Intel dishes new details on Apple-bound Silverthorne chip

[AppleInsider]

Intel has revealed some of the first official details of its upcoming lower-power Silverthorne processor that will soon turn up in a handful of next-generation mobile internet devices from Apple and other electronics manufacturers.

Presenting at the Solid State Circuits Conference (ISSCC) in San Francisco this week, the world's largest chipmaker issued 15 technical papers covering some of its most recent advancements. Atop that list is Silverthone, the code-name for the company's forthcoming 45 nanometer high-k metal gate, low-power processor architecture for ultra-mobile and mobile Internet devices.

Intel says the microarchitecture will be fully compatible with the Core 2 Duo instruction set, meaning Silverthorne chips will be capable of running the same applications written for Core 2 Duo-based notebook PCs.

At their peak, chips running on the new architecture will deliver performance inline with that of the Pentium M chips that powered the first array of Centrino notebooks, but consume only between 0.5 and 2 watts -- about 10 times less than a typical notebook chip.

Silverthorne processors will also boast support for a low power mode and hyperthreading, according to Intel. The first will allow the chips to shut down between tasks for optimal power savings, while the second will allow for execution of multiple simultaneous threads on a single core -- essentially emulating a dual-core chip.

Thus far, no specific chip numbers or clock frequencies have been released by Intel, though the firm maintains that the first production quality units should arrive some time during the second quarter of the year, with successive models eventually scaling up to 2.0GHz.



Intel chief executive Paul Otellini has already gone on record in saying that his firm plans a whole "product family" of 45 nm Silverthorne chips in the near future aimed at capturing the "top 10 to 20 percent of the cellphone market.”

Chief among the players aiding Intel on its quest for a slice of the high-end cell phone market is Apple, which AppleInsider reported in December would be among the first electronics makers to adopt Silverthorne chips.

[solipsism]

How do these compare to ARM power consumption? I thought ARM used quite a but less. If it does then these aren't a good candidate for iDevices.

[THT]

Um, there isn't much food on this dish. Nothing in fact. There isn't anything new here. I'd like to know say: 1 GHz Silverthorne = ?.? GHz Core 2 Duo at ?? Watts TDP; or, 1.8 GHz Silverthorne = ?.? Core 2 Duo at ?? Watt TDP. This will be new information.

Ie, if we put a 1.8 GHz Silverthorne into a MBA, will it be equivalent to the 1.8 GHz Core 2 Duo (highly unlikely). Will it be like a 1.2 GHz Core 2 Duo, but will have double/triple the battery life?

[nagromme]

One Silverthorne iPhone with GPS and 32 GB storage, please!

Also please implement Intel's new "widejack" headphone interface technology

[wizard69]

Not to reiterate the poster above but more info is required! Like what is the overall power draw of the chip set at a specific speed and performance level. Also what are the different variants and how do they accommodate the various radio hardware required.

As to ARM, I just don't see how Apple is going to be able to jam a Silverthorne into an iPhone. Especially considering that ARM aligned companies have dual core SOC implementations in the wings.

Now Newton 2 that is another matter. I could also see a bigger iPod to better deliver multimedia. Things do look good for Apple and Intel!

Heck if Apple put one in an AIR like portable and added the missing I/O I might even be interested.

Dave

[melgross]

Originally Posted by THT

Um, there isn't much food on this dish. Nothing in fact. There isn't anything new here. I'd like to know say: 1 GHz Silverthorne = ?.? GHz Core 2 Duo at ?? Watts TDP; or, 1.8 GHz Silverthorne = ?.? Core 2 Duo at ?? Watt TDP. This will be new information.

Ie, if we put a 1.8 GHz Silverthorne into a MBA, will it be equivalent to the 1.8 GHz Core 2 Duo (highly unlikely). Will it be like a 1.2 GHz Core 2 Duo, but will have double/triple the battery life?

you're asking the wrong questions.

Intel has clearly said that these were for small mobile devices, not notebooks, even ultralights.

Perhaps, several years down the line, they may be powerefull enough for that purpose, but then, more standard power consumption chips will also have become more powerful.

The MD Air has what would have been considered to be a very powerful cpu three years ago, but today, it is the least powerful machine in Apple's lineup.

The same thing will be true in the future. People want more powerful machines.

But for the less demanding handheld device market, these will work well.

[GQB]

Originally Posted by melgross

People want more powerful machines.

I really have to question that.
Certain categories of users (probably over-represented on this kind of forum) certainly do. No argument about that.
But for the broad sweet spot of users who need web, mail and heavy keyboard input to office applications, and some media organization/viewing, the steroidal increase in power over the past few years has been overkill.
I can work just fine at the 4-year-old iMac we've passed down to our daughter.
What most user want now is battery life. They want a 100 MPG plug-in Prius, not a 15 MPG Explorer or Vette.
I just think we've hit the point of diminishing returns regarding power, at least until we see some broad-based killer app that really does require another quantum leap in processing power.
Now steroidal increases in bandwidth? Yup.

[teckstud]

Originally Posted by melgross

People want more powerful machines.

People want more affordable machines -not $2000 sub-par ultra-light laptops.
Though I am not most people and therefore will get one!

[melgross]

Originally Posted by GQB

I really have to question that.
Certain categories of users (probably over-represented on this kind of forum) certainly do. No argument about that.
But for the broad sweet spot of users who need web, mail and heavy keyboard input to office applications, and some media organization/viewing, the steroidal increase in power over the past few years has been overkill.
I can work just fine at the 4-year-old iMac we've passed down to our daughter.
What most user want now is battery life. They want a 100 MPG plug-in Prius, not a 15 MPG Explorer or Vette.
I just think we've hit the point of diminishing returns regarding power, at least until we see some broad-based killer app that really does require another quantum leap in processing power.
Now steroidal increases in bandwidth? Yup.

The history is that people (unless they really can't afford to) overbuy their computer power.

By more powerful, I don't mean that people who buy Macbooks will opt instead for MB Pros. I mean that within the price catagory, people want more powerful machines.

The Air, for example, is already accused of being too weak for some tasks when compared to the Macbook.

[melgross]

Originally Posted by teckstud

People want more affordable machines -not $2000 laptops.

My answer to you is pretty much the same one I just gave to GQB.

I'm not saying that people will go for the $2,000 machine. Your interpretation of what I said was wrong.

I'm saying that when the Silverthorn, or ARM chips, are as powerful as the Core 2 Duo's now powering the $1,100 Macbook, the chips powering that Macbook will also have advanced.

People will choose the machine with the more powerful configuration, and so Apple won't make a laptop with chips that will be much less powerful than those used in the Air.

With media becoming more important to people all the time, and constantly requiring more powerful machines to run the content, moving backwards in power won't be popular.

Really cheap laptops such as the ASUS might choose these chips, but they will be in the lowest price category that Apple won't compete in.

That's why this, and the ARM, aren't intended to be used for laptops. They are intended for smaller machines.

I would think that Apple would maintain the Macbook at the level it is now in the lineup. There would also be a less powerful machine such as the Air (possibly more than one model eventually), and more powerful pro machines.

Then they might have handheld devices that use the Silverthorn.

Lastly, they could maintain the iPhone line with even smaller, less powerful chips. The top iPods would follow along with the iPhone.

[JeffDM]

Originally Posted by wizard69

As to ARM, I just don't see how Apple is going to be able to jam a Silverthorne into an iPhone. Especially considering that ARM aligned companies have dual core SOC implementations in the wings.

The PortalPlayer chips used in iPods have two ARM cores on a SOC. I think the newer ones probably do too, but Apple doesn't say anything about what's in their custom chips.

[Rot'nApple]

Originally Posted by teckstud

Though I am not most people and therefore will get one!

Luv it and don't we all!

[THT]

Originally Posted by melgross

you're asking the wrong questions.

Intel has clearly said that these were for small mobile devices, not notebooks, even ultralights.

Perhaps, several years down the line, they may be powerefull enough for that purpose, but then, more standard power consumption chips will also have become more powerful.

The MD Air has what would have been considered to be a very powerful cpu three years ago, but today, it is the least powerful machine in Apple's lineup.

The same thing will be true in the future. People want more powerful machines.

But for the less demanding handheld device market, these will work well.

I can't see how I'm asking the wrong questions. All I want to know are performance numbers. The "how fast is Silverthorne per Watt" question. I know Silverthorne/Menlow are from UMPC devices. The MBA question was just an example.

Is Silverthorne suitable for the MBA? Well, depends on the performance. From what I've read about it, it will definitely not be suitable for Mac OS X usage as it is only about as fast as a Pentium-M per MHz. It has the added wrinkle of 2-way SMT, so maybe a 1.8 GHz Silverthorne is like a 1.2 GHz Core 2 Duo in multi-thread, but like a 1.6 GHz Pentium-M in single thread. If it is, it will be an interesting proposition as a 1.8 GHz Menlow w/Silverthorne board will be on the order of 4-to-5 Watts while the SFF Merom board in the MBA is somewhere in the neighborhood 30 Watts. That's a 5 to 6x reduction in power versus with the MBA Merom with a ~50 reduction.

With the screen taking up a 1/3rd of the power, this could mean 2x or 3x times the battery life (7/8 hours) for half the performance. That's an interesting value proposition to consider.

[melgross]

Originally Posted by THT

I can't see how I'm asking the wrong questions. All I want to know are performance numbers. The "how fast is Silverthorne per Watt" question. I know Silverthorne/Menlow are from UMPC devices. The MBA question was just an example.

Is Silverthorne suitable for the MBA? Well, depends on the performance. From what I've read about it, it will definitely not be suitable for Mac OS X usage as it is only about as fast as a Pentium-M per MHz. It has the added wrinkle of 2-way SMT, so maybe a 1.8 GHz Silverthorne is like a 1.2 GHz Core 2 Duo in multi-thread, but like a 1.6 GHz Pentium-M in single thread. If it is, it will be an interesting proposition as a 1.8 GHz Menlow w/Silverthorne board will be on the order of 4-to-5 Watts while the SFF Merom board in the MBA is somewhere in the neighborhood 30 Watts. That's a 5 to 6x reduction in power versus with the MBA Merom with a ~50 reduction.

With the screen taking up a 1/3rd of the power, this could mean 2x or 3x times the battery life (7/8 hours) for half the performance. That's an interesting value proposition to consider.

You want to know how this compares to a Core 2 Duo, when it isn't intended to compete with that, either on price, perfprmance, or purpose. That's why it's the wrong question.

The question is how it compares to the various ARM chips, and any others that compete on that level.

The performance of this is below the ULW chips Intel produces for comnputer use. We kow that, and Intel is marketing it as such.

Moving it to a higher level is like asking a flyweight to do battle with a middleweight.

[Kolchak]

Originally Posted by THT

I can't see how I'm asking the wrong questions. All I want to know are performance numbers. The "how fast is Silverthorne per Watt" question. I know Silverthorne/Menlow are from UMPC devices. The MBA question was just an example.

I don't really understand this need for speed. I really don't care about the performance numbers. It's for an ultramobile device, not a Photoshop demon. I love my iPod Touch and thanks to it being basically all I need when I'm away from home (unless I'm away from my main Mac for an extended period like when I take my laptop on vacation), I don't even feel the urge to buy a Macbook Air. I don't need much more power than the Touch already has and it's unlikely to be slower. I'd love to have more battery life, though, and I'd upgrade to iPod Touch v2 in a heartbeat if this chip gave it that.

[planetWC]

Exactly. The question is WHAT is the device these chips are to be used in going to offer us?

Playing MP3s?
Playing Videos?
Reading PDF docs?
reading/writing "office" docs?
Web browsing?
Instant Messaging?

You don't need a C2D for that. So Apple was working on a Tablet prior to the iPhone and took the multituch UI from that for the iPhone. Perhaps silverthorne will be used for a larger sized iPod that has a bigger screen. With that we'd get an eBook reader for free. The cost then gets to the cost of the display and the technology used for it. Or perhaps silverthorne would be used in a next gen iPhone.

[Splinemodel]

Originally Posted by solipsism

How do these compare to ARM power consumption? I thought ARM used quite a but less. If it does then these aren't a good candidate for iDevices.

It depends on the ARM, and also how many peripherals are integrated onto the die. The ARM core is inherently a lot more efficient than any x86, although a lot of newer ARM cores are becoming relatively complex. If Intel has a big enough process advantage to make up the difference, maybe the power figures are comparable, but it's too hard to say with the limited information.

One aspect of the ARM that will be difficult to beat is that it's licensed, and hence it's pretty easy to cheaply get an ARM SoC just the way you want it.

[THT]

Originally Posted by melgross

You want to know how this compares to a Core 2 Duo, when it isn't intended to compete with that, either on price, perfprmance, or purpose. That's why it's the wrong question.

The question is how it compares to the various ARM chips, and any others that compete on that level.

The performance of this is below the ULW chips Intel produces for comnputer use. We kow that, and Intel is marketing it as such.

Moving it to a higher level is like asking a flyweight to do battle with a middleweight.

I don't get what your problem is with what is an extremely simple question. I don't have any ulterior motives but intellectual curiosity on microarchitecture design. I'm not even asking because I think it'll compete with the C2D or PC class MPUs. It's really obvious that I'm not, and that this is for UMPCs.

Even the MBA question is one born out of intellectual curiosity. It's like I'm speaking English and you're interpreting it through a marketing droid filter or something. The board in the MBA is extremely small. Menlow won't be much smaller at all. And since the MBA is almost a tweener UMPC-laptop due to extreme thinness, application of Silverthorne in the MBA is an interesting question.

Aren't you even curious at all on how a Pentium-like (in-order, 1 FPU, 1 IU, some SIMD) performs using a modern process and at 1.8 GHz? Well I am. I'm interested in how many execution pipeline stages it has. I'm curious about it's memory latencies. It also could be very very interesting if it encroaches upon low end C2Ds. Very interesting from a marketing and production cost perspective for Intel.

[THT]

Originally Posted by Kolchak

I don't really understand this need for speed.

You're applying a motive where there is none. It's really a simple question. It was also dig on Appleinsider for delivering no new information.

You know that Silverthorne threads before this one? Go back and read them on it in Future Hardware. It contained a ton more information and better informed speculation than this one is turning out to be. In fact, it contained more information than Appleinsider's "news".

I really don't care about the performance numbers. It's for an ultramobile device, not a Photoshop demon. I love my iPod Touch and thanks to it being basically all I need when I'm away from home (unless I'm away from my main Mac for an extended period like when I take my laptop on vacation), I don't even feel the urge to buy a Macbook Air. I don't need much more power than the Touch already has and it's unlikely to be slower. I'd love to have more battery life, though, and I'd upgrade to iPod Touch v2 in a heartbeat if this chip gave it that.

You do know that Menlow (Silverthorne + Poulsbo core logic ASIC + I/O ASIC) is for UMPCs? UMPCs are about 10x as large (by volume) as the iPod touch. It will never fit in an iPod touch or iPhone. The reference board layout is likely larger than the iPod touch itself. It'll barely fit in the MBA. Many of the UMPCs that ship with it will be on the order 7" x 4" x 1" using 5" or 6" LCDs in slider, clamshell, and big ol candy bar form factors. (The iPod touch is 4.4" x 2.4" x 0.3").

Not only that, I bet 75% of Silverthorne processors will be running MS Vista! You would think that it'll be important to know how fast they'll run Vista.

Even Moorestown, an evolved Silverthorne with memory controller and graphics on-die; and one I/O ASIC for even smaller footprint won't fit in an iPod touch. Mooretown platforms need like 5x volume of the iPod touch and targeted for MIDs (mobile internet devices) which is at least one class larger than an iPod touch.

Apple may use these two platforms for some future MID/UMPC. All the more power to them. Whether they'll run mobile OS-X or Mac OS X is an interesting question. Knowing how nice they'll be while browsing the internet with WiFi or HSPA is an interesting question with the performance/watt an important element. All it'll take is an embedded Flash website or Java website or opening multiples of them at the same to bring it to its knees.

[JeffDM]

So, in short, AppleInsider staff really have no idea whether this chip is really Apple-bound like the headline asserts. Intel makes a lot of different kinds of chips, and most of them probably won't be put into an Apple product, especially given that Intel's looking to have about eight different power grades for their mobile chips.

[Kolchak]

Originally Posted by THT

Even Moorestown, an evolved Silverthorne with memory controller and graphics on-die; and one I/O ASIC for even smaller footprint won't fit in an iPod touch. Mooretown platforms need like 5x volume of the iPod touch and targeted for MIDs (mobile internet devices) which is at least one class larger than an iPod touch.

Did you see the picture at the bottom of this article? That doesn't look like it's 5x the volume of a Touch. 2x maybe. And I'm sure they could get that down in time, although if it meant a slightly bigger screen, I wouldn't mind the larger size. The Touch/iPhone is what I've been waiting for for decades. Something I can carry with me everywhere, every day, and do most of what I need. Not even a Macbook Air is an everywhere machine.

[THT]

Originally Posted by Kolchak

Did you see the picture at the bottom of this article?

Dude, I've posted pictures of the Moorestown mockup in these very forums already, saying how I love it, and have inspected virtually every Youtube video of it. It's an RDF class looking device!

That doesn't look like it's 5x the volume of a Touch. 2x maybe. And I'm sure they could get that down in time, although if it meant a slightly bigger screen, I wouldn't mind the larger size.

The iPod touch is 8 mm thick! 0.31 inches! The Moorestown demo is like 4 iPod touchs in volume, probably 5, at least. It sure ain't 0.31 inch thick. It's on the order of 14 to 16 mm. Plus, I've got a sneaking suspicion that its a fake demo.

The Touch/iPhone is what I've been waiting for for decades. Something I can carry with me everywhere, every day, and do most of what I need. Not even a Macbook Air is an everywhere machine.

Well, Menlow (Silverthorne) won't be fitting inside a palm-size device and Moorestown isn't for an iPhone or iPod touch sized device either. 32 nm, probably. Intel is trying to get there. Moorestown in a 18 mm thick device the footprint of the iPhone is possible I think, but Apple won't do that.

[THT]

This the only picture that I can find of an iPhone and the Moorestown demo device in the same frame is at the bottom of this Anandtech article.

[ksec]

I am just wondering of software support for ARM. I know ARM owes 70% of mobile processor market. But is that big enough to complete with x86?

I know x86 has always win because it is backward compatible. But since this time around ARM actually has establish a huge market.

[samurai1999]

Originally Posted by THT

I don't get what your problem is with what is an extremely simple question. I don't have any ulterior motives but intellectual curiosity on microarchitecture design. I'm not even asking because I think it'll compete with the C2D or PC class MPUs. It's really obvious that I'm not, and that this is for UMPCs.

Even the MBA question is one born out of intellectual curiosity. It's like I'm speaking English and you're interpreting it through a marketing droid filter or something. The board in the MBA is extremely small. Menlow won't be much smaller at all. And since the MBA is almost a tweener UMPC-laptop due to extreme thinness, application of Silverthorne in the MBA is an interesting question.

Aren't you even curious at all on how a Pentium-like (in-order, 1 FPU, 1 IU, some SIMD) performs using a modern process and at 1.8 GHz? Well I am. I'm interested in how many execution pipeline stages it has. I'm curious about it's memory latencies. It also could be very very interesting if it encroaches upon low end C2Ds. Very interesting from a marketing and production cost perspective for Intel.

These are all valid questions, I think
- I think people looking for this device in the next iPhone are barking up the wrong tree
- as you point out, the physical size (at least 2 chips + RAM) and power consumption (2-3x ARM?) just aren't right
- so, probably, iPhone2 & Touch2 will use a next generation ARM part

- the Silverthorne chip has very lower power consumption for an x86 chip
- and is powerful when compared to mobile devices, and as powerful as the mobile x86 chips of 3-4 years ago.
- so, I think we should look for this turning up in devices that sit below the MacBook Air in terms of performance - i.e. some form of ultra-portable device, but not as portable as an iPhone.
- perhaps a 6" Touch screen device?
- just a guess

- of course, that could also be done with an ARM-based device, so Silverthorne doesn't really give Apple much that it doesn't have already - since it's already got it's OS running on ARM, and Mobile versions key apps like the browser etc.
- I can see that running on x86 will help with compatability for running Flash-based web-sites, or running full versions of Word, but then that's probably not what you want to do with a 6" touch-screen device most of the time.

[Hattig]

Originally Posted by Kolchak

Did you see the picture at the bottom of this article? That doesn't look like it's 5x the volume of a Touch. 2x maybe. And I'm sure they could get that down in time, although if it meant a slightly bigger screen, I wouldn't mind the larger size. The Touch/iPhone is what I've been waiting for for decades. Something I can carry with me everywhere, every day, and do most of what I need. Not even a Macbook Air is an everywhere machine.

By the time its power consumption and volume could compete with current ARM SoCs, the ARM SoCs will be either even smaller (and more powerful than today, or even less power draw), or vastly more powerful themselves (with software compatibility on their side).

And Apple knows what it is doing with ARM. It's got 15 years history with the architecture.

[melgross]

Originally Posted by THT

I don't get what your problem is with what is an extremely simple question. I don't have any ulterior motives but intellectual curiosity on microarchitecture design. I'm not even asking because I think it'll compete with the C2D or PC class MPUs. It's really obvious that I'm not, and that this is for UMPCs.

I don't doubt your motives. I just don't see the purpose of comparing two chipsets that are so different in power, and speed, and that are intended for two entirely different purposes.

It's like asking how a sportscar compares to a subcompact. Who cares? The two aren't comparable, so it doesn't matter. Compare two sportscars, or two subcompacts. That's valid.

Even the MBA question is one born out of intellectual curiosity. It's like I'm speaking English and you're interpreting it through a marketing droid filter or something. The board in the MBA is extremely small. Menlow won't be much smaller at all. And since the MBA is almost a tweener UMPC-laptop due to extreme thinness, application of Silverthorne in the MBA is an interesting question.

The board is only small when compared to that of the MacBook. Otherwise, when compared to that of a UMPC, which is where this chip is intended to go, it is a giant.

Aren't you even curious at all on how a Pentium-like (in-order, 1 FPU, 1 IU, some SIMD) performs using a modern process and at 1.8 GHz? Well I am. I'm interested in how many execution pipeline stages it has. I'm curious about it's memory latencies. It also could be very very interesting if it encroaches upon low end C2Ds. Very interesting from a marketing and production cost perspective for Intel.

I understand what you are saying. But, I know that it will never challenge the computing power of the Core 2 Duo, Penyrn, and later this year, Nehalem chips. Because of that, my interest in that comparison is little. I know it doesn't compare, and so I don't care by how much it falls behind.

What I am interested in, is how it will help to advance a handheld, where those other chips can't possibly be used. A totally different category of device, which Apple may offer later this year, or the next.

[melgross]

Originally Posted by THT

This the only picture that I can find of an iPhone and the Moorestown demo device in the same frame is at the bottom of this Anandtech article.

The chip is still very small, and so is the associated circutry.

The device they chow is about the size of the one we have been talking about for a while—the Newton-like device.

This would work very well in that, and is, after all, the type of device it is intended for.

I do agree with Gartner's prediction here. In a few years, people won't carry laptops around everywhere (most who own them don't do that now, too big and heavy), but they will carry these smaller devices around.

[melgross]

Originally Posted by Hattig

By the time its power consumption and volume could compete with current ARM SoCs, the ARM SoCs will be either even smaller (and more powerful than today, or even less power draw), or vastly more powerful themselves (with software compatibility on their side).

And Apple knows what it is doing with ARM. It's got 15 years history with the architecture.

Except that the x86 now fits within Apple's plans much more than does the ARM.

With Apple moving OS X into many more devices, the ARM becomes the odd man out.

The ease in programming for the x86 across all of Apple's lines is obvious.

Third party developers would much rather deal with one architecture than two. even though programs for a handheld would be much smaller, it's still easier to move code over, than to rewrite it anew, or have to undergo a major re-compile and fix-up.

Apple now obviously has an "OS X Everywhere" goal, and running x86 everywhere will help to further that goal much better than will continuing with the ARM.

Eventually, the iPhones and iTouch's will also move over, perhaps when the 32nm versions come out next year.

[solipsism]

Originally Posted by melgross

Except that the x86 now fits within Apple's plans much more than does the ARM.

With Apple moving OS X into many more devices, the ARM becomes the odd man out.

The ease in programming for the x86 across all of Apple's lines is obvious.

Third party developers would much rather deal with one architecture than two. even though programs for a handheld would be much smaller, it's still easier to move code over, than to rewrite it anew, or have to undergo a major re-compile and fix-up.

Apple now obviously has an "OS X Everywhere" goal, and running x86 everywhere will help to further that goal much better than will continuing with the ARM.

Eventually, the iPhones and iTouch's will also move over, perhaps when the 32nm versions come out next year.

I look forward to that. I wonder if OS X will make it's way into the rest of the iPod line. Having one OS to rule them all, sans the Shuffle, would be very efficient, especially on x86.

[backtomac]

Much better article on Silverthorne over at Ars.

Stokes is very good at analyzing hardware IMO. I read his articles closely.

[solipsism]

Originally Posted by backtomac

Much better article on Silverthorne over at Ars.

Stokes is very good at analyzing hardware IMO. I read his articles closely.

That sums it up nicely:

"Ultimately, Silverthorne could be compelling for the Asus Eee PC form factor, and at 2GHz there's an outside possibility that it might find a home in a MacBook Air that's relatively underpowered, but has great battery life. But the MID form factor, at least in its Silverthorne combination, is dead on arrival."

[melgross]

Originally Posted by solipsism

I look forward to that. I wonder if OS X will make it's way into the rest of the iPod line. Having one OS to rule them all, sans the Shuffle, would be very efficient, especially on x86.

Eventually, processing will become small enough, and cheap enough, and will have enough memory available for even the cheapest, and tiniest devices to run a full OS.

We might see that between five and ten years from now. A device the size of the Shuffle will be a full fledged computer.

The challenge is then the interfacing. We've seen Si/Fi with people having one eye covered with the input (and eye controlled output) device. Maybe the current, primitive products, that do that will become ubiquitous. Maybe other, better, methods will become available. As we now know that direct thought can control prosthetics directly, that may become a new, widely used method.

[melgross]

Originally Posted by solipsism

That sums it up nicely:

"Ultimately, Silverthorne could be compelling for the Asus Eee PC form factor, and at 2GHz there's an outside possibility that it might find a home in a MacBook Air that's relatively underpowered, but has great battery life. But the MID form factor, at least in its Silverthorne combination, is dead on arrival."

His statement "relatively underpowered" says it all. It's why I don't agree that we will see it in an Air.

Right now, with the much more powerful chips, tha Air is considered to be "relatively underpowered" for the price, and size. I can't imagine the reception it would get if it were slower by half, or more.

Jon is very good at describing technology, as he should be, given that his degree is in computing architecture, but his predictions have been only so so over the years.

[backtomac]

Originally Posted by melgross

His statement "relatively underpowered" says it all. It's why I don't agree that we will see it in an Air.

Right now, with the much more powerful chips, tha Air is considered to be "relatively underpowered" for the price, and size. I can't imagine the reception it would get if it were slower by half, or more.

Jon is very good at describing technology, as he should be, given that his degree is in computing architecture, but his predictions have been only so so over the years.

It's interesting that Silverthorne appears ill suited for MIDs (according to Stokes) and would seem underpowered for something like the MBA.

What's the market for this chip?

[melgross]

Originally Posted by backtomac

It's interesting that Silverthorne appears ill suited for MIDs (according to Stokes) and would seem underpowered for something like the MBA.

What's the market for this chip?

This is the first incarnation. It's a very good first step. Intel will quickly improve it.

rememver that ARM has been this space for a long time. This is new. I hope that by now, no one here really doubts Intel's prowess.

I also doubt very much that Stoke's "...is deal on arrival" (for now) will be true. There will be a lot of designs done around this, and products will arrive.

The problem for those products is that they will be the same old thing from the same old manufactures, and so THEY will be dead on arrival, as have all UMPC's been so far.

Apple might wait for the 32nm version next year (or, who knows, perhaps late this year). By then, it will be much more viable, and no other chipmaker can stay on the schedule Intel can.

[MacGregor]

Man, the folks at Intel need to use Keynote and watch Jobs make informational slides for a while. What an ugly ppt mess that image is.

[Splinemodel]

I just looked into this today: For what it's worth, the PPC 750 variant in the Nintedo Wii would probably score the same (if not better) power figures than the Silverthorne if it were made at 45nm, and it would likely have superior performance.

I don't see it happening, but it's an interesting thing to note.

[superkaratemonkeydeathcar]

Originally Posted by THT

I don't get what your problem is with what is an extremely simple question. I don't have any ulterior motives but intellectual curiosity on microarchitecture design. I'm not even asking because I think it'll compete with the C2D or PC class MPUs. It's really obvious that I'm not, and that this is for UMPCs.

Even the MBA question is one born out of intellectual curiosity. It's like I'm speaking English and you're interpreting it through a marketing droid filter or something. The board in the MBA is extremely small. Menlow won't be much smaller at all. And since the MBA is almost a tweener UMPC-laptop due to extreme thinness, application of Silverthorne in the MBA is an interesting question.

Aren't you even curious at all on how a Pentium-like (in-order, 1 FPU, 1 IU, some SIMD) performs using a modern process and at 1.8 GHz? Well I am. I'm interested in how many execution pipeline stages it has. I'm curious about it's memory latencies. It also could be very very interesting if it encroaches upon low end C2Ds. Very interesting from a marketing and production cost perspective for Intel.

I understand your question completely.
0-60 times & hp#s are still fun to know, even if you're comparing a Smart Two to a Gallardo Superleggera.

melgross' problem is he has no life and the only way he can dhv or achieve any self-esteem is to be a gigantic fart-mongering condescension wonk here on AI, which is why he's not actually Listening to what you're saying.

You should see his google images photo. --Hell I'd be a fart-wonking Mac tool if I looked like "The Librarian" from the original "Blade" movie, too.

[JeffDM]

Originally Posted by superkaratemonkeydeathcar

You should see his google images photo. --Hell I'd be a fart-wonking Mac tool if I looked like "The Librarian" from the original "Blade" movie, too.

That's not the same person.