Pentium M Dual Core in January 2006 / Summary of the Intel Mobile Roadmap

With the flurry of new products from Intel coming up, we thought it best to recap a little bit on a CPU that is our personal favorite: Yonah. Intel will be introducing this highly anticipated dual core Pentium M CPU in January 2006. Yonah is based on the 65nm process technology. For the first few months of next year, it will coexist with the current Pentium M generation codenamed Dothan (based on the 90nm process technology). The Yonah based Pentium M's core also has architectural improvements for better SSE and floating point performance, two major weak points of the Dothan based Pentium M.

Intel currently identifies its dual core Pentium M series by using the letter "x" followed by the two digits. The chips with a leading "x" do not have a digit assigned to them yet, although we're already hearing some OEMs call these chips the "800" series. That will be a little odd, as there will be overlap with the already existing Pentium D line. However, Intel has done this in the past with the Celeron M and Desktop Celeron parts, so placing the chips in the 800 series seems likely.

Pentium M Roadmap
CPU Model Codename Clock (GHz) Cores FSB (MHz) Launch
TBD Merom TBD 2 667 2H'06
x60 Yonah 2.33 2 667 2H'06
x58 Yonah LV 1.83 2 667 2H'06
x50 Yonah 2.16 2 667 Jan'06
x48 Yonah LV 1.66 2 667 Jan'06
x40 Yonah 2 2 667 Jan'06
x38 Yonah LV 1.5 2 667 Jan'06
x30 Yonah 1.83 2 667 Jan'06
x20 Yonah 1.66 2 667 Jan'06
780 Dothan 2.26 1 533 Jul'05
778 Dothan LV 1.6 1 400 Jul'05
773 Dothan ULV 1.3 1 400 Jan'06
766 Yonah 1.83 1 667 2H'06
756 Yonah 1.66 1 667 Jan'06
TBD Yonah ULV 1.2 1 533 Q2'06
TBD Yonah ULV 1.06 2 533 2H'06
TBD Yonah ULV 1.06 1 533 Q2'06

The mainstream Pentium M dual core CPUs to be introduced in January will range from the 1.66GHz (Pentium M x20) to the 2.16GHz (Pentium M x50), similar in speed to all but the fastest current single core Pentium M CPUs. Intel will also be introducing two low voltage (LV) Pentium M dual core CPUs, the x38 (1.5GHz) and the x48 (1.66GHz), for the thin and light notebook and tablet PC markets. The front bus speed (FSB) for these new Pentium M CPUs will also be increased from 533 to 667.

Intel has not yet indicated if these dual cores will include Intel's Virtualization Technology (VT), but roadmaps did state that VT will be available in future Pentium M dual core CPUs such as the x60 (2.33GHz) and LV X58 (1.83GHz), currently scheduled to be released in H2'06. As with most of the Dothan Pentium Ms, all Yonah based Pentium M's will also have XD (Execute Disable Bit), EIST (Enhanced Intel Speedstep® Technology) and a 2MB L2 cache. The L2 cache for a dual core Pentium M will be shared between both cores, unlike the current Pentium D and Athlon X2 designs.

Unlike AMD's Turion mobile CPU, the Yonah based Pentium M's will not have 64 bit capability - at least initially. 64-bit will have to wait until the second half of 2006 with the arrival of the second-generation dual core Pentium M CPUs, codenamed Merom. Merom also doubles the L2 cache to 4MB. If you aren't aware, Merom is really just the mobile version of the new Conroe architecture that Intel recently announced, so it will have all of the other architectural changes planned for Conroe, and likely some low power performance tweaks as well.

Although dual-core will be the primary focus of the Yonah based Pentium M CPUs, there will also be a few single core versions of this CPU. Intel continues with the existing Pentium M numbering scheme for these CPUs, starting with the Pentium M 756 (1.66GHz) scheduled to be released in January, followed by the Pentium M 766 (1.83GHz) in H2 '06. They will not include VT.

The ULV Pentium M Yonah based CPUs, aimed at the imbedded market, will be introduced in the second half of '06. They are still unnamed, but will be 1.06GHz in both single and dual core and 1.2GHz in single core only. They will have an FSB of 533 and a 2MB L2 cache.

Celeron M Roadmap
CPU Model Codename Clock (GHz) Cores FSB (MHz) Launch
430 Yonah TBD 1 533 2H'06
423 Yonah TBD 1 533 Jan'06
420 Yonah TBD 1 533 2H'06
410 Yonah TBD 1 533 Jan'06
390 Dothan 1.6 1 400 Jan'06

The Celeron M 410 and the Celeron M ULV 423 CPUs, based on the Yonah single core, will be released in Q1 of 2006. Clock speed has not yet been determined for these CPUs. The main difference from the Dothan based Celeron M is the increase of the FSB from 400 to 533 and the improved core. In 2H '06, the Celeron M 420 and 430 will also be released.

The last two Dothan based M chips will also be released in January '06. They are the Pentium M 773 ULV (1.3GHz) and the Celeron M 390 (1.7GHz). Intel will price the Pentium M dual core processors competitively with the existing single core Pentium M, for a rapid transition from single core to dual core. The dual core Pentium M, with the dual core AMD Turion, will ensure that the notebook market will not be left behind in the dual core revolution.

Mobile Chipsets

There are four new mobile chipsets for the new Yonah based Pentium M CPUs. The two high performance chipsets are the 945PM and the 945GM, the difference being that the 945GM adds the Intel's GMA 950 (Graphic Media Accelerator). These two chipsets have a PCIe x16 connection, and support an FSB of 533/667MHz and up to 4GB of 533/667 dual channel DDR2 RAM. These chipsets will be released in January 2006.

Besides the high performance chipsets, Intel will also be offering a couple of lower-end chipsets. The third chipset is the 945GMS. This chipset supports an FSB of 533/667MHz with a single channel of DDR2-533 RAM. This will be released in Q2 '06 and is aimed primarily at small designs such as tablet PCs. The last chipset, the 945GML, will also be released in Q2 '06. This is a value chipset supporting an FSB of 533MHz and up to 2GB of DDR2-400 memory, again in single-channel operation. It will have the Intel's GMA 950 video solution, but will not have a PCIe x16 connection.

Comments Locked

34 Comments

View All Comments

  • StriderGT - Tuesday, October 11, 2005 - link

    Everybody sans the "intel"igent? fanatics seems to be really yawning with every next new spectacular intel river, whose bottom seems well bottomless...
    Moderate speed bumps, astronomical caches hiding the bandwidth limitation problems, no intergrated mem controllers (double the same problem with each stiched core), lack of 64bits (I remember the "who needs the 32bit" moto by the now defunct 3dfx in the 3d arenas), DDR2 that never delivered and now forced on us all since the memory foundries need to cater for the ignorant everyday intel shoppers. Well its sad but intel's enthusiast room is really empty nowadays...

    I know we must repent for defying the allmighty intel lord, but what can we do with such procs?
  • Xenoterranos - Wednesday, October 12, 2005 - link

    We can feed them to the beasts of incredulity and pander then unto those ignorant masses. Lo unto thee, for Dell is among you, and his mouth is vast and terrible to behold.
  • Anemone - Saturday, October 8, 2005 - link

    AMD is working on clockrate. Nobody yet has seen a Dothan go past 2.8 (assuredly) yet and frankly don't believe that Merom is going to suddenly allow Intel to beat that. Merom will up the complexity, aim at a wider (more operations per clock) method of operation and Intel (Itanium) hasn't done well in getting such designs up in clockspeed.

    That said, a current Dothan about equals an equivalemt 1mb cache A64 of the same clockspeed. So at 2.33ghz, you'd have to have a 2.4 (4800 x2 equiv) in a mobile format for AMD to keep up. They'll probably do that, but don't expect that AMD is going to have so much easier of a time than Intel. On the desktop AMD will be pushing 2.8 - 3.0 out of dual cores, which with the higher 667 bus should do quite well against a Conroe @ say 2.67 (remember Intel always clocks conservatively). So the only thing Intel is going to do on the desktop is "catch up" really, and you're going to have to wait till about a year from now to see it. Sorry, not very impressive for the size and resource rich company they are.

    As for chipsets changing, Intel will change Napa again, in H2 (which we currently understand is far into the last quarter of 2006) when they bring Merom. So their chipset change cycles are going to be less than a year, and yes, that qualifies as "frequent changing" especially since there is zero upgrade path for older chipsets. 1 speed grade self upgradeable is not an "upgrade path" but for some it is I guess. Oh well.

    Yonah will be quite the spritely chip and quite a few dual core aware apps should be hitting the shelves during 2006, so inevitably most users will find a use for it. That's one thing you can count on both companies to promote, software that can make each one's dual cores shine. However, Yonah is 32 bit. That saves energy, which is precious to a mobile user, but if current schedules maintain, you are going to be looking at less than 9 months before a "64 bit first, 32 bit as an afterthought" operating system will be introduced. That being Vista. So you'll have to ask yourself, is is worth the 1500-4k to spend on a system that in 9 months will look like last year's news? Many will argue that 64 bit won't be a big deal. Having watched the 16-32 bit revolution the first time, I've heard that argument before and it was bogus. Current tests on hastily written 64 bit software don't show tons of differences. Same for many apps on dual cores, no difference. However give dev's a year or bit more to sit and work on new 64 bit apps for Vista (written, according to MS, from the ground up for 64 bit machines) and you can bet the revolution will be swift and absolute.

    I just guess it's going to depend on how people want to spend their money, but honestly, underestimating the need for 64 bit on any pc purchase after Jan 2006, would be a grave mistake.

    :)
  • IntelUser2000 - Monday, October 10, 2005 - link

    Uh-huh. Conroe!!=Yonah. Plus, indications are that Conroe will run up to 2.93GHz initially. Conroe is wider, higher clocked, better architecture. We'll see how it turns out.

    667FSB on AMD CPUs won't do much, as that's HTT speed, and as some may know, HTT increase did nothing. There is DDR2 support, but we'll see whether X2's need more bandwidth.

    Intel only clocks conservatively for laptop chips, and mostly for Pentium M's.

    Merom: ~2GHz
    Conroe: Close to ~3GHz
  • stateofbeasley - Sunday, October 9, 2005 - link

    (1) Raising complexity does NOT mean that clockspeed will be lower. Compare the K7 Athlon to the Katmai P6. At 0.25u, the Athlon smoked the Pentium III in clockspeed, despite having a far wider execution engine and 4x the L1 cache. Again, wider execution DOES NOT necessarily mean lower clockspeed.
    (2) Napa/'s chipest will not change. The 945xM chipset was already modified some time ago to support 64-bit. (http://www.theinquirer.net/?article=21666">http://www.theinquirer.net/?article=21666, reporting in March that prototype Napa64 boards were already up and running).
    (3) Vista cannot be considered 32-bit as afterthought, given that x86 and x64 builds are concurrent.

    People need to check their facts before posting. This area of the site is starting to look like the P&N forum, with ignorance becoming widespread among the population.

    For the enthusiast, a Yonah machine may not satisfy the irrational craving for the latest and greatest, but by and large the average user does not care. 64-bit will not make Microsoft Office or Internet Explorer go any faster.
  • Anemone - Monday, October 10, 2005 - link

    Conroe @ 3ghz not in '06.

    Napa64 and Napa32 are two different chipsets.

    For Vista 32 bit is a "secondary consideration".

    I'd be careful quoting a year old Inquirer article as "fact"

    Comparing wider is not slower across AMD and Intel is not the statement I made. For INTEL, wider is always slower.

    If you are going to contradict facts, try using ones that make sense. And please don't quote your source as a rumour mag.

    thanks :)
  • stateofbeasley - Tuesday, October 11, 2005 - link

    You attack my source, and provide no source of your own, which is not very credible.

    Napa is NOT a chipset, it is a platform which uses the 945PM/GM series of chipset. The 945 chipset remains constant for Napa and Napa64.

    The general proposition is that Wider doesn't mean Slower. P6 was much wider than P5, but it was able to clock much higher. This defeats your proposition that for Intel, wider = slower.

    Your statements are factually incorrect, your conclusions are inaccurate, and you cite no sources. Therefore, your post is entirely lacking in credibility.
  • Anemone - Tuesday, October 11, 2005 - link

    Ok wider = slower
    um not to refute your 10 year old "P5" comparison but how about something oh a bit more recent, say Itanium?

    Napa is the reference to the chipset, however it can refer to the whole platform, but only when its brought to market, such as centrino. For now, all references are to the Yonah cpu running on the Napa platform. Don't know, but guess that would be a bit advanced for you.

    Quoting sources was only done by you to add weight to your arguments. Sadly your arguments are years out of date and supported by very little.

    Oh well.
  • IntelUser2000 - Thursday, October 13, 2005 - link

    quote:

    Ok wider = slower
    um not to refute your 10 year old "P5" comparison but how about something oh a bit more recent, say Itanium?


    Yes. Look at this though.

    Pipeline Stages:

    P5: 5 stages, 2-wide
    P6: 10 stages, 3-wide

    IA-64: 8 stages, 6-wide

    P5, the Pentium, clocked at highest 233MHz at 0.35 micron, with 5 stages
    P6, the Pentium II, clocked at 300MHz at 0.35 micron, with 10 stages]

    P6, with Pentium III, reached 1.1GHz with 0.18 micron on 10 stages(or higher according to some, Intel says otherwise: http://www.anandtech.com/cpuchipsets/showdoc.aspx?...">http://www.anandtech.com/cpuchipsets/showdoc.aspx?...
    P6, with Pentium M iteration, and with at LEAST 12 stage pipeline(much as 14 stages) reached 1.7GHz on 0.13 micron.

    IA-64, with Itanium 2, reached 1GHz with 0.18 micron on 8 stage pipeline.
    Itanium 2, reached 1.66GHz with 0.13 micron on 8 stage pipeline.




  • Shintai - Saturday, October 15, 2005 - link

    Dont use a bigtin CPU to compare. Intel (and others) clocks a servercpu under what it does for a desktop or even "small servers".

    It´s all about extra reliability and safeguards.

Log in

Don't have an account? Sign up now