Simple Intel answer to AMD Zacate

[randomaeiou]

Up to now, Intel has been deliberately handicapping Atoms to prevent them from cannibalizing more lucrative sales of other products. However, with the en-masse shift in that market segment to AMD's Zacate, such a strategy is realistically no longer viable.

A very simple fix, using existing technology, would be to create a Sandy Bridge (or even Ivy Bridge!) equivalent of the single ULV core Core2Solo SU3500, with HD3000 or whatever they have on Ivy Bridge, with HyperThreading and Turbo enabled. Given the 17w TDP of the i5-2537M, a halved version of it should take somewhere around 8-10w, which competes well against the C-series Fusions and should blow the E-series Fusions out of the water. If we talk about 22nm Ivy Bridge and FinFETs, the TDP of this chip could get even uglier for AMD...

With Turbo and HT, the single-threaded performance of the vastly more powerful SB/IB core will blow Ontario/Zacate, or likely even Wichita/Krishna out of the water. Aggressive downclocking and power management combined with a generous Turbo boost should provide an excellent mix of battery life and power on demand. Multi threaded performance is likely to be very similar, at least to Zacate and Wichita, maybe not the possible quad core Krishna's. Given that the target of this chip would be low power devices, possibly even tablets, the hope would be that multithreaded, CPU intensive applications would form a relatively tiny proportion of the expected workload, which further highlights the IPC advantage of SB/IB. With Sandy Bridge HD3000, the raw graphics power of this chip will also be better than a Zacate's 6310 - unfortunately driver support might be a sore point.

Not sure about costings though... Given that the 1k units price of i5-2537M is $250, estimated somewhere between $150 and $200? Unfortunately the e350 will probably beat that hands down. However, to Intel's advantage is that all the R&D and fab design work has already been done, thereby massively reducing production costs as well as allowing a product to come to market quickly. Cutting out however much die area a SB core takes up, along with its cache, along with the reduction in complexity from having a single-core setup, will further drop production costs.

In summary, an Intel chip to counter the Bobcat threat would be something that looked like a ~1.0-1.3GHz single core SB with hyperthreading, turbo to 1.5-1.8GHz (okay, maybe not turbo since it's single core, but very aggressive power gating and ramping, you get my drift) and HD3000 graphics, with a sub-10w TDP and hopefully a sub-$150 1k unit price. Quick back-of-napkin calculations based upon 4C12EU, 2C12EU and 2C6EU die area estimates puts a 1C12EU processor at ~100mm2 on 32nm, which is quite a bit bigger than Zacate's 75mm2 on 40nm, but hey...

Whaddaya thunk?

And if they REALLY want to get funky, incorporate a single 32GB or 64GB flash chip into the companion southbridge, particularly if the single core is going to be Ivy Bridge, with its rumoured inherent ability to use SSD for write caching. This way, no matter how you look at it, the small SSD will be beneficial, either as a standalone drive in tablet-style devices, thus removing the need for another drive, or as a cache for a spinning drive in laptop and HTPC form factors, as the rumoured upcoming 20GB Intel SSD is going to be for.

 

[pxc]

Whaddaya thunk?

32nm Atoms are coming later this year, with clock speed bumps. The 2GHz+ Atoms will match the Zacate C50 in single threaded CPU performance and add a nice (further) boost to multi-threaded performance. The Zacate C60 will probably be out first. Graphics performance is supposed to double GMA 3150 (GMA 950) performance in those 32nm Atom chips.

I don't think there is much Intel can do on the mini ITX side with Atom. Brazos uses 2x more power and can crank up performance because of that.

Atoms are very cheap for Intel to produce. The performance is segmented to prevent cannibalization of higher speed processors. AMD is dealing with the opposite problem. Since Zacate was released, it has taken 50% of AMD's mobile sales without expanding market share. IOW, the "success" of Zacate is what caused AMD's mobile ASPs to fall from $42 to $37 in just one quarter. And that doesn't count the losses for IGP chipsets or discrete GPUs. Ugly times ugly.

Any way you cut it, a much larger and more expensive SB processor is not going to fit into the Atom's price segment, and definitely not into Atom's power segment. The just released Atom Z670 + chipset = 3.75W TDP, with typical power usage while running non-gaming apps is under 1W. Manufacturers are free to make other low cost computers from low end Nehalem-based or upcoming SB-based chips. IIRC, the cheapest LV chips are 2-3x more expensive than Atom chips and that will cause price increases for any laptop that uses those.

Atom is not going to become a high performance processor until it hits 22nm in a couple more years.

 

[Abula]

I think intel is fine, they will have new atoms for 2012 that might fight the price side, but atm intel has

Intel® Core™ i5-2537M Processor (1.4ghz 2.3ghz with turbo).
Intel® Core™ i7-2617M Processor (1.5ghz, 2.6ghz with turbo)

we might see something like a 2400 or 2300 series, maybe 1.0ghz and 2.0ghz with turbo, but atm i think both of the above are really nice opitons in the mean time the new atoms are launched.

I don't think there is much Intel can do on the mini ITX side with Atom. Brazos uses 2x more power and can crank up performance because of that.

Well intel on mITX also has the i3 2100T which for cpu is much faster than brazos, but also more expensive, i still think they are no matching with products, but there are alternatives.

 

[pxc]

Well intel on mITX also has the i3 2100T which for cpu is much faster than brazos, but also more expensive, i still think they are no matching with products, but there are alternatives.

I meant any low priced Atom to compete against Brazos in the higher performance low end.

I'll probably upgrade the i5 661 in my miniITX PC when low power Ivy Bridge processors arrive.

 

[randomaeiou]

I think intel is fine, they will have new atoms for 2012 that might fight the price side, but atm intel has

Intel® Core™ i5-2537M Processor (1.4ghz 2.3ghz with turbo).
Intel® Core™ i7-2617M Processor (1.5ghz, 2.6ghz with turbo)

we might see something like a 2400 or 2300 series, maybe 1.0ghz and 2.0ghz with turbo, but atm i think both of the above are really nice opitons in the mean time the new atoms are launched.

Well intel on mITX also has the i3 2100T which for cpu is much faster than brazos, but also more expensive, i still think they are no matching with products, but there are alternatives.

Well, what I'm suggesting is essentially Intel® Core™ i3 (Pentium branding maybe?) -2307M Processor, Single Core, HT and Turbo enabled, (1.0GHz, 2.0GHz with turbo), ideally 6-7W TDP. (Realistic considering the 5.5W TDP of the C2D SU3500 clocked at 1.4GHz).

Yes, Atom has a TDP of 3.75W, however, there are articles out there about how much energy it takes to complete a task (a combination of BOTH time and processor speed), and Atom actually comes out most energy INefficient, and ironically the fastest processors (2600K, 980X) are the most efficient - they get it done quickly then power down again.

Therefore, in terms of battery life, my proposed single core ULV SB processor should not be that far behind, if at all, when compared under heavy load to Atom/Brazos. The main challenge is power draw at idle, hence the suggestion of a very low base clock which can then be aggressively upclocked via Turbo, thus conserving energy at idle and providing power when needed.

As I said again, cost is likely an issue. However, Intel is currently pursuing next-gen 32nm Atoms for tablet/embedded computing, and that seems like a year or so away to market. A single core SB processor could realistically come to market within weeks to months, as a lot of the work is just thinking about what to remove, rather than a redesign, a new manufacturing process, and all that. This will allow Intel to compete here and NOW, and prevent ARM-based devices from grabbing further mindshare, and consolidating its position, particularly with the imminent arrival of ARM support on Windows 8 further eroding the Wintel ecosystem.

The 12 inch Asus tablet based upon an ULV i5 is something I had in mind, but the price is slightly steep. Intel will have a winner if they can combine the proposed single core, with a good form factor (slim, maybe 12 inch screen with a thin bezel that is close in size to the typical 10.1 inch tablet, thanks to LG's new ultrathin panels?). The first "golden point" in battery life is approx 4-5h light use (video, flash, surfing, word processing, etc.) AND ~8-10h total battery life (about 5-6h of standby on top of 4-5h of light use), which is about the length of an average workday with a bit for a commute on either side of it - if this combination can hit that target, then the product will sell, no matter the *small* premium on it, particularly since it can double up as a fully fledged laptop with some sort of keyboard dock - something ARM processors are not so good for at the moment with patchy office suite and general software support.

 

[Abula]

I see your point, and while its a good one and valid, we are just in a time that what you are looking for doesnt exist, Intel has no plans to release a more capable ATOM this year, there will be reviesions with higher clocks but still its going to be based on the same arq, until 2012 i woudlnt touch atoms.

I do agree into the prices of this low clocks SB mobile cpus are too high, and a release of an i3 with the caracteristics you suggest could be a good thing, but at the same time CULV cpus are already on touching the toes of atom, SU2400/3500 laptops were getting close to ATOM dual core netbooks, if a chip is released with such eficiency and low price, then not much reason to devolop atom.

Hope you find what you looking for, just not the right time atm with intel releases.

 

[pxc]

Yes, Atom has a TDP of 3.75W, however, there are articles out there about how much energy it takes to complete a task (a combination of BOTH time and processor speed), and Atom actually comes out most energy INefficient, and ironically the fastest processors (2600K, 980X) are the most efficient - they get it done quickly then power down again.

Therefore, in terms of battery life, my proposed single core ULV SB processor should not be that far behind, if at all, when compared under heavy load to Atom/Brazos.

You're trying to have your cake and eat it too.

Nothing is free. Atom is energy efficient, and so are Sandy Bridge chips. Lowering the clock speed and decreasing voltage on a SB down to Atom competitive power levels will give you a CPU with Atom-like performance. You mention that as much in your scenario. Measuring whole computer energy efficiency takes all power into account, including keeping the video card and drive(s) running. A slower SB will sink in that measurement exactly as Atom or Zacate does.

Zacate has the same problems. When AMD clocks it down to fit a 5W TDP, the CPU core is *slower* than an Atom at the same power level and now with 45nm (45LP) 3W Atoms +0.65W chipset, the power consumption difference is even bigger (5W Zacate went from a few % to 40% less power efficient than Atom). When 32nm SoC Atoms are released later this year it will widen that lead (performance/W) even more. Of course AMD's own 28nm Zacate successor next year should help keep it competitive.

Intel doesn't make single core SB chips. A dual core SB is approximately 5x larger than the cheaper Atom. What you're proposing is something that wouldn't benefit Intel at all. I agree that it would be nice to have a cheaper Pentium G/Core i3 LV/ULV mobile CPUs, but it isn't going to happen the way you propose.

 

[randomaeiou]

You're trying to have your cake and eat it too.

I'll take a black forest cake with icing on top please

Nothing is free. Atom is energy efficient, and so are Sandy Bridge chips. Lowering the clock speed and decreasing voltage on a SB down to Atom competitive power levels will give you a CPU with Atom-like performance. You mention that as much in your scenario. Measuring whole computer energy efficiency takes all power into account, including keeping the video card and drive(s) running. A slower SB will sink in that measurement exactly as Atom or Zacate does.

Zacate has the same problems. When AMD clocks it down to fit a 5W TDP, the CPU core is *slower* than an Atom at the same power level and now with 45nm (45LP) 3W Atoms +0.65W chipset, the power consumption difference is even bigger (5W Zacate went from a few % to 40% less power efficient than Atom). When 32nm SoC Atoms are released later this year it will widen that lead (performance/W) even more. Of course AMD's own 28nm Zacate successor next year should help keep it competitive.

I know that nothing is "free" - computational power involves pushing electrons around, which involves energy. However, it's about reducing wastage - no point keeping a relatively high baseline idle computational power active doing nothing, particularly as a lot of tasks nowadays require but a fraction of the maximum computational power of even a Brazos/Atom class chip, for example reading pdfs and editing simple office suite documents on word, powerpoint etc. Hence my argument for a SB processor downclocked *at rest* to Atom-class levels of efficiency/power consumption to maintain brilliant idle battery life, but with the *potential* to Turbo up to SB class processor levels to complete more computationally challenging tasks quickly and efficiently.

We all know that Intel spends craploads on R+D, and it shows in the general efficiency and power of their x86 cores - I doubt there's much dispute there. Their Achilles heel up to now has been graphics - traditionally not worth the silicon they are printed on and the watts they burn, and still no match for NVidia/AMD offerings. However, HOWEVER, Brazos has shown the utility of a 6310 class GPU in a cheapish offering. Now, an Intel HD2000/3000 class GPU is actually right around that ballpark, so there's not much to say there - they DO have the capability to match AMD on their turf now (caveat: software driver support a bit dodgy). Hence my argument of a single core ULV SB with HD3000 which will annihilate Brazos on every one of its selling points with the single exception of price. Brazos is around 75mm2, based on earlier estimates in my post above, a single core SB will come around 100mm2, which really isn't all that far off. Sure, it's 33% more than Brazos, but it sure as hell is quite a bit more powerful too!

I'm well aware of Intel's policy of segmenting Atom from CULV. My argument is that Atom, in its current incarnation, is fatally flawed when compared to Brazos, and I personally believe that Wichita/Krishna will make 32nm Atoms obsolete rather quickly once again. And the main point of this, is that Atoms are not a shoddy CPU given their cost and power consumption, but that the GRAPHICS SUXXXXXORS! With a decent HD2000/3000 class GPU on board, that would altogether neutralize half of Brazos' main selling points - GPU performance. The other being cost, which, frankly, would you rather sell less (relatively) expensive chips, or have a warehouse full of really cheap chips no one is going to buy?

The other possibility, which just occurred to me typing the previous paragraph, would be to add 6 or 12 EUs to the Atom - problem sorted. Can you imagine a 2C4T Atom with 12 EUs in a 32nm package? That'd give Brazos and likely Wichita/Krishna a serious run for their money...

Intel doesn't make single core SB chips. A dual core SB is approximately 5x larger than the cheaper Atom. What you're proposing is something that wouldn't benefit Intel at all. I agree that it would be nice to have a cheaper Pentium G/Core i3 LV/ULV mobile CPUs, but it isn't going to happen the way you propose.

I know single core ULV SB doesn't exist at the moment. However, Intel have done SU3500 before, and it co-existed with Atom N330. Given the "threat" of Brazos and Fusion at the moment, why not go back to that strategy, particularly with improvements in the IGP/IPG?

 

[pxc]

Given the "threat" of Brazos and Fusion at the moment, why not go back to that strategy, particularly with improvements in the IGP/IPG?

Neither of those is a "threat" to Intel. Both of those products simply cannibalize AMD's own sales of higher priced CPUs, IGP capable chipsets and/or discrete GPUs.

The graphics performance of Llano is somewhat poor, considering that the whole Fusion fiasco nearly killed the company. AMD claims the 400SP Llano GPU is 3x faster than HD Graphics 2000 (6 EU). Great really, but HD Graphics 2000 is 1/2 the speed of HD Graphics 3000 (12 EU). In several months when Ivy Bridge is released, the graphics performance difference will be pretty small on average vs Llano. $6 billion of debt and becoming a fabless semiconductor company for a 6-7 month lead. Ruiz was an idiot.

LOL, I'm sure Intel is begging for mercy at this point.

What is a threat in general to AMD and nvidia GPU sales is Bobcat, Llano, 32nm Atom, Sandy Bridge and Ivy Bridge. For many typical uses, none of those require a discrete GPU anymore.

Anyways, back on topic... the SU3500 is a dual core chip with one core disabled. It came out 2 years ago and had a list price of over $260. If you look at the current Intel price list, there are newer processors that have surpassed it at lower prices (although not at the same TDP... the much cheaper and similar Penryn-based Celeron ULV 763 1.4GHz has a 10W TDP): http://files.shareholder.com/downlo...C32-B57C-48A17B68AB72/Apr_11_11_1ku_Price.pdf If there were any kind of demand for a particular core/speed/power config, Intel would make it available. The SU3500 was a very expensive niche product.

 

[tybert7]

The tone of the op is almost regret that amd has a more compelling product in the low end.

Oh no! intel is not way ahead of amd at EVERY segment of the market anymore! que x/y/z contingency for the atoms STAT!


...

Let them have a brief moment in the sunlight, intel will do their thing on their own.

 

[tybert7]

... AMD is dealing with the opposite problem. Since Zacate was released, it has taken 50% of AMD's mobile sales without expanding market share. IOW, the "success" of Zacate is what caused AMD's mobile ASPs to fall from $42 to $37 in just one quarter. And that doesn't count the losses for IGP chipsets or discrete GPUs. Ugly times ugly.
...

For now zacate is the only truly new chip amd has on the market, when llano hits I think that could do well for amd as well. It cedes pure cpu power to intel where it cannot compete, but offers good enough cpu power plus much more powerful base graphics power. I think gamers will enjoy having cheaper notebook options that can handle games, especially if adding in a discreet gpu can work in tandem with the graphics component of the apu for an even greater boost.

Lets say there is a llano chip with cpu performance on par with pre sandy bridge i3/i5 speeds but with vastly improved graphics?

That seems like a compelling offering to me, but maybe that's just me.

http://www.youtube.com/watch?v=XqBk0uHrxII

 

[pxc]

Lets say there is a llano chip with cpu performance on par with pre sandy bridge i3/i5 speeds but with vastly improved graphics?

If Intel stopped releasing new processors after the 2+ year old Nehalem-based CPUs, I think Llano would have been very compelling. It would have been cool too if Llano didn't have to recycle the old, tired K10 core and had a new architecture like the competition does every 2 years.

What Llano needs to do to be successful is reverse AMD's general trend of losing desktop/laptop market share (or stemming losses using massive price cuts) since it is the new mainstream volume processor. Not only that, but it also needs to command a price to cover the loss of low end discrete GPU sales that are no longer required.

The problem though is that Llano will be competing against Sandy Bridge chips now, and Ivy Bridge chips in several months, and GPU performance is not the #1 selling point for systems to the general public. Unfortunately for AMD, Llano, like Bulldozer will face low price ceilings due to CPU performance. Llano's die size is close to 2x bigger than dual core SB chips it will be competing against. Not exactly a recipe for success: high manufacturing cost, low price it can sell for.

 

[tybert7]

If Intel stopped releasing new processors after the 2+ year old Nehalem-based CPUs, I think Llano would have been very compelling. It would have been cool too if Llano didn't have to recycle the old, tired K10 core and had a new architecture like the competition does every 2 years.

What Llano needs to do to be successful is reverse AMD's general trend of losing desktop/laptop market share (or stemming losses using massive price cuts) since it is the new mainstream volume processor. Not only that, but it also needs to command a price to cover the loss of low end discrete GPU sales that are no longer required.

The problem though is that Llano will be competing against Sandy Bridge chips now, and Ivy Bridge chips in several months, and GPU performance is not the #1 selling point for systems to the general public. Unfortunately for AMD, Llano, like Bulldozer will face low price ceilings due to CPU performance. Llano's die size is close to 2x bigger than dual core SB chips it will be competing against. Not exactly a recipe for success: high manufacturing cost, low price it can sell for.

Sandy bridge is the faster cpu, llano will be the more capable chip though for a certain segment of people who want to ever play a recent game. The latter group will lose out on what by sacrificing cpu performance? lower cpu benchmark scores? After a certain performance level, I am not sure the difference will be that noticeable for people.


For people who never intend to play a graphically intensive game on their notebook, then sandy bridge is the way to go. For the smaller segment that does intend to do so? Especially at a certain lower price segment? llano seems to be the best option as the baseline graphics performance is so much better, and THAT is the bottleneck for those people, not the cpu.


intel makes the fastest performing x86 chips on the market, they have a process technology lead that cannot be closed any time soon, they focus much more of their die size on the x86 portion of their chips as well. They are not going to be caught by AMD anytime soon. What is amd going to do? Go home?

They seem to know they are not going to go toe to toe anytime soon on the x86 front, so they shifted focus to a segment that was an actual performance bottleneck for many people, graphics performance. Something more and more software takes advantage of, and something that seems to have a much higher threshold to travel before diminishing returns are reached.

how fast of a chip do you need to run excel? office?

is the i7 vs the i3 going to make THAT big of a functional difference?


diminishing returns




Now a much more potent graphics card is going to make a huge difference when playing mass effect on your laptop, its not a useless strategy for them to play to their strengths.

I don't know how the llano chips will do in the market, but I am sure about one thing, they will have the more rounded performance offerings from the major chip makers.


And it may be that most people don't give a crap about being a jack of all trades vs a master of the x86 game, but the truth is many people would have better overall usability if they went that route for what they do @ certain price levels.

 

[randomaeiou]

The tone of the op is almost regret that amd has a more compelling product in the low end.

Oh no! intel is not way ahead of amd at EVERY segment of the market anymore! que x/y/z contingency for the atoms STAT!


...

Let them have a brief moment in the sunlight, intel will do their thing on their own.

To be perfectly honest, I couldn't care less if my chip was fabbed in Timbuktu by VIA, if it got me the performance that I desired. Unfortunately, Intel has, for the largest part of PC history, been the leader in chip technology, hence my observations/suggestions.

The reason I got into this whole idea in the first place is because of my own personal needs, and seeking some sort of product that would answer it. Allow me to explain:

Reducing number of electronic devices - phone + laptop is more than enough thank you. Phone + laptop + home desktop + work desktop/laptop + mp3 player + GPS + tablet + whatever else commonly used = WAY too much luggage. Thank goodness a decent smartphone nowadays can be a decent GPS and a good mp3 player.

However, laptop+desktop - I hate transferring data, and having to duplicate programmes, and maintaining two systems, both hardware and software. Throw in a tablet, which straddles the laptop-smartphone divide, and it gets even messier trying to consolidate data. I know there are lots of sync programs and cloud based computing, but sometimes secure data just can't be trusted to the cloud, and wireless syncing can be a pain on battery life and slow transfer speeds and getting the machines to talk to each other in the first place, not to mention either having to manually do it, or jump through extreme hoops getting it to work reliably in an automated fashion, and wired syncing, well, you end up with black spaghetti/medusa in the briefcase/bag.

I'm by no means a power user in terms of my computing demands - I'm currently running an ASUS UL50vt based upon a SU7300(oc@1.9GHz)+GT210m(mild oc) with a hybrid 500GB drive and a 4GB ramdisk on 8GB total ram, and it satisfies my computing needs. I game lightly, and generally on older titles (not Crysis!), hence no need for HD 9990/ GTX995 class graphics. I work a usual working day, getting to work ~8am and leaving ~6pm, with less than an hour public transport commute either end of it, and the UL50vt can survive that long on average use (surfing/email, productivity suite, proprietary x86 programmes that aren't particularly demanding), with media consumption during the commute on each end, with a bit of power savvy management.

That said, when I'm on the go, a 15 inch laptop is probably not the most portable. In fact, I've had the opportunity to play with a MBA 11", and still didn't like having to open up the keyboard - loved the iPad for that. Unfortunately, iPad is ARM/iOS based - I frequently walk alongside my bosses from point A to point B having to edit/show office documents, pdf stuff, proprietary x86 software the company uses, etc.

So, in summary, I'm looking for: a tablet form factor 10-12inch x86 computer, with low-mid level CPU ability, half decent (~6310/HD2000/HD3000 class) GPU (integrated or not, I couldn't care less, as long as battery life is up to par), and ~6-8h battery life which includes ~2h media consumption, with wireless and bluetooth ON all this time - bluetooth for a keyboard, mouse and headphones, as necessary, and HDMI out for linking to my large screen at home. Ideally sub $800-$900.

Based on current/imminent technology, such demands would be met by:

a) 32nm 2C Atom with a few EUs (HD class graphics)
b) 2C AMD Wichita (assuming TDP in the 5W ballpark, and around 1.5GHz clockspeed)
c) single core ULV SB with HD2000/HD3000

all of the above ideally with a PixelQi screen for further power savings/improved battery life, particularly since during my normal working hours, most work done on the computer will be fairly static and does not need a high refresh rate, but gaming/media out-of-hours does.

Currently available tablets do NOT satisfy this criteria to become my one and only computer, because:

a) Atoms have crap graphics, and tablet Atoms do not ship with ION/ION2.
b) AMD Ontario is too slow CPU wise (I have fairly low requirements, but even then the Ontario just doesn't cut the mustard as a primary computer).
c) AMD Zacate tablets have yet to make their appearance, and are probably too power hungry to maintain decent battery life.
c) i5-470UM in ASUS EP121 has crap graphics, no battery life to speak of, and out of budget, making me pay TOO MUCH for computational power I WILL NOT NEED/USE.

Now, I'd like to think I'm not so special/demanding in my needs - battery life to fit around a workday+commute, x86 support for commercial/proprietary software and full featured office suite work, basic wireless connectivity, and a half decent GPU for an enjoyable *light* gaming experience (<25fps@720p does NOT count as enjoyable ).

Open to suggestions...

 

[YeuEmMaiMai]

IMHO Atom has a long way to go to catch up to AMD. just the fact that AMD can run X64 is enough reason to choose it over Atom any day of the week. Not to mention the massive memory space advantage

the ONLY thing I would ask amd to do is make the platform dual channel as to make full use of the intergrated cpu/graphics core.

 

[pxc]

how fast of a chip do you need to run excel? office

You make a very compelling argument why a company using an ancient core + generally poor product execution and third best performance (Intel is basically topping its own last generation with SB and will do the same with IB) should be able to do well. GLWT, Llano.

Bobcat CPUs OTOH are fine in the segments where it is being sold. Unfortunately it is not growing a market like Atom did and is simply cannibalizing AMD's own sales.