xorg +XGI Volari XP5

Matthew Dillon dillon at apollo.backplane.com
Tue Apr 26 15:10:52 PDT 2005


:Matt:
:>a competitive offering in ~12 months or so (the Turion isn't it).
:
:Why you think Turion isn't competive?

    It's too new, for one thing.  It was barely launched a month ago.
    Secondly, the Turion is basically just a rebranded notebook 
    athlon 64.  The pentium-M is a major redesign of the pentium-III
    core with power consumption in mind.

    The turion clearly outperforms the pentium-M, but the low voltage
    version of the pentium-M beats the shit out of AMD's best turion
    offering in regards to power consumption.  The pentium-M also
    has a bunch of major frequency-reducing technologies built into it 
    (such as clocking different parts of the core at reduced frequencies),
    above and beyond basic cpu frequency controls, and I'm fairly sure
    that the Turion just has the basic stuff.

    Slashdot quoted a The Register article about this a while back, but
    they pointed to the wrong The Register URL so you will have to 
    search The Register's site for the article.

	http://slashdot.org/article.pl?sid=05/03/16/0219208&from=rss

    Don't get me wrong, I believe that AMD has a superior NON-laptop cpu.
    But Intel definitely has the superior laptop cpu.  You don't buy a
    laptop for performance, you buy it for how long it will last doing
    normal tasks on battery.  It need only have 'adequate' performance.
    If you want performance in a laptop you can always get a pentium-4
    or an AMD64 monster, with a battery life of 1-2 hours.  If you want
    a low power offering then the Pentium-M centrino based laptops, with
    battery lifes approaching 4 hours (at least running a FreeOS as I have
    a centrino laptop) wins.

    I also believe that AMDs technology will ultimately prove to be
    superior for laptops, ONCE they've done a few major engineering turns
    on it to reduce power consumption.  It's 1-2 years away at the very
    least.

:Thomas:
:> Yes, but it's not as efficient as a Pentium-M, and you can't buy it 
:> everywhere yet.
:
:I can buy one for 929EUR, 512 MiB Ram, a Turion64 1.6Ghz CPU, 80 GB, 
:DVD±RW DL Writer, Modem, Ethernet 10/100Mb, 801.11g, USB2.0
:
:The shop is having it in storage.
:
:> That's just thermal dissipation, not real power usage.
:
:I gave you evicence on the validiy of these number, what you give me is 
:a vague expectation based on nothing... please back up your claims with 
:numbers, or don't claim anything.

    To the original poster re: thermal dissipation.  Thermal dissipation
    (i.e. heat generated) is *DIRECTLY* related to the amount of power the
    cpu (or any device) consumes.  It's a basic law of physics.  Since
    a computer is not converting energy to potential energy (i.e. consuming
    power to bond matter which then becomes potential energy that can be
    later released, or to raise itself up a gravity well, or accellerate 
    the computer in space, or the myrid other forms of potential energy),
    then only place for that power consumption to go is as kinetic energy,
    which in this case is basically just heat.

    However, TDP (Thermal Design Power) has very little to do with actual
    power use.  The power used by a laptop cpu depends heavily on work 
    load and frequency reducing technologies, and very little to do with
    the maximum heat dissipation of the cpu.

					-Matt
					Matthew Dillon 
					<dillon at xxxxxxxxxxxxx>

:> I'm wondering what stresstest you did on that winny? Running SuperPI, 
:> 3DSmax renders etc. produces heat sometimes in excess of the TDP 
:> spedified for said processor (I tested on a Winchester 3000+).
:
:http://www.hardtecs4u.com/reviews/2005/amd_90nm/index5.php
:Official TDP by AMD: 67 Watt
:Testet with Prime95: 38.65 Watt
:
:If you look at the old Athlon64 3500+, you will notice AMD's TDP is set 
:at 89, the test measured 59.32. Same goes for the FX-55 offical number 
:is 104Watts, and measured in this test 72.99...
:
:I'm curious what you tested.






More information about the Users mailing list