Dragonfly and Hyperthreading....

[Matthew Dillon]

:Intel's Pentium-M has similar performance, which I why I asked Matt
:if he thought Intel would use that technology to replace P4s. It appears 
:that is in the works to some degree. I have a 2.0Ghz Pentium-M notebook
:and it cooks, and I get 5 hours of continuous use on a standard battery 
:charge. A 2Ghz Pentium-M is about as fast as a 3.4Ghz P4. Here is an 
:interesting article on it:
:
:http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2342

    That's a pretty good article.  The pentium-M is clearly an excellent
    notebook cpu, using only 1/4 the power of a desktop Athlon64.  I am
    quite impressed.  Their comments about its lack of scaleability (other
    than die shrinks) also make a lot of sense.  Clock distribution is the
    single most difficult architectural problem in any digital chip design
    and it gets much worse when you start playing with clocks running at
    different frequencies as the article indicates they are doing in the M.
    It's a classic tradeoff.

    All the other graphs are in line with what I've seen before.   Basically
    the P4's have great FPUs but pretty much lose everywhere else, and have
    significantly higher power consumption.  That said the performance
    results (ignoring power consumption) of the high-end cpus are still
    typically within 15% of each other, an almost unnoticeable difference.

					-Matt
					Matthew Dillon 
					<dillon at xxxxxxxxxxxxx>