Replies to post #2366 on Intel Corp (INTC)

[Bonefish]

Elmer,

Ref:"Legitimately? I don't think so. 100 quasihurts but not legitimately."

Well, you can argue all that quantipoints/Mhz stuff until the cows come home. The rating seem about fair to me.

Ref:"AMD has not shown they can do this with even a large penalty. AMD has never been able to produce a large cache die"

This is incorrect, the K6-3 had a 512k L2 cache. It did have a slight Mhz penalty however 5% maybe. Unfortunately the K6-3 came at the end of a product cycle and was phased out rather quickly.

Ref:"With their surprisingly poor fab output, they aren't likely to readily move to larger die unless they have no other choice. Come to think of it, I guess they have no other choice."

Using SOI and Low k the Hammer die won't yield well enough for AMD to offer the Clawhammer (and make money at it) before the Hammer core has been tweaked a while. So you are correct. AMD has no choice but to jack up the rating of the throroughbred by adding cache > Barton. This is not unlike the situation at the end of the K6-x product cycle when AMD came out with the K6-3 only to barely make it to Athlon 6 months later. As a matter of fact, the seasonal timing is about the same. Rather eery.
AMD blows process and then has to fall back to tweaking the Athlon just to barely make it to Opteron. This time however AMD gambled with SOI and unlike the Athlon, Opteron will have to be offered as a server chip first.

Bone

PS. I have to wonder the Mhz penalty the P4 took when it added up to 512k cache. Doen't really seem to matter since the P4 appears to still have enough left in the tank.