;p

[Metroid]

I have done my best to focus on each chip, CPUs are in the same position in each pic:

Left: Conroe E6700 ES
Right: Kentsfield "dual E6700" ES (quad 2.66GHz)

As you can see, the Kentsfield LGA775 package is visibly thicker. If my memory serves me right the package has been built on a maximum of 8 layers up until Kentsfield. Looks like an additional 2 layers have been added (additional signal layers for the 2 additional cores would make sense)....


-FCG

It looks as you said between 2 or 3 layers have been added to to it.

I hope the final model will be reduced with a 45nm increment.

It looks brighter than the normal one.

[taemun]

I don't suppose that you have a dial caliper (micrometer, w/e) handy FCG?

[Vric]

Interesting. It could be a problem with some heatsink..

[eva2000]

wow interesting info.. still would be fine for swiftech g4 storm users i guess

[dinos22]

you're famous

waves at The Inquirer readers

Hello Fudo you legend heheheh

[longshot]

haha waving at The Inquirer readers thats a good one!

[Greenhead]

so does this mean that the retail version of kentsfield will not fit the 975 motheboards?

[EnJoY]

It will be interesting to see how this will effect the use of current lga775 heatsinks.

[Vric]

It will be interesting to see how this will effect the use of current lga775 heatsinks.

and the lga775 socket itself !

[Activate: AMD]

now thats just nitpicking!

congrats on becoming a "spod" FCG

[nn_step]

Another Intel Socket change...

[PGHammer]

so does this mean that the retail version of kentsfield will not fit the 975 motheboards?

No, it doesn't. It shouldn't even affect most heatsinks. The issue would only occur with certain types of enclosures (SFF, for example).

[PGHammer]

and the lga775 socket itself !

It won't effect the socket. Not a whit. The only real issues are with heatsinks and enclosures.

[WeedJunk]

Nice observation FCG

[Mr.President]

No, it doesn't. It shouldn't even affect most heatsinks. The issue would only occur with certain types of enclosures (SFF, for example).

Thats wrong, the p5wdh for example or the bad axe 304 supporst kentsfield.

i think all conroe ready boards are also kentsfiel ready


and the heatsinks will fit for sure, it´s only ~1 mm differnce

[PGHammer]

Thats wrong, the p5wdh for example or the bad axe 304 supporst kentsfield.

i think all conroe ready boards are also kentsfiel ready


and the heatsinks will fit for sure, it´s only ~1 mm differnce

That was the point I was trying to make.

Even though Kentsfield is basically two Conroes, it uses the same LGA775 socket, and is no wider (instead of going across, like the Pentium Pro did, Intel went vertical). This also means *no* socket change (and much worse news for AMD if the yields hold up).

The issue of fab yields, of course, leads to the biggest worry that Intel (and naturally AMD) will have. If Kentsfield yields are high, this will, naturally, push Core 2 Duo pricing heavily south (as much as a third south from launch pricing, even for the X6800). AMD will have to re-jigger K8L to be socket-compatible with AM2 (which it isn't planned to be for now), so as to avoid pissed AMD users deserting in droves (two socket changes in a year is a bit much to swallow, and is something that even Intel never tried to do) which will, naturally, result in more delays, which gives Intel a greater incentive to utterly *swamp* AMD in Core product (Solo, Duo, Core 2 Duo, and Kentsfield/Core 2 Quad). If Kentsfield yields are low, however, AMD would be correct to be screaming *paper launch*; internally, AMD's board will be realizing that they dodged a major bullet.

[VictorWang]

FCG, you can compare with 965XE, same PCB as kentsfield there