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

[informal]

PhysX.

And the only reason NV uses x87 is because SSE based one would probably trounce GPU Physx

[Chumbucket843]

Are there a lot of real world modern consumer software that still use x87 code? I can't think of any off the top of my head, but maybe i've missed a niche that you could pigeon hole my argument into.

yes, depending on your definition of modern. i'll assume that it means the program is still in use. i could disassemble a shatload of binaries to show you the "f instructions" (x87 instructions begin w/ the letter f.) which imo are as bad as the f word, but the x87 is there none the less.

furthermore x87 does not make anything synthetic. it makes it legacy software or at the least implies it. do i think superpi is a good bench? no. but is it synthetic? no. i'm not really into extreme OCing due to "budgetary constraints" but superpi is a de facto benchmark that the OCers like and there is nothing wrong with it for what they do.

[nn_step]

yes, depending on your definition of modern. i'll assume that it means the program is still in use. i could disassemble a shatload of binaries to show you the "f instructions" (x87 instructions begin w/ the letter f.) which imo are as bad as the f word, but the x87 is there none the less.

furthermore x87 does not make anything synthetic. it makes it legacy software or at the least implies it. do i think superpi is a good bench? no. but is it synthetic? no. i'm not really into extreme OCing due to "budgetary constraints" but superpi is a de facto benchmark that the OCers like and there is nothing wrong with it for what they do.

Let us assume for a second, that the term modern means using an optimizing compiler of equal or greater sophistication than Microsoft's C compiler and/or GCC.

Which in short means does not generate x87 instructions except in cases where data types are defined at 80bit floating point or when hand coded assembly demands use of x87 operations.

Legacy support is cheap in instruction sets and there is absolutely no requirement or need to optimize performance for less important legacy instructions.

[Nintendork]

Some bobcat p@rn:

City of heroes ran smoothly, and there were no glitches or hiccups in the demo we saw. This should change the way integrated graphics are thought of.

http://www.semiaccurate.com/2010/09/...obcat-systems/

[qcmadness]

Some bobcat p@rn:



http://www.semiaccurate.com/2010/09/...obcat-systems/

still requires a fan

i want a fanless bobcat itx board as my new home server

[Nintendork]

It's Zacate, the "High-End" platform.

Other thread videos:
http://www.youtube.com/watch?v=NZE2SuJlJCw
http://www.youtube.com/watch?v=-lLNHTu4oLk

[nn_step]

still requires a fan

i want a fanless bobcat itx board as my new home server

prototype

Doesn't imply that retail version will require a fan.

[Calmatory]

still requires a fan

i want a fanless bobcat itx board as my new home server

Just remove the fan? I ran an overclocked Celeron 333(@416 MHz) in a small ATX case which only had PSU fan. Sure it ran hot but it was rock solid for over 2 years with over 500 days of uptime.

[savantu]

Where it is? on the die of course!

Well, the AMD engineers who validated the 5w 800MHz part as the lowest power part did not know that. Write them an email about this.

Leakage itself in such a part is higher than 1w and you know it.

The sub 1W capable core is on the 10 times larger SOC die of course.
Do you really think AMD is lying in all these presentations?

Not really. They said " sub 1w capable". That doesn't commit to anything. It could be 3rd generation on 28nm of the uarch that reaches sub 1w cpu power.

Do you realize that driving the memory bus to a SODIMM or other
external memory may already take several Watt? That's why smart
phones have the DRAM in the package.

Have you done your homework and read all the fine print of how
some ultra low TDP's are measured and defined. Do you understand
that you are comparing with an SKU which is NOT binned for ultra
low power consumption at all but is just the cheapest bin for small
embedded applications. The 5W part is defined with (cheaper)
1.5V memory I/O devices and not even low power 1.35V LPDDR or
a more appropriate mobile POP setup.

And that low power DDR interface will make it shave half the TDP. What are you smoking ?

You found this on an investor board as always didn't you? From
"somebody" who claims that an Atom core is an order of magnitude
more power efficient as a Bobcat core, yes? and you do believe this.....

Actually not ( I can't pin point anything I've read with what you're saying ) . It seems you're reading Intel's invest boards more than me.

[RaV[666]]

Cooler looks like from a pentium 166mmx days ;-) .Im sure you can cool it passively with something just a tad bigger.

[Manicdan]

still requires a fan

i want a fanless bobcat itx board as my new home server

what your looking at is a laptop setup with an ALUMINUM heatsink with fins so thick you can count them in that horrible scaled image. i assure you that they can make them fanless for anything with an inch of room. and yes i bet miniITX can easily support fanless. but a netbook i would recommend a small fan that you never hear

[-Sweeper_]

llano-based systems available in summer-2011?

I guess terrace wasnt so wrong after all

[LightSpeed]

savantu, i really want a reply to my post. dont ignore it

[SimBy]

Zacate in action.

http://www.youtube.com/watch?v=XH5A4D9qoDQ

[Lightman]

Just once thing to communicate.

@savantu

I know which slide you're talking about and making big fuss.
AMD claimed sub 1W capable for Bobcat CORE (single processing core without IMC and the rest). You're simply stretching that a bit too much and implying they meant APU which is simply not true.
So leakage of single Bobcat core is a lot lower than 1W which you can't comprehend. Leakage of whole Ontario or Zacate APU is higher than 1W which we all know/can accept.

PS. Please show me slide in which AMD claimed they are targeting Smartphone market with this core.

[Hans de Vries]

Well, the AMD engineers who validated the 5w 800MHz part as the lowest power part did not know that. Write them an email about this.

Leakage itself in such a part is higher than 1w and you know it.



Not really. They said " sub 1w capable". That doesn't commit to anything. It could be 3rd generation on 28nm of the uarch that reaches sub 1w cpu power.

No, they are saying that Bobcat (40nm) has sub 1W capable cores.
They specifically say that the entire Ontario System-On-Chip is designed
for the netbook segment and the value/mainstream notebook segment
and that it wasn't designed with tablets in mind and certainly not for
smart phones.

And that low power DDR interface will make it shave half the TDP. What are you smoking ?

No, the difference between driving 64bit GDDR3 and GDDR5 (P2P) is 2W TDP
for example. You are comparing a retired device which didn't even had a
memory interface but a 400MHz FSB instead against a 64bit DDR3 bus at
1033MHz which needs to be able to drive multiple SODIMMS.

[MrMojoZ]

llano-based systems available in summer-2011?

I guess terrace wasnt so wrong after all

Too bad he had act like a child and get punished, eh?

[savantu]

You want mine? 70%. Im not here to see who was right or wrong in 4 months, might be you or me or neither of us, we'll know that in time. However, let me paste the case you pointed out again:



So over here, the TDP is the same, performance is the same (by your predictions), and Bobcat has a much faster GPU. Clean win I would say, atleast in this case?
So is AMD superior to Intel in this occasion ? Or are you having a hard time admitting it? Or are you gonna say we cant predict the performance of Bobcat's GPU, the Atom's might be superior? You also completely ignored by next paragraph, let me paste it again for you:

Clear win indeed. The higher you go in power the bigger the gap between Bobcat and Atom. Conversely, the lower you go, the opposite happens, until Bobcat can no longer compete ( sub 5w ).

Just once thing to communicate.

@savantu

I know which slide you're talking about and making big fuss.
AMD claimed sub 1W capable for Bobcat CORE (single processing core without IMC and the rest). You're simply stretching that a bit too much and implying they meant APU which is simply not true.
So leakage of single Bobcat core is a lot lower than 1W which you can't comprehend. Leakage of whole Ontario or Zacate APU is higher than 1W which we all know/can accept.

So now we're moving from CPU power to core power. What's next ? They'll put forward FPU and ALU power ?
If an APU has disabled everything else except one core and the IMC, by using power gating technology, you should not have a difference compared to the single core.

PS. Please show me slide in which AMD claimed they are targeting Smartphone market with this core.

Well, none. Because they know they can't. Which also means they know it is not "sub 1 w capable" , at least in its current generation.

No, they are saying that Bobcat (40nm) has sub 1W capable cores.
They specifically say that the entire Ontario System-On-Chip is designed
for the netbook segment and the value/mainstream notebook segment
and that it wasn't designed with tablets in mind and certainly not for
smart phones.

And you're telling me the core burns sub 1w and the IMC is over 2w ?

No, the difference between driving 64bit GDDR3 and GDDR5 (P2P) is 2W TDP
for example. You are comparing a retired device which didn't even had a
memory interface but a 400MHz FSB instead against a 64bit DDR3 bus at
1033MHz which needs to be able to drive multiple SODIMMS.

So according to you, memory I/O is the main culprit for high power consumption.

[mAJORD]

Clear win indeed. The higher you go in power the bigger the gap between Bobcat and Atom. Conversely, the lower you go, the opposite happens, until Bobcat can no longer compete ( sub 5w ).



So now we're moving from CPU power to core power. What's next ? They'll put forward FPU and ALU power ?
If an APU has disabled everything else except one core and the IMC, by using power gating technology, you should not have a difference compared to the single core.


Well, none. Because they know they can't. Which also means they know it is not "sub 1 w capable" , at least in its current generation.



And you're telling me the core burns sub 1w and the IMC is over 2w ?



So according to you, memory I/O is the main culprit for high power consumption.

"the use of .5v DDR3 will incur a power adder"

If standard DDR3 will change the TDP by an amount worth mentioning (perhaps 1-2w) then yes, the DDR3 IMC likely does contribute over 2w.

I would have though that was common knowledge to you anyway.

As Hans said, a lower spec, lower power DRAM interface would be used if the core ever made its way into devices that sensitive to power draw.


By the way.

[qcmadness]

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)

I don't think Atom is competitive at all.

[Mechanical Man]

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)

I don't think Atom is competitive at all.

Those 6W chipsets are very power costly, i think AMD should target in power reduction in that segment next.

[mAJORD]

I'm not sure we know the TDP of the Hudson chipset yet. What's it consist of?

SB850 @55nm is 4w, which is quite bloated compared to what you'd want for a netbook platform (too many sata, raid, 1 billion billion USB ports, etc)

[savantu]

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)

I don't think Atom is competitive at all.

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 ? ).

[Mechanical Man]

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 ? ).

Just remember savantu that amd is not targeting that market, so there is no point comparing ontario to these, but compare them to n and d series.

There is potential for AMD to get major "win". But we can't really know until more performance data is available.

Ps. If it is true what ppl say and you are financial analyst. I see problem with your very negative view on amd and very positive view on intel. It wont let you properly analyze potential of both companies. It is true intel is big and can handle failures better, but for amd the result of major win is also bigger in stock prize than for intel. That is why i find your biased view strange.

[savantu]

Just remember savantu that amd is not targeting that market, so there is no point comparing ontario to these, but compare them to n and d series.

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 ? ).

But I'm preparing the popcorn to see this new war unfold in the next years.

Ps. If it is true what ppl say and you are financial analyst. I see problem with your very negative view on amd and very positive view on intel. It wont let you properly analyze potential of both companies. It is true intel is big and can handle failures better, but for amd the result of major win is also bigger in stock prize than for intel. That is why i find your biased view strange.

I have no financial interest in the matter, it is simply a hobby for fun.