AMD Ontario APU pictured,die size ~77mm^2

[qcmadness]

You do realize that for Atom the D stands for desktop replacement ? You don't find D parts in mobile, they are for always powered embeded form factors, controllers, telecom,etc.
The Ds are basically the failed Atoms, with the highest leakage and power consumption. Not wanting to throw them away, Intel sells those cheaply in always plugged in systems.

The mobile netbook parts are the Ns, while tablets and MIDs go for Zxx series.
The new Z6xx series target the smartphones, Intel announced they have 2 major wins in this field ( Motorola, Nokia ? ).

Not of Bobcat / Ontario / Zacate target market here.
Irrelevant with topic.

[savantu]

Not of Bobcat / Ontario / Zacate target market here.
Irrelevant with topic.

I don't know whether to yawn or ,

So you make this comparison

AMD current gen mobile platform:
x2 L625 (18W), 780G + SB710 (11.5W+4.5W)

AMD next gen mobile platform:
Ontario (9W), Hudson M1 / D1 (~6W)
Zacate (18W), Hudson M1 / D1 (~6W)

Intel Atom mobile platform:
Atom D51x - Pineview (13W), NM10 (2W)

Intel Core ULV platform:
i3-330UM - Clarsdale (18W), Ibex Peak (3.5W)

Than draw this conclusion :

I don't think Atom is competitive at all.

I tell you the D series aren't part of the mobile platform

You do realize that for Atom the D stands for desktop replacement ? You don't find D parts in mobile, they are for always powered embeded form factors, controllers, telecom,etc. ....The mobile netbook parts are the Ns, while tablets and MIDs go for Zxx series.

Let me refresh your memory :
N550 - 8.5W, 1.50 GHz, dual core, GPU
N475 - 6.5W, 1.83 GHz, single core, GPU
N450 - 5.5W, 1.66 GHz, single core, GPU

And you reply

Not of Bobcat / Ontario / Zacate target market here.
Irrelevant with topic.

Do you need any further assistance ?

[Florinmocanu]

Savantu, is he Paul Demone? Because if they are the same person, much would be explained.

I don't know whether to yawn or ,

So you make this comparison


Than draw this conclusion :



I tell you the D series aren't part of the mobile platform



Let me refresh your memory :
N550 - 8.5W, 1.50 GHz, dual core, GPU
N475 - 6.5W, 1.83 GHz, single core, GPU
N450 - 5.5W, 1.66 GHz, single core, GPU

And you reply


Do you need any further assistance ?

But the GPU is not on Die, it's quite important to say that.

Ontario's TDP is CPU + GPU, while Atom is only CPU. So while Ontario is 9W, its CPU + GPU = 9W. While Atom is CPU only.

Let me know the combined TDP of ATOM N550 + integrated GPU from intel and it's level of perf/wat, i cannot find such an info yet.

[-Boris-]

Let me refresh your memory :
N550 - 8.5W, 1.50 GHz, dual core, GPU
N475 - 6.5W, 1.83 GHz, single core, GPU
N450 - 5.5W, 1.66 GHz, single core, GPU

And you reply


Do you need any further assistance ?

And they shall compete with this one.
T40N - 9W, 1.0 GHz, dual core, GPU, LVDDR3

And the only thing we now about the power draw of T40N is that it's theoretical maximum power draw is somewhere between 5W and 9W.

Saying that it's typical power draw is at 9W is BS, and maximum power draw at 9W is absolute worst case scenario.

[savantu]

But the GPU is not on Die, it's quite important to say that.

Ontario's TDP is CPU + GPU, while Atom is only CPU. So while Ontario is 9W, its CPU + GPU = 9W. While Atom is CPU only.

Let me know the combined TDP of ATOM N550 + integrated GPU from intel and it's level of perf/wat, i cannot find such an info yet.

Reading comprehension problems ?

N550 - 8.5W, 1.50 GHz, dual core, GPU
N475 - 6.5W, 1.83 GHz, single core, GPU
N450 - 5.5W, 1.66 GHz, single core, GPU


Let me repeat : you're stuck in 2008. We are in 2010. Today's Atom that your find all over the place ( cast a look emag ) integrates on the same die, the CPU+GPU+NB. Clear ?

So total power for the T40N would be 9+6w=15w ( SB ) while for the N550 it would be 8.5+2=10.5w

30% less power at platform level which with the same battery and screen would translate into 30% more endurance. Using the other N models only increases this advantage.

Current Atom : everything on a single die.

And they shall compete with this one.
T40N - 9W, 1.0 GHz, dual core, GPU, LVDDR3

And the only thing we now about the power draw of T40N is that it's theoretical maximum power draw is somewhere between 5W and 9W.

Saying that it's typical power draw is at 9W is BS, and maximum power draw at 9W is absolute worst case scenario.

You do understand that what you're saying equally applies to Atom chips also ?

[Solus Corvus]

IMO, when the players will overlap, AMD could find itself as a collateral victim. Intel has the resources and is willing ( already its marketing machine turned on ARM ) to steam roll ARM on its turf. I wouldn't be surprised if Intel allocated to Atom and other small factor SoC an order of magnitude more resources than ARM has.
Yet, ARM has an aces on its sleeve : it is a pure design house which licenses its designs and other carry the burden to implement them. ARM gets royalties of a few cents for each chip. That's extremely low, but when your designs are produced at a rate of 5m a day it adds up.
Basically ARM brings together Qualcom, Freescale, Samsung,etc. Intel isn't in battle only with ARM, but with those companies also.

Why could AMD end up as a collateral victim ? ARM is so widespread because it is cheap. ARM based CPUs sells for cents and rarely above a few $. It is safe to assume its trust in low power and low end x86 territory will be price driven ( they don't have the SW base ). Who is the low cost king of x86 ? AMD.
Intel doesn't go after every unit in the market for a very simple reason, its margins would crater from the lovely 60%+ it enjoys now. Every time AMD took a beating from Intel it retreated in the low cost part of the market where Intel wasn't willing to enter.
Well, that is about to change since ARM targets specifically that market. ARM won't bother to design the next Nehalem, but it will produce an ARM based Ontario-like CPU ( AR15 ? ).

I find your analysis to be flawed. For one thing, exactly the opposite of what you describe is actually happening. Intel is competing with ARM for the low margin parts and AMD is filling the gap in higher margin parts.

ARM might eventually make inroads into the laptop market, but it's already a force in the low power market. Intel might have a lot of disposable income but throwing a bunch of money at it won't necessarily make x86 better than ARM for low-power. I think AMD is smart not to try to enter that segment yet. Their little ontario would get strangled in the crib by ARM. AMD might have to compete with future products from ARM but Intel has to compete with current products from ARM.

[Manicdan]

Intel and AMD are looking for different markets right now. AMD needed a good mobile part for netbooks and above, Intel needs something good below netbooks ( mids, tablets and ultimately smartphones ).

ARM is picking up steam and aiming to go from smartphones to mids/tablets/laptops and ultimately high density servers.

A clash is on the horizon as Intel thrusts in ARM territory and viceversa. AMD at the moment is sitting on the sidelines, but they need to be prepared for this war. Current Ontario landed just above where ARM is trying to go in the next step.

IMO, when the players will overlap, AMD could find itself as a collateral victim. Intel has the resources and is willing ( already its marketing machine turned on ARM ) to steam roll ARM on its turf. I wouldn't be surprised if Intel allocated to Atom and other small factor SoC an order of magnitude more resources than ARM has.
Yet, ARM has an aces on its sleeve : it is a pure design house which licenses its designs and other carry the burden to implement them. ARM gets royalties of a few cents for each chip. That's extremely low, but when your designs are produced at a rate of 5m a day it adds up.
Basically ARM brings together Qualcom, Freescale, Samsung,etc. Intel isn't in battle only with ARM, but with those companies also.

Why could AMD end up as a collateral victim ? ARM is so widespread because it is cheap. ARM based CPUs sells for cents and rarely above a few $. It is safe to assume its trust in low power and low end x86 territory will be price driven ( they don't have the SW base ). Who is the low cost king of x86 ? AMD.
Intel doesn't go after every unit in the market for a very simple reason, its margins would crater from the lovely 60%+ it enjoys now. Every time AMD took a beating from Intel it retreated in the low cost part of the market where Intel wasn't willing to enter.
Well, that is about to change since ARM targets specifically that market. ARM won't bother to design the next Nehalem, but it will produce an ARM based Ontario-like CPU ( AR15 ? ).

so much of this i do not agree with
first intel is moving down into a market to fight with ARM that ARM is moving out of.

think for a second, in the 10W market, ontario is going to do great.
in the 2-5W market, atom is kinda by itself
in the 1W market and lower ARM is doing wonders

now how long until ARM moves up to 10W and starts competing with AMD? and what will it take to get there? were not talking about an ipad device thats just a big screen with a weak chip, were talking about full computers that need to be able to do everything.

it seems that in this case, intel gave up on the 10W market cause atom is a pretty weak chip, and decided to do everything to make it use less power and get into cell phones and hand helds. but whats it going to do when it gets there? a good cell phone already has gpus to help out the cpu, is x86 going to fix that aswell? are we going to see an atom based phone with like 5 LRB cores just so it can compete with the next waves of phones that will have graphics better than a PSP or DS.

my opinion is that atom by itself is too weak to do anything worth while, and it will always need the help of something else, primarily a gpu, to let it even do any damage. and i think we all know that intel is not really into the whole gpu thing.

also about the whole market vs margins thing, when i7 first came out, it was less than 1% of the sales. they worked so hard to make their 500$ chips, that almost no one wanted. this is XS and many of us do like the newest stuff, but average joe consumers rarely see any point in buying something 20% faster for 300% the price. there is nothing wrong with making stuff for mainstream

[-Boris-]

You do understand that what you're saying equally applies to Atom chips also ?

You can't compare them since intel has more steps, N550 has a power consumption between 6.5W and 8.5W. T40N has a maximum power consumption between 5W and 9W. I think that's enough to say that you don't have enough to say anything about how one compares to the other. Or do you know something I don't?

[savantu]

I find your analysis to be flawed. For one thing, exactly the opposite of what you describe is actually happening. Intel is competing with ARM for the low margin parts and AMD is filling the gap in higher margin parts.

Allow me to disagree. Here are my reasons :

1. Intel isn't competing for the low margin products. For example, Atom has higher gross margins than Intel's Celeron,single and double core Pentiums. Why ? Because you get 2000 from a wafer and you sell them for $40 in average. With Celeron/Pentium you get 400 and sell them for $60.
Same story is happening at MID/tablet/smartphone level. Intel isn't competing for $15 a pop cellphones, but for those which cost over $200 where they can enjoy high margins. They don't want to get in the price slime that ARM loves.
2. AMD isn't filling any high margin gap. I'm sure they'll price their parts at Atom like level or even lower. Where Ontario is competing you have current single and dual core Atom, superseded at the top by ULV I3/5s,than LV I3/I5s which are going to be replaced by SB very soon. I've heard that Intel has a backup solution a single core SB. Probably it will end up in the 10-20w range and things will be really intersting : 2 small Bobcat cores + large GPU vs. large SB core + medium GPU.
3. ARM targets the netbooks, a market currently worth 30-40m units. That's where Bobcat also tries to reach. Intel is attacked from both sides, true, but hey, if they can't handle the pressure who can ?

ARM might eventually make inroads into the laptop market, but it's already a force in the low power market. Intel might have a lot of disposable income but throwing a bunch of money at it won't necessarily make x86 better than ARM for low-power. I think AMD is smart not to try to enter that segment yet. Their little ontario would get strangled in the crib by ARM. AMD might have to compete with future products from ARM but Intel has to compete with current products from ARM.

The growth is in low power market. That's why Intel is going after it. The traditional PC market is yesterday news.
Intel covers the very top of the low power market with netbooks and Atom. There tablets below, MIDs and smartphone where Intel is a non-player. Their attempt to increase revenue by going into new markets is very logical.

AMD can sit on the sidelines and be confined to the x86 market or try to do what Intel does and go for ARM's throat. It is their choice after all.

[savantu]

You can't compare them since intel has more steps, N550 has a power consumption between 6.5W and 8.5W. T40N has a maximum power consumption between 5W and 9W. I think that's enough to say that you don't have enough to say anything about how one compares to the other. Or do you know something I don't?

What the is that ? From where do you pull this ranges ?
Do you understand what TDP is ?

If a chip has a TDP of 10w it means, that's the maximal power it will dissipate running power intensive commercial SW. The cooling solution needs to be able to remove 10w 24/7. The CPU itself can exceed its TDP in certain cases ( thermal virus for example ). Another metric is idle power, when the whole chip or parts of it are in deep power states.

So if T40N has a higher TDP than N550 it means at similar load levels it will burn more power. It should be fairly obvious.

[Manicdan]

What the is that ? From where do you pull this ranges ?
Do you understand what TDP is ?

If a chip has a TDP of 10w it means, that's the maximal power it will dissipate running power intensive commercial SW. The cooling solution needs to be able to remove 10w 24/7. The CPU itself can exceed its TDP in certain cases ( thermal virus for example ). Another metric is idle power, when the whole chip or parts of it are in deep power states.

So if T40N has a higher TDP than N550 it means at similar load levels it will burn more power. It should be fairly obvious.

you completely missed his point, ill wait for him to set you strait though

[ajaidev]

Humm i have no idea why everyone is comparing Ontario with Atom ?

Instead of Atom comparison should be made between Ontario and Arrandale based Celeron/Pentium "Coming soon" and Core i3-330UM, given that all these are suppose to have 18TDP.

[marten_larsson]

What the is that ? From where do you pull this ranges ?
Do you understand what TDP is ?

If a chip has a TDP of 10w it means, that's the maximal power it will dissipate running power intensive commercial SW. The cooling solution needs to be able to remove 10w 24/7. The CPU itself can exceed its TDP in certain cases ( thermal virus for example ). Another metric is idle power, when the whole chip or parts of it are in deep power states.

So if T40N has a higher TDP than N550 it means at similar load levels it will burn more power. It should be fairly obvious.

Maybe you should chill out a bit. I'm fairly certain he meant the Atom has "maximum" power of 6.5 - 8.5W while the Zacate has 5 - 9W - you won't get closer than that in an average situation on either solution. So while the Zacate TDP is higher, the actual "maximum" power could be lower (maximum being fully loaded 24/7 in a normal app) than Atom - otherwise the Atom part would have been binned <6.5W.

Also the fact that Zacate has the same die size (actually smaller if 77mm² is true) as dual core Pineview Atom should give AMD the same high margins (I'm don't know if the wafer size is similiar though) but you fail to mention this.

[-Boris-]

What the is that ? From where do you pull this ranges ?
Do you understand what TDP is ?

If a chip has a TDP of 10w it means, that's the maximal power it will dissipate running power intensive commercial SW. The cooling solution needs to be able to remove 10w 24/7. The CPU itself can exceed its TDP in certain cases ( thermal virus for example ). Another metric is idle power, when the whole chip or parts of it are in deep power states.

So if T40N has a higher TDP than N550 it means at similar load levels it will burn more power. It should be fairly obvious.

You don't just have TDP, you have TDP ranges, If you haven't noticed AMD has close to 100 different models rated at 65, 80, 95 or 125W TDP. A coincidence that all these chips has a TDP at these 4 exact numbers? Of course not, if a chip has a maximum power consumption of 69W, it's rated at 80W. And you know this. And for Bobcats AMD has only three TDPs to date, 5, 9, and 18W. That means that a processor rated at one of these can be anywhere between that number and the number below.

A T40N could be 1W away from being rated at 5W, and there is no way you could know if it is. Just as well as it could be at 8.9W.

That's why you can't do your comparisons and expect to get a useful conclusion.

[RaV[666]]

Humm i have no idea why everyone is comparing Ontario with Atom ?

Instead of Atom comparison should be made between Ontario and Arrandale based Celeron/Pentium "Coming soon" and Core i3-330UM, given that all these are suppose to have 18TDP.

Ontario is lower power version, it competes with high power atom.
Zacate is the 18W version, and it will compete somewhat with "i" intel series.

[savantu]

You don't just have TDP, you have TDP ranges, If you haven't noticed AMD has close to 100 different models rated at 65, 80, 95 or 125W TDP. A coincidence that all these chips has a TDP at these 4 exact numbers? Of course not, if a chip has a maximum power consumption of 69W, it's rated at 80W. And you know this. And for Bobcats AMD has only three TDPs to date, 5, 9, and 18W. That means that a processor rated at one of these can be anywhere between that number and the number below.

A T40N could be 1W away from being rated at 5W, and there is no way you could know if it is. Just as well as it could be at 8.9W.

That's why you can't do your comparisons and expect to get a useful conclusion.

Parts distribution based on power is what drives the TDP. If 95% of the parts would be under 7w then they would set that as TDP. The fact that they choose 9w it means most of the parts are closer to 9 than to any other arbitrary figure you've chosen ( the next range at 5w is for a single core no GPU device , hardly relevant ).

[Manicdan]

Parts distribution based on power is what drives the TDP. If 95% of the parts would be under 7w then they would set that as TDP. The fact that they choose 9w it means most of the parts are closer to 9 than to any other arbitrary figure you've chosen ( the next range at 5w is for a single core no GPU device , hardly relevant ).

and i think your wrong

they are distributed to the highest in the range where many fall

if the tpd values are as follows per chip:
5
7
8
9
13
14
17
18

the 3 best numbers are 5, 9 and 18,

if their goal was to make things more attractive then it would be like
5, 8, 9, 14, 18. way to many extra things thrown in for a value that really only affects the size of the heatsink and power brick. why have 5 values when the difference is that all ontario netbooks can accept anything 9 and under. imagine looking at one where they choose the 8w tdp to build everything around, all that extra time wasted to build a platform that is almost no different. its a wasted value and serves no real purpose.

its not too hard to keep it simple and be open minded is it? no need to assume a million other things just because they didnt provide as many tdp ranges as intel

[-Boris-]

Parts distribution based on power is what drives the TDP. If 95% of the parts would be under 7w then they would set that as TDP. The fact that they choose 9w it means most of the parts are closer to 9 than to any other arbitrary figure you've chosen ( the next range at 5w is for a single core no GPU device , hardly relevant ).

The first two sentences is just plain irrelevant, you don't have to tell me those things. And they don't choose 9W because it's closest only, they choose 9W because it's the closest upwards.

So if AMD have a new bobcat based SOC that isn't listed above, with all parts between 10W and 11.5W. What TDP do you think they would give that product?

[savantu]

The first two sentences is just plain irrelevant, you don't have to tell me those things. And they don't choose 9W because it's closest only, they choose 9W because it's the closest upwards.

So if AMD have a new bobcat based SOC that isn't listed above, with all parts between 10W and 11.5W. What TDP do you think they would give that product?

11.5w

Anyway, your assumption implies the process to be rubbish with power variation of more than 30% between the parts and distribution doesn't follow any pattern. I do not believe that.

[qcmadness]

11.5w

Anyway, your assumption implies the process to be rubbish with power variation of more than 30% between the parts and distribution doesn't follow any pattern. I do not believe that.

You don't have to believe that.

Most Intel Q9xx0 quad-core CPUs can fit into 65W power envelope, yet they are marked as 95W TDP CPUs.

[-Boris-]

11.5w

Anyway, your assumption implies the process to be rubbish with power variation of more than 30% between the parts and distribution doesn't follow any pattern. I do not believe that.

So by stating that they would sell this one marked as a 11.5W TDP chip you deny their use of TDP ranges?

My assumption was purely fictional and I didn't even care what it would imply on the process. But if want to bring it up it's 15% and it's completely normal. It's nothing unusual with processors sold at slight different voltages, where some has 1.2V others of the same model can have 1.3V. And that is much more than 15% power difference.

[savantu]

So by stating that they would sell this one marked as a 11.5W TDP chip you deny their use of TDP ranges?

My assumption was purely fictional and I didn't even care what it would imply on the process. But if want to bring it up it's 15% and it's completely normal. It's nothing unusual with processors sold at slight different voltages, where some has 1.2V others of the same model can have 1.3V. And that is much more than 15% power difference.

I believe we are in a stalemate with the discussion and only reviews done this winter will show how much headroom there is. We'll revisit it then.

[-Boris-]

I believe we are in a stalemate with the discussion and only reviews done this winter will show how much headroom there is. We'll revisit it then.

You couldn't avoid my point forever.
But that is exactly what I've said, you can't do any conclusions based on the numbers we have now.

[Solus Corvus]

Allow me to disagree. Here are my reasons :

1. Intel isn't competing for the low margin products. For example, Atom has higher gross margins than Intel's Celeron,single and double core Pentiums. Why ? Because you get 2000 from a wafer and you sell them for $40 in average. With Celeron/Pentium you get 400 and sell them for $60.

I'm not talking gross margins, just margins. Your example doesn't disagree with what I said. Atom in this example is still a lower margin product.

Same story is happening at MID/tablet/smartphone level. Intel isn't competing for $15 a pop cellphones, but for those which cost over $200 where they can enjoy high margins. They don't want to get in the price slime that ARM loves.

I think that ARM is likely more suitable for the >$200 cellphone/MID market than x86. Of course, time will tell. For one thing, backwards comparability on those formats isn't as important.
For the tablet/netbook level Atom is too weak for a decent experience, IME. I'm curious to see how ontario fits in those formats. They are just a little too small for CULV.

2. AMD isn't filling any high margin gap. I'm sure they'll price their parts at Atom like level or even lower. Where Ontario is competing you have current single and dual core Atom, superseded at the top by ULV I3/5s,than LV I3/I5s which are going to be replaced by SB very soon. I've heard that Intel has a backup solution a single core SB. Probably it will end up in the 10-20w range and things will be really intersting : 2 small Bobcat cores + large GPU vs. large SB core + medium GPU.

AMD won't get high margins for a few reasons. For one, they have to outsource their chip production. They also can't price as high because they don't have the brand recognition. We will have to see what the margins end up like. Probably not great, but for them some revenue is better then none.

I think Ontario will fit in the gap between Atom and I3/I5/SB. It should be significantly more powerful then Atom. That's good for the formats, like netbooks and tablets, where Atom is weak now. It will overlap with I3/I5/SB at the Zecate end of the range. My bet is that I3/I5/SB will be quite a bit faster, but Zecate will have significantly lower real world power usage. GPU will probably be about the same. So it would depend on your needs as to which one is better.

3. ARM targets the netbooks, a market currently worth 30-40m units. That's where Bobcat also tries to reach. Intel is attacked from both sides, true, but hey, if they can't handle the pressure who can ?

ARM could do some damage in the netbook/tablet sectors if they designed a slightly beefier chip. I hope they do. Though for some people in this range software compatibility is still important.

Of course Intel can handle the pressure. AMD beat on them for years and Intel easily recovered. The real question is if x86 is ever going to be competitive for MIDs and smartphones. Because when Intel demonstrated the capability to take a beating and hardly notice they also demonstrated the capability to stubbornly go down a futile path. Personally I feel that ARM (or other low-power specific architectures) will always have a major advantage in this segment and what Intel has to fight with is process.

The growth is in low power market. That's why Intel is going after it. The traditional PC market is yesterday news.
Intel covers the very top of the low power market with netbooks and Atom. There tablets below, MIDs and smartphone where Intel is a non-player. Their attempt to increase revenue by going into new markets is very logical.

The whole market of nettops/netbooks/tablets/mids/smartphones etc is growing. Traditional desktops are in decline. What could be more logical then for AMD to go after a segment of that market that neither ARM or Intel are filling effectively? At least as a first step.

AMD can sit on the sidelines and be confined to the x86 market or try to do what Intel does and go for ARM's throat. It is their choice after all.

You make it sound like those are the only two options. Personally I'd like to see Intel and AMD both release low-power non-x86 architectures to more effectively compete with ARM in these segments. I think this is where things are going to heat up in the next few years.

[savantu]

I'm not talking gross margins, just margins. Your example doesn't disagree with what I said. Atom in this example is still a lower margin product.

True. But it is incremental to current business and it is an opportunity that needs to be pursued.

For the tablet/netbook level Atom is too weak for a decent experience, IME. I'm curious to see how ontario fits in those formats. They are just a little too small for CULV.

If Atom is weak, what about current ARM designs like Ipad and Iphone ? Those CPUs are much slower than Atom. Maybe the whole point is not performance, but power ? Atom has still too much performance per clock to be able to go down to a few hundred milliwats of power. The uarch needs to be simplified or moved aggressively to new processes. Performance is the last issue.

You make it sound like those are the only two options. Personally I'd like to see Intel and AMD both release low-power non-x86 architectures to more effectively compete with ARM in these segments. I think this is where things are going to heat up in the next few years.

Why bother and reinvent the wheel when you already have x86 ? The whole problem is getting it into the right envelope. And from then on, for Intel at least, its process advantage will leave ARM&foundries in the dust.