[rescue] OT: Linux and USB on Intel

[Dave McGuire]

On Sunday, April 20, 2003, at 04:42 PM, David Passmore wrote:
>>    Well the x86 problem will take care of itself.  The architecture
>> itself hit a dead end two generations ago.  While I'm sure they'll do
>> their best to get another iteration out of it, I'll be shocked if they
>> can milk it for two, then it's done.  I'm not worried about it at all.
>>
>>    x86 is a self-limiting problem.
>
> I'm not sure I agree, Intel engineers have proven to be very clever 
> (and
> Intel's marketing very good at reaching managers), so time will tell.

   They can't out-clever the laws of physics, nor are they even 
particularly clever...that is, unless you consider a sledge hammer to 
be clever.  The walls they're up against are defined by physics.  
They're close to the limits beyond which they can't get the heat out of 
the chips as fast as they produce it, they can't push the voltages down 
much further, and they can't push the clocks up much faster.  The era 
of brute-force performance scaling by dropping core voltages, adding 
heat sink area, and bumping up clock rates is drawing to a close.

   The last truly clever thing those guys did was uOps in the PPro, and 
maybe the semi-async functional unit stuff in the P-4...but the latter 
was a very, very old idea anyway.

   But yes, time will tell.

         -Dave

--
Dave McGuire           "She's a cheek pincher.  I have scars."
St. Petersburg, FL                          -Gary Nichols