The Inquirer-Home

2006 notebook chips mean an Intel-AMD standoff

2006 Ahoy! Same as it ever was, same as it ever was
Fri Dec 30 2005, 10:16
FOR THE SECOND part of our yearly look ahead, we go from desktop chips to laptop parts. This one is a little more confusing, and has a lot more cloud and smoke surrounding it than either desktop or server chips.

Either way, it starts out with Intel having a pretty clear lead in a lot of ways, but AMD gaining market share anyway. Turion was launched late, slow, and with all the furore of a frightened child hiding in a corner. The supply situation has gotten vastly better over the past several months, but only in the retail channel. For corporate laptops, Intel and the Pentium M still rule.

There are three main factors people buy a laptop for, size/weight, battery life and performance. Smaller is better, as is lighter, unless you have a desktop replacement - there, bigger screens are better. Lighter is always better regardless of the rest. Performance means CPU power in the corporate world, and GPU power in the retail sector. Kiddies don't crunch 10MB Excel spreadsheets, CxOs don't give a flying faggot about frame rates in Unreal 2007. Battery life is of paramount importance on the business side, a distant second to performance on the retail side.

If you look at where Intel starts the year, the Yonahs will be a little worse than Dothan on power consumption, 31W vs 27W TDP, but most likely a little better in real world battery use due to the next generation of power saving tech. Performance will be a lot better than Dothan, especially in the multimedia area. Rejoice all of you that rip DVDs on your laptop, unless you are smart enough to realise you have a single slow HD. Raw performance will lag the AMD Turions, but it will be a bit more efficient. The glaring elephant in the corner is the 32-bitness of the chip. If you are going Vista in a few months, does Yonah make sense?

AMD starts the year with Turions, the mainstream 35W ML series, and the scarce as hen's teeth 25W MTs. These chips may bracket Yonah on power, but real world power use is not TDP, and I would have to give the edge to the Pentium M team there. Against Dothan, AMD wins the performance race. Against Yonah, it doesn't have a chance, but Turions are 64-bit, and useless as that may be until Microsoft gets its act together on Win64, it is still a strong bullet point marketing tool.

One monkey wrench thrown into the mix here is chipsets. AMD incorporates the memory controller and the FSB into the CPU, Intel has an external FSB and memory controller. If you look at the power jump from the 400 to 533 FSB Dothans, you can see that the FSB is much more than a trivial power draw. The memory controller also sucks a few watts, so the 31W vs 25/35W comparison is a bit misleading.

One more thing to consider, Intel has ground up mobile chipsets with all the latest and greatest power savings features baked in from the very start. AMD has third party chipsets with often conflicting goals. Does Nvidia want an all out performer or a battery sipper? Most of the AMD based chipsets are rehashed desktop parts, I don't know of a single chipset that is ground up developed for laptops.

Together, they about cancel out, with a possible tip of the real world hat to Intel. This is one area where unlike the desktop, Intel's platformisation strategy pays off. AMD is showing signs of jumping on this bandwagon, but it will be a while before that bears fruit.

I would expect AMD to grow a little in the business side of things, more due to the 'not Intel' factor than any inherent goodness. On the retail side, AMD will probably take a lot of market share. Yonah is good, but 64-bit, a butt-kicking GPU and a shiny case with a teen-angst logo will sell more here than 30 min of battery life. The vast majority of sales will be Intel when all is said and done.

In the spring, AMD launches dual core Turions, and they will catch up with Yonah parts on raw performance, possibly moving far ahead if AMD is willing to eat a little battery life. We hear the name will be the Turion X2 MA- series, but that may be an inside joke. Actually, the dual cores will be competitive at retail, but not for corporate use, I think they won't have all that much of an impact there, but for gaming laptops, they are the silicon to show up with at the LAN party.

As with Conroe, the Merom launch at the beginning of Q3 will be a shot in the arm for Intel. It will be a little less potent than a desktop Conroe, again coming in a few speed bins down for battery purposes. Intel will also hamstring the FSB for battery life, and that will cost quite a bit in the benchmarks. Even with that in mind, expect the top parts to be in the low 2.x GHz range, and have a draw of 35W. We am hearing conflicting rumours of average power usage, some say that it will be less than Dothan, but have a higher TDP, others say it will take more. I would lean towards the less average use because of the design tradeoffs, specifically power down as the default state.

Even with all that effort, the 35W draw will keep it out of the thin and light category, no question there. Look for only mid sized and DTR Meroms until the 45 nanometre shrink. The lowest end of the market, 12-inch screen one spindle notebooks will all be Yonahs, and those chips will continue to be sold until 45nm parts are common.

After the Merom launch, Intel will have a pretty convincing lead in most areas until AMD jumps to 65nm too. Again, the F-Step (FS) goodies we talked about yesterday, and the lower power draw of 65nm will both add up to a more efficient mobile CPU for AMD. Don't expect miracles though, it will be enough to keep them in the game, but not to vault into the lead.

We would expect the added headroom from FS and 65nm to not translate directly into a new gen of parts, or a lower power draw grade, IE Mx where x>T. Instead, AMD will probably skew production to raise yields and profits, those whacky capitalists, isn't this a technocracy?

2006 will close with Intel about where it is now in the mobile world, ahead in corporate sectors by a large margin, AMD with a healthy chunk of the retail sector. The same caveats apply as they did on the desktop, Intel has to deliver on the cores, AMD on 65nm. A screw up for either would hand the sector to the other side in short order.

Intel will have the lead, AMD will be a serious player, and the sun will continue to rise in the morning, set in the evening. We hope. Nothing earth shattering here, the battleground is 45nm, things are coasting until then. µ


Share this:

blog comments powered by Disqus
Subscribe to INQ newsletters

Sign up for INQbot – a weekly roundup of the best from the INQ

INQ Poll

Heartbleed bug discovered in OpenSSL

Have you reacted to Heartbleed?