Too little too late?
The single-core (two threads) Z670 runs at 1.5 GHz and has 512 KB L2 cache. The 45 nm CPU is combined with the SM35 I/O controller and comes with the GMA 600 graphics core (which is based on a PowerVR SGX 535 chip) as well as support for up to 2 GB of DDR2-800 memory. Oak Trail runs at about 3 watts TDP, which is not really low in comparison to the original Atoms and not low in comparison to its ARM rivals.
It's good news that Intel has finally released an tablet processor, but I doubt that this platform can dent ARM's dominance in the market. The Z670, which continues the rather unsuccessful original Atom Z500 series with Silverthorne core, has not the specs that would allow it to attract much interest - it is state of the industry at best (while we do have to wait for first benchmark results.) It is a single-core processor that will run 1080p video, but that's about it. Atom's rivals are dual-cores right now, and they are moving to triple- and quad-core versions soon. It will be difficult to score any meaningful design wins right now, even if Intel says that Oak Trail will be found in 35 tablet designs soon.
Its biggest problem may be its price. The Z670 is priced at $75, which is not only more expensive than the comparable (but more power-hungry) 1.83 GHz N470, which targets netbooks. I am not quite sure who is setting the pricing at Intel, but a chip that costs Intel about $5 to produce (according to our industry sources) and a chip that needs to compete with the iPad in price regions as low as $500 for entire tablets is not exactly aggressive. Intel may have killed Oak Trail right out of the gate. These processors need to be extremely cheap as they have become commodity devices that are not really in the focus of consumers anymore and therefore cannot carry a premium.
Oak Trail won't have a shot at the market today. We will wait for the 32 nm Cedar Trail platform in H2 of this year and hope that the chip will be faster, will come as a dual-core version, consumer less than half the power of Oak Trail and cost about one third of today's product.
Thats what they did. They took the basic P54C arch and stripped it of a few things and put it on 32nm.
Thats Atom.
What they need is to push out 22nm Atoms ASAP. That would help power drain a lot.
*cough* 45nm *cough*
Anyway, Brazos is taking away the light on low power set ups IMO. Atom has little to nothing to do now, even more when you think about "GPU".
Cheers!
Z670/650 is supposed to compete with ARM chips. IE A8/A9 Cortex and the variants of that. Mainly a tablet CPU I would expect.
Personally, this isn't going to be able to compete with ARM. It has a higher power consumption than ARM, and ARM is already ahead of the game with the A15 coming out next year.
I say Adreno 220 + A9. That or next year with the Adreno 330 and the A15 Cortex.
Cheers!
+1
I do think that tablets will require a CPU/GPU with more power than the ones currently on the market, but battery life is still king when it comes to tablets and phones. I do think we'll see intel-powered tablets and phones hit the market in mass quantity, but 45nm just won't cut it in terms of power consumption. I think Intel went that route to decrease the cost, but when they're already so damn cheap I don't see why they're charging so much.
If not, then I'll go with ARM and Android. I became rather fond of the Android system (where it not for the programs that autoboot into sys memory).!
No offense, but you're clueless, and you're spreading misinformation. The Atom is NOTHING like a Pentium. Where do we start? For one, the Atom has over three times as many stages. It runs it's pipelines independently, whereas the Pentium pipelines were lock stepped, and the V pipeline couldn't even execute that many instructions at all. Since we're talking about instructions, the Pentium actually executed 386 instructions, instead of changing them into other instructions and executing them (called decoupled architecture). Pentium also had no support for any version of SSE, or x86-64.
In short, they were virtually completely different.
I don't think the P5 could be used successfully, however modified, because the days of actually executing x86 instructions are probably gone for good on CPUs. Larabee was going to use modified P5s though, but that was intended as a GPU.
I always thought the K6 would come back as a low power chip, and although I've heard rumors that the Bobcat is based on it, when you look closely, if it is, it's old very moderately. I guess it's just better to start from scratch.
I think the new oak trail 45nm is more of a test bench to see the market audience. Frankly, despite the low power consumption with respect to the atom predecessor it's still out of the ballpark in comparison to the ARM-based alternatives. Marketing it as a windows-based tablet I don't think is a good idea either. The computing power is almost certainly decreased, and windows isn't exactly known for being a sleek and power-sipping, OS. I foresee somewhat of a resource hog that has the appeal of running your desktop apps at a snail's pace.
http://www.xbitlabs.com/news/mobile/display/20110412170720_Intel_Designs_Cloverview_System_on_Chip_for_Tablets.html
the future seems far more brighter for intel, though. These planned chips seem to be far more suited to the environment they're currently targeting. Bear in mind, though, AMD has wichita as well and other chips based on similar nm-scale with what will almost certainly be better graphical/gaming performance. In the tablet world that seems to be the latest trend.
We all know once the comparison charts come out everyone will want it. In comparison the the ARM processor the Oak Trail will most likely be 2x faster in most benchmarks.