Slashdot Mirror
AMD Releases 12 New Chips at CeBIT
SuperDuG writes "AMD now offers three categories of processor for notebooks grouped under the Athlon XP-M brand. It labels them "desktop replacement," "standard," and the new "low-voltage". AMD plans to make a desktop replacement in the notebook computer market using the Barton Core, a technology designed to double the CPU Cache. Looks like yet another case of AMD being one-up on Intel."
What about the Centrino processors released last night? 1.6GHz performing equal to that of a 2.6GHz P4? AMD is way behind in the Mobile race. Does anyone know what voltage the XP-M even runs on?
Kristopher
AMD isn't "one up" on Intel until these chips are benchmarked and compared side by side with offerings from Intel. Until then, "low-voltage" is just as nonsensical as "centrino".
It would be interesting to see how the low-voltage Athlon XP-M processors compare to the vaunted Centrinos. Seems to be shaping up to be an interesting battle. Still, the categorization is a good move for AMD, I think -- it's a much more intuitive naming convention than the confusion that's doubtless going to be caused by Intel's Pentium 4M/Pentium M/Centrino names.
-- shayborg
I bought a p-2000 from fujitsu last year and one odd thing is they charge sales taxes for all 50 states(I doubt they are giving that money back to the states). So expect to pay 100+ more for anything you buy from fujitsu direct.
The centrino based X31 from IBM is new release too. Thinkpads cost more but they are built like tanks and come with a 3 year warranty compared to the usual one year from many other manufactures. The X31 would be a much more attractive package with 802.11G and legacy free. Who the hell needs a parallel port on a subnotebook?
I'm currently using a laptop with an Athlon XP-M chip and I love it, better price and great power management. I'm glad to hear that AMD is continuing to inovate in the mobil processing arena. If this continues, I'll definitly purchase another AMD based laptop.
There is no dark side of the moon really, matter of fact it's all dark
Cache memory tends to be section of the CPU that fails, it's hard to manufacture and this results in high failure rates when you have stacks of it. This is typically why Celerons and Durons have been so cheap. A poor yield puts up the cost of the good uns, hence why Xeons have always costs so much (plus they're aimed at high end servers)
I would like to see some of those 1400+ and faster low voltage chips in a desktop machine. I know there are relatively low power alternatives like the VIA C3/EDEN processors and the tualatin-based Celerons, but for some things like games and high-res video, those processors are kind of lacking oomph.
As a rather frivolous example, with neither mplayer+Quicktime dlls, nor the Quicktime player for Windows, my 1.0 GHz Celeron could not play the 1024x464 trailer of the Matrix Reloaded without dropping every 2nd or 3rd frame. A 1.533GHz (1800XP) Athlon chewed through the Quicktime with mplayer without any problems at all.
That machine however, uses significantly more power, generates way more heat and requires more noisy cooling gear.
Those low voltage cpus would be the cats ass for building a small, quiet, cool and still very powerful desktop machine that is a little easier on the power bill. Anything that uses less power is good in my mind.
"but lately Intel has been steadily ahead with clock cycles that even AMD's "2800+" marketing cant compete with. "
0 ,0 0.asp
That's where your wrong. The new 3000+ Barton is faster then the Intel 3.06. The same will be true of a 2800+ Barton vs. a P4 2.8. The fact that a P4 3GHz gets beat by a chip running almost a 1GHz slower is embarassing for Intel to say the least and most certainly confirms AMD's "number+" processor naming convention. It's Intel marketing that doesn't live up to the hype, not AMD's.
http://www.pcworld.com/news/article/0,aid,10958
If you wanna get rich, you know that payback is a bitch
Cache is less and less useful, the more you have of it. with 64K, you only cache miss, say, 10% of the time. Or your data is in cache, 90% of the time, say. With 256K, you only miss... 5% of the time. Adding another 768K, and you might get down to only 2%. Yay. You still have to load data in the first time - so it can't be in cache - it's all the way out on disk. Adding another 50 MB really won't help too much - especially for how much it costs - it will only snag another 1% tops. These are just numbers i'm making up - but that's pretty much how it goes.
Now - cache gets cheaper, and it makes it worth while to add more. But the "throw more cache at it" approach doesn't help a whole lot. And the more cache you have, the more expensive managing it becomes.
And just throwing on more CPUs doesn't help either - like another poster stated - you can add more processors, but managing them, making sure they are doing something useful and different, and not running into each others' memories...that's a trick too. 2-way SMP might be useful these days, but only for power users who really use their machines. Most people still only run one program at a time...it's usually not economically viable to ship smp systems for people that won't ever use them.
Who is this Anonymous Coward character, how does he post so much, and why is he always such a whore?