View Single Post
Unread 07-26-2003, 09:55 PM   #80
Cathar
Thermophile
 
Cathar's Avatar
 
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
Default

Quote:
Originally posted by copyman
I wonder if it would not be possible in the cascate or lrww, you optimize them a little more to the format incapsulated of the die p4, increasing the efective thermal change area of them
To optimise more for the P4's will result in less performance for bare-die CPU's.

I figured it like this. All the really hard-core P4 overclocker's pop the heat-spreader off the P4 anyway, which essentially turns a P4 into a bare-die CPU again.

I guess it's just a matter of targetting what matters most when it comes to balancing performance across a range of applications. There is a slight preferential focus to bare-die CPU's, but I was careful to ensure that for IHS-based CPU's the Cascade also gives very good performance.

The IHS really is a cooling obstacle. I can work with it slightly better (maybe 1C), but to the detriment of bare-die cooling performance, and doing so will also increase the cost of the waterblock as there will be extra machining time.
Cathar is offline   Reply With Quote