AMD CPU speculation... and expert conjecture

[Cazalan]

And I'm telling you there is no smartphone docking solution right now that remotely replaces a desktop. Samsung's been working on several for awhile now and they just never work right. Cellphone hardware is really weak, we don't notice it because with such a limited interface there isn't much for it to do. The moment you try to go multi-task it really slows down, neither Android nor iOS is capable of preemptive multi-tasking which is a requirement for Office Automation work. Windows RT is a wildcard as the NT kernel used can technically do that, though there haven't been any solid platforms made to do that.

So it's primarily a software issue not hardware issue. No one said it's perfect. Yes it's considerably weaker than a desktop but people live with those limitation for the convenience of mobility. Desktop PCs didn't use to be preemptive multitasking either but business still ran. I had the privileged of running Windows 3.1.

Ubuntu seems to be doing a good job of adding better multitasking. It may not be 100% there yet but the mobile platform will continue to improve as the compute power goes up. The tricks for dock/undocked profiles is moving into phones/tablets.
http://www.ubuntu.com/phone/design

From what I've read Android 4.2 has preemptive multitasking.

 

[jdwii]

Also someone mentioned HP and Dell losing so much money now compared to before maybe its because their deals suck and everyone knows they suck i don't know anyone who likes Dell or HP hopping until Newegg real quick shows their deals are extremely bad, all this shows is if you keep producing unreliable crap and overcharge it people will learn. HP wants 500$ for a A4 laptop. Lol you have to spend 700+$ to get a A8 laptop from them i hope they do go under.

 

[-Fran-]

Also someone mentioned HP and Dell losing so much money now compared to before maybe its because their deals suck and everyone knows they suck i don't know anyone who likes Dell or HP hopping until Newegg real quick shows their deals are extremely bad, all this shows is if you keep producing unreliable crap and overcharge it people will learn. HP wants 500$ for a A4 laptop. Lol you have to spend 700+$ to get a A8 laptop from them i hope they do go under.

I think Dell and HP have another problem. I'd say that current gen teenagers are more aware of what a computer really is and go to, for example, Newegg + friend "who knows" or BestBuy/Fry's + assembly service. And that permates to current adult gen as well. Nowadays, every one has a friend or relative that "knows about computers", right?

Dell and HP don't cater that much now for that type of people.

Lappies are another beast though. I'd say in lappies the competition is even more fierce, since most companies have free reign on how to design theirs unlike desktop PCs (like, the only benefit for non-standarization, haha). They have to thin the profit margins (I think) to the point where they must be the best not only at product level, but at manufacturing level (costs). And that's a lot of hard work, I'd say. Boutique ones are excluded from this Logic, I think.

Cheers!

 

[JAYDEEJOHN]

They want to keep their margins, and where those margins are thin, they create larger ones elsewheres to round out their projections.
This whole thing including some HW makers are set for certain margins, and when economies change, markets change, and competition becomes viable, the top usually doesnt get hit right away, but does come soon enough.

 

[blackkstar]

Also someone mentioned HP and Dell losing so much money now compared to before maybe its because their deals suck and everyone knows they suck i don't know anyone who likes Dell or HP hopping until Newegg real quick shows their deals are extremely bad, all this shows is if you keep producing unreliable crap and overcharge it people will learn. HP wants 500$ for a A4 laptop. Lol you have to spend 700+$ to get a A8 laptop from them i hope they do go under.

I think Dell and HP have another problem. I'd say that current gen teenagers are more aware of what a computer really is and go to, for example, Newegg + friend "who knows" or BestBuy/Fry's + assembly service. And that permates to current adult gen as well. Nowadays, every one has a friend or relative that "knows about computers", right?

Dell and HP don't cater that much now for that type of people.

Lappies are another beast though. I'd say in lappies the competition is even more fierce, since most companies have free reign on how to design theirs unlike desktop PCs (like, the only benefit for non-standarization, haha). They have to thin the profit margins (I think) to the point where they must be the best not only at product level, but at manufacturing level (costs). And that's a lot of hard work, I'd say. Boutique ones are excluded from this Logic, I think.

Cheers!

Personally I can say it's changed a lot. We actually had a Dell back in the day, like when I was in middle school and not building my own. I started building in high school and now ordering from Dell or anything is simply out of the question.

I just built a wicked office computer for my parents, FX 6300, 5450, 970FXA-D3, Mushkin SSD, 1080p monitor, RAM, etc and it was about $600. Dell can't come close to that no matter what they have.

The whole "PC is DYING!" thing is more about Intel and AMD not offering products that entice builders to upgrade, and Dell/etc desktop sales plummeting are due to the fact that even if they offer something for cheaper than custom built, it's of vastly inverior quality (think motherboards so cheap and low end you can't even buy them on your own if you wanted, IE OEM only) and there's massive short cuts taken.

 

[Blandge]

So, haswell is, essentially, a 16.54 nm FinFET...hmm...my how they resort to fallacy to maintain an air of supremacy.

Why does it not surprise me though...considering their track record for truth.

Intel claims Haswell will use 22nm finFETs. So 16.54nm would be significantly better than expected.

 

[8350rocks]

So, haswell is, essentially, a 16.54 nm FinFET...hmm...my how they resort to fallacy to maintain an air of supremacy.

Why does it not surprise me though...considering their track record for truth.

Intel claims Haswell will use 22nm finFETs. So 16.54nm would be significantly better than expected.

I meant next hop after that...was thinking skylake and typing haswell...thanks for pointing that out.

 

[Cazalan]

The APU motherload, but don't get too excited it's probably fake.

Unknown AMD SoC Features 8 Bulldozer Modules, 1024 Stream Processors, 512-Bit Memory Bus.

 

[8350rocks]

The APU motherload, but don't get too excited it's probably fake.

Unknown AMD SoC Features 8 Bulldozer Modules, 1024 Stream Processors, 512-Bit Memory Bus.

Possibly a future Opteron APU? Not likely, but not out of the question either.

 

[anxiousinfusion]

Regarding the cellphone "killing" desktop debate, David Lynch describes perfectly why the desktop platform will never be supplanted for media use: http://www.youtube.com/watch?v=wKiIroiCvZ0

Also Xbit talks about this supposed monster APU: http://www.xbitlabs.com/news/other/display/20130508194707_Unknown_AMD_SoC_Features_8_Bulldozer_Modules_1024_Stream_Processors_512_Bit_Memory_Bus.html

 

[palladin9479]

Well, here at Experian we just had a PC refresh and they all gave us lappies, replacing desktops... So now we have to take them home and I hate it.

That is why their giving you a notebook instead of a desktop. Their hoping to *cough* "leverage mobile computing to optimize worker contribution", aka their trying to wrangle more work out of you for less money. Most likely a HR monkey suggested it and used a powerpoint slide to convince people that the new free man hours would be worth the expense.

Notebooks are inefficient for office work, cramped keyboards and poorly positioned screens are the reason. If you connect a proper mouse + keyboard + monitor to them then they function as desktops but at a considerably higher cost. For quick connect / disconnect your not talking further cost per unit by adding in port replicators. Unless there is a need to go mobile then your whomever is doing your engineering is throwing money into a fire.

 

[palladin9479]

Also someone mentioned HP and Dell losing so much money now compared to before maybe its because their deals suck and everyone knows they suck i don't know anyone who likes Dell or HP hopping until Newegg real quick shows their deals are extremely bad, all this shows is if you keep producing unreliable crap and overcharge it people will learn. HP wants 500$ for a A4 laptop. Lol you have to spend 700+$ to get a A8 laptop from them i hope they do go under.

I think Dell and HP have another problem. I'd say that current gen teenagers are more aware of what a computer really is and go to, for example, Newegg + friend "who knows" or BestBuy/Fry's + assembly service. And that permates to current adult gen as well. Nowadays, every one has a friend or relative that "knows about computers", right?

Dell and HP don't cater that much now for that type of people.

Lappies are another beast though. I'd say in lappies the competition is even more fierce, since most companies have free reign on how to design theirs unlike desktop PCs (like, the only benefit for non-standarization, haha). They have to thin the profit margins (I think) to the point where they must be the best not only at product level, but at manufacturing level (costs). And that's a lot of hard work, I'd say. Boutique ones are excluded from this Logic, I think.

Cheers!

I can't speak for HP as we don't do server with them but Dell is still making a killing in the business sector. Every hardware refresh cycle we engineer for and procure another hardware baseline. Typically optiplex's for Office Automation (OA) work and Precissions for special applications work (they need to be mobile and have removable HDDs for secure storage). We tend to buy them by the thousands and have the pricier support agreement. Something breaks and Dell immediately sends out Mr. Wei (their Enterprise support guy for the area) to replace it.

You gotta separate consumer and business markets, they have different needs and different budget constraints.

 

[Cazalan]

The APU motherload, but don't get too excited it's probably fake.

Unknown AMD SoC Features 8 Bulldozer Modules, 1024 Stream Processors, 512-Bit Memory Bus.

Possibly a future Opteron APU? Not likely, but not out of the question either.

That would be like an Opteron 6380 fused with a Radeon 7850.
Transistor count. 2.4B + 2.8B - 1B (L3 cache) = 4.2B

That's a big boy. Maybe at 20nm.

 

[Blandge]

Well, here at Experian we just had a PC refresh and they all gave us lappies, replacing desktops... So now we have to take them home and I hate it.

That is why their giving you a notebook instead of a desktop. Their hoping to *cough* "leverage mobile computing to optimize worker contribution", aka their trying to wrangle more work out of you for less money. Most likely a HR monkey suggested it and used a powerpoint slide to convince people that the new free man hours would be worth the expense.

Notebooks are inefficient for office work, cramped keyboards and poorly positioned screens are the reason. If you connect a proper mouse + keyboard + monitor to them then they function as desktops but at a considerably higher cost. For quick connect / disconnect your not talking further cost per unit by adding in port replicators. Unless there is a need to go mobile then your whomever is doing your engineering is throwing money into a fire.

I actually prefer a laptop for office use. My employer gives each employee a docking station as well as any peripherals that may increase ergonomics. Certain jobs may require desktops, and those are distributed as needed, but
I work in a lab, so it would be very inconvenient to have to go to my desk (several floors away) to get any documentation work done.

 

[Cazalan]

Regarding the cellphone "killing" desktop debate, David Lynch describes perfectly why the desktop platform will never be supplanted for media use: http://www.youtube.com/watch?v=wKiIroiCvZ0

That was pretty funny but phones already have enough power for 1080p decoding. You just need to doc to your HDMI TV.

 

[juanrga]

Mike Bryant writes:
I’ll repeat some useful numbers I’ve given before at our events. As can be seen for SRAM the Intel 22nm is almost but not quite twice as dense as their 32nm process, whilst TSMC and GF processes fit in roughly to the Intel densities as given by the names of their nodes.

Once upon a time half the metal M1 pitch defined the node but in recent years metal pitch reduction has not kept pace with other features so fabs have tended to reference the node as the ratio to the density of SRAM at 65 or 90nm.
Certainly there is a LOT of market BS taking place as well (Intel included) but in reality Intel does lead the way in density at least.

Of course SRAM density isn’t everything but with BEOL limited by the desire to stay with single patterning, for logic the metal routing now begins to dominate over transistor size.

The drawn gate length is not really a relevant measure any more and in any case has never been the official definition of the node. It is generally used to set the performance of the transistor, shorter is faster whilst longer is slower but less leakage. The move to FinFETs allowed the performance and density of the next node to be achieved without going to shorter gate lengths which would have excessive leakage.

The additional complication of FD-SOI as a competing process technology will prove interesting. Despite the hype, FD-SOI SRAM densities won’t match the densest FinFET SRAMs at the same node, but leakage will be noticeably better, causing a definite process branch applicable to numerous applications.


The stats are :

SRAMs
(HP and special low power versions are larger for all fabs)

Intel 45nm SRAM cell – 0.346um^2
Intel 32nm SRAM cell – 0.171um^2
Intel 22nm SRAM cell – 0.092um^2

TSMC 40nm SRAM cell – 0.290um^2
TSMC 28nm SRAM cell – 0.127um^2
TSMC 20nm SRAM cell – 0.090um^2

GF 28nm SRAM cell – 0.120um^2

ST 28nm FD-SOI SRAM cell – 0.120um^2 (this is believed to be the version with no back-gate and uncompetitive leakage – 0.152um^2 for back-gate and lowest power/leakage but of course LP versions of other processes are also larger)


Metal pitch
(once upon a time half this was the node size but as can be seen the transistors have shrunk a lot more than the metal)

Intel 22nm metal pitch – 64nm
Intel 14nm metal pitch – 48nm

TSMC 28nm metal pitch – 64nm
TSMC 20nm metal pitch – 64nm

GF 14nm metal pitch – 48nm (predicted)


Contacted gate pitch
(this is a key dimension in that it is no point making transistor gates shorter unless you can reduce this number to match)

Intel 32nm contacted gate pitch – 112nm
Intel 22nm contacted gate pitch – 90nm (80nm with special processing)
Intel 14nm contacted gate pitch – unknown

TSMC 40nm contacted gate pitch – 160nm
TSMC 28nm contacted gate pitch – 118nm
TSMC 20nm contacted gate pitch – unknown
TSMC 14nm contacted gate pitch – unknown

GF 28nm contacted gate pitch – 113nm
GF 20nm contacted gate pitch – 80nm
GF 14nm contacted gate pitch – not relevant as BEOL is stated as being the same as 20nm. However this may change in due course.


I hope this helps

Mike Bryant
http://www.electronicsweekly.com/mannerisms/general/the-intel-nanometre-2013-02/

From the comments section below from your link.
The density can no longer be described as the gate length, simple as that, so dont let marketing lead you.

I find other interesting comments in the same link.

Asked the question: ‘What it is that 22nm, in the context of Intel’s 22nm finfet process, is a measurement of,’ Intel tell me: “These processes are very complex, and there are many feature sizes. The details we’ve given in terms of fin width, gate pitch and interconnect pitch are all we are willing to share.” Not exactly an answer to the question: ‘What is it that 22nm is measuring? Could it be that nothing measures 22nm on the Intel 22nm finfet IC process?

I asked Intel two more questions: So 22nm doesn’t relate to anything in particular? and Do you know why that number was chosen?
The answer I have received to those questions from Intel is: “Quite honestly I don’t know.”

What? They don't know how they got the 22nm figure that they are marketing?

Also relevant is the quote from Mojy Chian (Globalfoundries):

Intel’s terminology doesn’t typically correlate with the terminology used by the foundry industry. For instance Intel’s 22nm in terms of the back-end metallisation is similar to the foundry industry’s 28nm. The design rules and pitch for Intel’s 22nm are very similar to those for foundries’ 28nm processes.

Meanwhile I found this

http://www.xbitlabs.com/news/other/display/20130211140309_GlobalFoundries_10nm_Process_on_Track_for_2015_7nm_Fabrication_Process_Due_in_2017.html

2014 --> 14 nm
2015 --> 10 nm

WOW

 

[jimmysmitty]

Steamroller will be 28nm Right?

I knew that Intel lies about benchmarks. I have just discovered that Intel also lies about the size of the process/node:

As the world now knows, there’s a nanometre and then there’s an Intel nanometre.

To most of us a nanometre is a nanometre but, to Intel, a nanometre measures 1.182 nanometres.

http://www.electronicsweekly.com/mannerisms/general/the-intel-nanometre-2013-02/

Intel’s 22-nm node is really 26 nm

http://www.eetimes.com/electronics-news/4403320/Fur-flies-over-FinFETs-and-future-in-IEDM-panel

They all follow the same naming scheme.

Either way the 22nm stems fro the smallest part. Not all of the CPU is 22nm just like not all of the CPU in BD/SB is 32nm.

Its always been this way from what I remember.

Still what's more interesting is what is going to push beyond 14nm as silicon is hitting its limits.

 

[JAYDEEJOHN]

OK, it is 22nm, as each aspect of the transistors are shrunk, but having options such as doping, you can change gate length for speed/tdp/perf, depending on design, combined with the density, or total number you want to achieve.
Its a marriage of space for best compatibility for density/speeds and leakage

 

[JAYDEEJOHN]

While it is crucially important to develop leading-edge fabrication processes, it is also essential to ensure high production volumes. While GlobalFoundries will be able to process around 190 thousand 300mm wafers per month (including 20nm, 28nm, 32nm, 40nm and other process technologies) by late 2013, its arch-rival Taiwan Semiconductor Manufacturing Company can already process 1.076 million 300mm wafers a month and will further boost output by the end of the year.
http://www.xbitlabs.com/news/other/display/20130211140309_GlobalFoundries_10nm_Process_on_Track_for_2015_7nm_Fabrication_Process_Due_in_2017.html
Again juan, from your link, some tasty infos there, look at TSMCs output, and this was dated before the announcement of their new fab coming on line, where it was hinted at here.

 

[-Fran-]

That is why their giving you a notebook instead of a desktop. Their hoping to *cough* "leverage mobile computing to optimize worker contribution", aka their trying to wrangle more work out of you for less money. Most likely a HR monkey suggested it and used a powerpoint slide to convince people that the new free man hours would be worth the expense.

Notebooks are inefficient for office work, cramped keyboards and poorly positioned screens are the reason. If you connect a proper mouse + keyboard + monitor to them then they function as desktops but at a considerably higher cost. For quick connect / disconnect your not talking further cost per unit by adding in port replicators. Unless there is a need to go mobile then your whomever is doing your engineering is throwing money into a fire.

This down here:

I actually prefer a laptop for office use. My employer gives each employee a docking station as well as any peripherals that may increase ergonomics. Certain jobs may require desktops, and those are distributed as needed, but
I work in a lab, so it would be very inconvenient to have to go to my desk (several floors away) to get any documentation work done.

In Experian, it's the same thing: Lappies came with a Dock station (no complaints, since it's movable in every way) and Mouse + Keyboard + Extra Monitor. Plus, they gave us the extra charger (home) and a travel bag for the lappy.

And regarding the different markets: of course there's a huge difference, I was just referring on my impression of the consumer market only. I really don't know how servers are panning out, but I do know IBM and Dell are the biggest players still.

Cheers!

 

[JAYDEEJOHN]

You divide the current node by about thirty %, every other node is half that of two generations before.
This isnt exact, but every other node is exactly half of two nodes prior.

Lets say, at 22nm, the density is fine, you have plenty of room, but the gate length at a real 22nm is too hot, or it leaks too much, or you get tunneling, by adjusting its length, you can get the desired results, leakage you can live with, speeds that are acceptable etc.
Some things dont shrink as well, or not at all, and have to remain at a given size.
But yes, as jimmy points out, the smallest structures are the size of the node

 

[JAYDEEJOHN]

http://translate.google.com/translate?sl=auto&tl=en&js=n&prev=_t&hl=en&ie=UTF-8&eotf=1&u=http%3A%2F%2Fwww.planet3dnow.de%2Fcgi-bin%2Fnewspub%2Fviewnews.cgi%3Fid%3D1368122313
Another link, now it just has to be double confirmed

 

[palladin9479]

That is why their giving you a notebook instead of a desktop. Their hoping to *cough* "leverage mobile computing to optimize worker contribution", aka their trying to wrangle more work out of you for less money. Most likely a HR monkey suggested it and used a powerpoint slide to convince people that the new free man hours would be worth the expense.

Notebooks are inefficient for office work, cramped keyboards and poorly positioned screens are the reason. If you connect a proper mouse + keyboard + monitor to them then they function as desktops but at a considerably higher cost. For quick connect / disconnect your not talking further cost per unit by adding in port replicators. Unless there is a need to go mobile then your whomever is doing your engineering is throwing money into a fire.

This down here:

I actually prefer a laptop for office use. My employer gives each employee a docking station as well as any peripherals that may increase ergonomics. Certain jobs may require desktops, and those are distributed as needed, but
I work in a lab, so it would be very inconvenient to have to go to my desk (several floors away) to get any documentation work done.

In Experian, it's the same thing: Lappies came with a Dock station (no complaints, since it's movable in every way) and Mouse + Keyboard + Extra Monitor. Plus, they gave us the extra charger (home) and a travel bag for the lappy.

And regarding the different markets: of course there's a huge difference, I was just referring on my impression of the consumer market only. I really don't know how servers are panning out, but I do know IBM and Dell are the biggest players still.

Cheers!

Now think real quick, how much more expensive was that laptop from an equivalent desktop? How much extra for the docking station (should be a port replicator, docking stations have their own internal PCIe bus amongst other things)? The keyboard + mouse + monitor are necessary for office work anyway. Essentially your paying significantly more for the exact same capabilities, unless there is a need to "work from home after work" aka providing free hours.

And now the issue becomes clear, only those people who need travel a lot or otherwise need to perform their work after hours / from home need to have the extra expensive incurred from a mobile solution. Mobile is NOT replacing desktop, it's complimenting it, which has been my assertion the entire time.

 

[esrever]

The APU motherload, but don't get too excited it's probably fake.

Unknown AMD SoC Features 8 Bulldozer Modules, 1024 Stream Processors, 512-Bit Memory Bus.

Possibly a future Opteron APU? Not likely, but not out of the question either.

That would be like an Opteron 6380 fused with a Radeon 7850.
Transistor count. 2.4B + 2.8B - 1B (L3 cache) = 4.2B

That's a big boy. Maybe at 20nm.

its likely something to compete with xeon phi if its real. It would be closer to 3.5B transistors and probably does 500 mflop or 1gflop DP. They can put 2 of them on 1 card if they wanted to.

Its unlikely to be real tho.

that picture was shopped from this one. Someone went and blurred it and ctrl shift I on it.

 

[sarinaide]

The APU motherload, but don't get too excited it's probably fake.

Unknown AMD SoC Features 8 Bulldozer Modules, 1024 Stream Processors, 512-Bit Memory Bus.

That looks like the HD9970 There are no 16 Core APU's and that kind of GPU is never going to be found on a APU for a while yet. That was the exact same die chart used for Volcanic Islands so if it claims to be an APU then its fake.