Intel’s Centrino brand has developed extremely well since its introduction less than two years ago.  You could go as far as to say that Centrino has been Intel’s quickest zero-to-success brand that we’ve ever seen in the history of the company.  A very large part of the success is due to the strong technology behind Centrino

AMD doesn’t have the resources of Intel, that’s plain and simple.  They have done an excellent job with their K8 architecture, but for AMD to devote additional resources to developing another architecture, strictly mobile oriented, it’s just not possible at this time.  At the same time, AMD’s Athlon 64 is far more suited for mobile environments than Intel’s Pentium 4 ever was, so the need for a mobile specific architecture isn’t as great for AMD as it was for Intel.  So when AMD announced their Turion 64 “Mobile Technology” we had a feeling it would be something very similar to their desktop Athlon 64s, today AMD confirmed that.

Much like Intel’s Centrino, AMD is referring to their Turion 64 as a “Mobile Technology” and not just a microprocessor platform.  AMD has seen the success of Intel’s Centrino Mobile Technology and is hoping to capitalize on some of that success.  Unlike Centrino however, Turion 64 doesn’t rely on AMD-supplied chips to receive the branding.  AMD will provide the Turion 64 processors, but chipsets can be provided by a number of vendors as can the LAN and wireless controllers.  AMD is far less restricting on the components that make up Turion 64 enabled notebooks, which means that they will inevitably be cheaper than Centrino platforms, but it also means that they may not be as power efficient as Centrino platforms - it’s a tradeoff that AMD honestly had to make, as they are far from being in Intel’s position. 

The Turion 64 processor is basically a 90nm mobile Athlon 64, so all of the architectural features of the Athlon 64 make their way to the Turion 64.  One advantage that the Turion 64 has is that with an on-die memory controller, AMD can potentially offer lower memory controller power consumption than Intel. 

The Turion 64 is based on the latest revision E4 of the K8 core, meaning that it supports SSE3 instructions as well as lower power states.  The Turion 64 line also supports AMD’s PowerNow technology (known as Cool’N’Quiet on the desktop), which allows for clock speed (and voltage) modulation between 1.0GHz and the processor’s maximum frequency based on load. 

The Turion 64 will be available in both 1MB L2 and 512KB L2 cache models, but both models will only support a 64-bit (single channel) DDR400 memory controller.  The first Turion 64s will be available in speeds ranging from 1.6GHz up to 2.0GHz.

With the Turion 64 AMD is introducing a new model numbering system to help differentiate various Turion CPUs from one another (and to separate the Turion 64 line from the Athlon 64 line):

As the chart above shows, currently AMD has two Turion lines - a 35W TDP and a 25W TDP line (note that Intel’s Pentium M 755 has a 22W TDP but they are measured differently from AMD).  The second letter in the model number indicates the level of power consumption of the processor, with “higher” letters denoting lower power consumption (e.g. MT-34 has lower power consumption than the ML-34).  AMD is clearly doing a bit of chip binning, with chips capable of running at lower voltages (and thus lower power) being set aside for the Turion 64 line.  This isn’t much of a surprise as they do a similar thing on the server side to create their low power Opteron HE and EE chips (50W and 30W respectively). 

The two digits, as usual, indicate clock speed/cache size and other performance impacting features. 

Performance, Availability and Final Words
POST A COMMENT

43 Comments

View All Comments

  • yanon - Friday, March 11, 2005 - link

    "ASUS, Averatec, BenQ, MSI and Packard Bell are among the leading, global computer manufacturers who have indicated they will support AMD Turion 64 mobile technology."
    Packard Bell !!! Does it still exist?
    Reply
  • RockHydra11 - Thursday, March 10, 2005 - link

    If these are anything like Athlon 64s, they will be more balanced than Pentium M chips performance-wise in the big picture. This should give AMD an edge. Reply
  • alangeering - Thursday, March 10, 2005 - link

    #21, #24, #30, et al,

    Thanks guys, I spotted the single channel mem contorller soon after I posted so realised that it had to be the 754 part. I just wondered whether AMD would go to another socket for these processors to stop their use in other desktop PCs. It seems from what I'm reading that it is just a S754 arrangement.

    Oh well, this, I guess, means no dual core Turion? Or are AMD going to release a 754 dual core part? I thought dual core was going to be a S939 and newer thing.

    So when we see dual core Pentium-M, will AMD be able to respond?

    I do like the idea of 1Mb L2 cache, my S939 A64 3000+ is rather lacking in that area, but I think I will stick with S939 so I can upgrade in 2 years time (hoping for cheap dual core).

    Thanks again,
    Alan
    Reply
  • JarredWalton - Thursday, March 10, 2005 - link

    #11 - as #26 pointed out, nF4 is certainly capable of being put into a small space. The Shuttle XPC SN25P I just reviewed definitely has a smaller motherboard than uATX, an it could be made smaller yet if it were targeting a laptop configuration. Pictures are available in the article:

    http://www.anandtech.com/systems/showdoc.aspx?i=23...
    Reply
  • bersl2 - Thursday, March 10, 2005 - link

    #28: The thing that gives the P-M such performance might be its low latency---at least on cache hits, that is.

    http://www.anandtech.com/cpuchipsets/showdoc.aspx?...

    I think it will perform more or less equally well, with somewhat worse battery life. Hopefully, though, this will be offset by a lower price, which AMD often brings.
    Reply
  • abakshi - Thursday, March 10, 2005 - link

    Well this is probably perfect for AMD -- not many resources involved in developing it (just in adjusting A64 for a lower TDP and overall better mobile package), and a decent effect on the market.

    Dothan is a nice chip - at 1.6 Ghz, it matches most P4's in the low 2 Ghz range (on the P4 533FSB). At 2.0, on the 400 FSB, it matches upper level P4's, is furthered by the 533 FSB, and OC'ed, in many cases, it can beat the top performers in the market (A64 4000+, P4 EE) in certain areas, all at a power/heat level far lower than that of the Oven (oops I mean Prescott), Northwood, or AMD64 lineup.

    But P-M has some issues - floating point performance is weak, as is multitasking (which is not aided by the lack of HT). These two issues should be fixed by the dual-core setup. Now that will be very interesting...

    So I think Turion will be somewhere in the middle -- won't match P-M's power or heat efficiency, or its top-end performance, but it'll be a solid stopgap measure in between. Athlon 64 isn't the best multitasking performer either, and that'll pass on to the Turion, but it's not bad, and if adopted may give P-M a run for its money.

    Will I buy it, probably not :-) But it might be effective in the mid-range.

    Until we see dual-core P-M's on the market, I'll enjoy my P4 3.0 / 533 Northwood on my laptop. 2 hour battery is not too bad, if I can do some serious stuff while it's lasts :-)
    Reply
  • ZobarStyl - Thursday, March 10, 2005 - link

    While this isn't the greatest thing since sliced bread, I like the idea, since we will actually get performance and decent battery life. And for the guy who says that if Intel had just slapped a Centrino sticker on a P4, well, of course people would be mad, it would sear the flesh off your legs. AMD has a cooler, lower voltage part for desktop; it makes perfect sense to simply adapt it to mobile. Dothan is very nice but it has minor flaws; the A64 may just give Centrino as run for it's money. Reply
  • glennpratt - Thursday, March 10, 2005 - link

    #11 filterxg... Your wrong... Thier certainly seems to be no technical reason for not making miniITX NF4 motherboards.

    http://sys.us.shuttle.com/ModelsP_5.aspx
    Reply
  • bwall04 - Thursday, March 10, 2005 - link

    Reply
  • kb3edk - Thursday, March 10, 2005 - link

    #8 (alangeering) -

    As ozzimark mentions, this is going to be Socket 754. I think the chip has really good potential for SFF mATX systems as well as laptops.
    Reply

Log in

Don't have an account? Sign up now