AMD CPU speculation... and expert conjecture

[skitz9417]

http://www.xbitlabs.com/news/cpu/display/20121122235832_AMD_s_Steamroller_High_Performance_Core_Slips_to_2014_Excavator_May_Face_Delays.html
http://www.xbitlabs.com/news/cpu/display/20121105000202_AMD_s_Roadmap_Slide_Does_Not_Predict_Steamroller_in_2013.html]
http://www.theinquirer.net/inquirer/news/2208525/amd-sticks-with-socket-am3-for-steamroller
http://vr-zone.com/articles/amd-pushes-steamroller-and-excavator-forward-bullish-about-performance-increases/17088.html
http://www.anandtech.com/show/5491/amds-2012-2013-client-cpugpuapu-roadmap-revealed
http://www.overclock.net/t/1310350/amd-steamroller-on-am3/0_100
http://forums.anandtech.com/showthread.php?t=2276782

 

[fazers_on_stun]

Advanced Micro Devices Is Already Inside A 4G Tablet And Nobody Has Noticed

Not exactly Steamy news but still good news for AMD, according to one "analyst" with suspect spelling:

Advanced Micro Devices (AMD) has been battered of late, with some believing it will go bankrupt in due coarse. Plenty of older articles on Seeking Alpha were bullish but many have stopped being bulls given the dire situation of the PC market. However, we believe there is a resurgence of PC sales bubbling from under the ashes. This resurgence will bring back a market where AMD, like Intel (INTC) and Nvidia (NVDA), get more than three fourths of their revenue. This resurgence of PCs will be tablet lead. Wait, did we say the tablet? Isn't that what will finally kill the laptop? Didn't you already see the commemoration funeral in Barron's months back? No we are not mad, let us introduce the Fujitsu Stylistic-Q572, which is available now.

Technically, the tablet is the new PC and AMD is already in one. We were surprised to find AMD with a tablet design win in the Fujitsu Stylistic-Q572; it was spotted with an AMD Hondo (Z-60) chip. The Stylistic even includes cellular 4G broadband access. In that version, it is AMD's first ever cellular tablet design win.

According to Fujitsu it is:

"The ideal lightweight 25.7 cm (10.1-inch) tablet PC for mobile professionals in business, government and SME. With pen and finger support, the slim 820 g device provides perfect flexibility to accomplish your daily tasks."

Certainly looks very desirable. AMD, unlike rivals Qualcom (QCOM), Nvidia and Intel, is the only x86 CPU designer which does not provide baseband chips at all. Thus, it would have likely increased the total cost to the provider to use an AMD chip unless AMD gave them a good discount. This will be a very low cost and similar low margin sale. But it is a sale, and this is important even if only a million units are sold.

Present Revenue Is Important - Isn't Low Margin Bad?

Low margin is not bad, as long as it is free cash flow generating in the case of a company like AMD. When AMD is trading below a $1.7 billion market value, with 2012 revenues between $5.5-6 billion and a P/S less than 0.3, it is hard to rate it a sell. This low valuation has made many professional investment bank analysts to adjust their ratings. They now agree with us. AMD is a weak buy if you take an average of the analyst calls according to Nasdaq.

 

[mayankleoboy1]

Hondo will be 40nm, right ? So power/performance should be around 40nm A9 chips. Which were a generation old.

 

[de5_Roy]

AMD Publicly Releases CodeXL 1.0
http://semiaccurate.com/2012/12/21/amd-publicly-releases-codexl-1-0/

 

[fazers_on_stun]

Hondo will be 40nm, right ? So power/performance should be around 40nm A9 chips. Which were a generation old.

From Softpedia:

Much like the Z-01 accelerated processing unit that it’s designated to replace, Hondo will also use the Bobcat CPU core, which will be paired with a DirectX 11 capable onboard GPU.

According to AMD, Hondo APUs can include either one or two x86 computing cores and will be built on the 40nm node, while its TDP is estimated at a rather low 4.5W.

This is a significant improvement when compared to the 5.9W TDP of Desna, and should help tablets powered by AMD’s next-gen Z-series accelerated processing units to deliver even better battery life.

Going forward, AnandTech revealed that AMD wants to get into the sub-2W market with a new APU design, most probably based on the 2013 Tamesh processor, which will be AMD’s first tablet APU to use a true system-on-a-chip (SoC) design.

I think Temash is the 28nm CPU.

 

[viridiancrystal]

From Softpedia:

Much like the Z-01 accelerated processing unit that it’s designated to replace, Hondo will also use the Bobcat CPU core, which will be paired with a DirectX 11 capable onboard GPU.

According to AMD, Hondo APUs can include either one or two x86 computing cores and will be built on the 40nm node, while its TDP is estimated at a rather low 4.5W.

This is a significant improvement when compared to the 5.9W TDP of Desna, and should help tablets powered by AMD’s next-gen Z-series accelerated processing units to deliver even better battery life.

Going forward, AnandTech revealed that AMD wants to get into the sub-2W market with a new APU design, most probably based on the 2013 Tamesh processor, which will be AMD’s first tablet APU to use a true system-on-a-chip (SoC) design.

I think Temash is the 28nm CPU.

well according to this

Temash will be using GCN, which would imply 28nm.

 

[anxiousinfusion]

This resurgence of PCs will be tablet lead. Wait, did we say the tablet? Isn't that what will finally kill the laptop?

Maybe I'm alone on this but I feel like tablets will ultimately exist *along side of* laptops and desktops. Wasn't it only a few years ago laptops where supposed to destroy and replace the entire desktop market?

 

[esrever]

This resurgence of PCs will be tablet lead. Wait, did we say the tablet? Isn't that what will finally kill the laptop?

Maybe I'm alone on this but I feel like tablets will ultimately exist *along side of* laptops and desktops. Wasn't it only a few years ago laptops where supposed to destroy and replace the entire desktop market?

people were saying PDAs would kill desktops a bit more than 10 years ago

 

[fazers_on_stun]

This resurgence of PCs will be tablet lead. Wait, did we say the tablet? Isn't that what will finally kill the laptop?

Maybe I'm alone on this but I feel like tablets will ultimately exist *along side of* laptops and desktops. Wasn't it only a few years ago laptops where supposed to destroy and replace the entire desktop market?

I think the consensus is that tablets (& smartphones) are for media consumption, desktops (and laptops) are for content creation. Or something like that .

Personally I am 90% consumption at home, and unfortunately 90% creation at work. The first is called entertainment and the second, "work" .

 

[gamerk316]

This resurgence of PCs will be tablet lead. Wait, did we say the tablet? Isn't that what will finally kill the laptop?

Maybe I'm alone on this but I feel like tablets will ultimately exist *along side of* laptops and desktops. Wasn't it only a few years ago laptops where supposed to destroy and replace the entire desktop market?

Mobile will never really obsolete the stand alone or mainframe based architecture, simply due to performance/storage/battery reasons. I do see the number of form factors shrinking a bit though; netbooks are basically obsoleted by tablets, and I suspect laptops aren't long before they also go. I basically see three tiers: Smartphone, Tablet, and Desktop.

 

[mayankleoboy1]

Performance/power is definitely going to improve in the coming years.
Its the battery that has not really improved in the quite some time. Almost every week we hear news about some revolutionary battery tech. But when will it become production ready?

 

[fazers_on_stun]

Performance/power is definitely going to improve in the coming years.
Its the battery that has not really improved in the quite some time. Almost every week we hear news about some revolutionary battery tech. But when will it become production ready?

Geez, hafta agree with you there. For the last 10 years every other battery news story has been "nanotube this" or "nanotube that". So where are all them dang nanotubes improving my battery life by 100%??

Too bad cold fusion never panned out - just think, a 10ml supply of heavy water would power your laptop continously for 20 years ..

The world doesn't have an energy crisis - it has an energy density crisis. Unfortunately fossil fuels are still the most dense form of energy, short of nuclear of course.

 

[mayankleoboy1]

the only thing that makes things impossible is believing that the impossible is in fact impossible and never trying.

Intel believes that they can make trouble free 14nm fabrication tech, such that thermals will be under control.

 

[mayankleoboy1]

edited.

 

[noob2222]

Intel believes that they can make trouble free 14nm fabrication tech, such that thermals will be under control.

sure, just have to adjust things to make it feasable.

the problem lies in making the possible into practical. The question remains at what point is 14nm practical or just simply possible, 6+ghz, 200+ watts is possible just requires LN2. Its not practical. You could add a phase change cooling tower, adding power consumption, but that defeats the purpose of reducing power in the first place.

I guess one thing that is impossible is having enough time to see the impossible become possible. I will likely never see humans travel out of our solar system.

 

[Chad Boga]

Intel believes that they can make trouble free 14nm fabrication tech, such that thermals will be under control.

:lol: :lol: :lol:

 

[noob2222]

:lol: :lol: :lol:

lol, there he is ... so how fast is that 57W broadwell part? Think we will see DT speeds? Should be interesting.

 

[R0ck3tm4n]

Well we can read about how cold fusion, although it has already been done, cannot be done because modern science doesn't understand how it can be done and therefore can't be done or we can let gamerk316 tell us about how a computer program cannot use more than 4 cores because programmers don't understand how it can be done, yet.

I don't have any news or speculation on Steamroller right, maybe we should ask Rory Reed to come here and tell us how all that is going. What do you say, Mr. Reed, are there any engineers working on that still?

 

[viridiancrystal]

Perhaps we can add a comma after AMD in the thread title. That opens up a little more room for discussion.

 

[ebalong]

Well we can read about how cold fusion, although it has already been done, cannot be done because modern science doesn't understand how it can be done and therefore can't be done or we can let gamerk316 tell us about how a computer program cannot use more than 4 cores because programmers don't understand how it can be done, yet.

I don't have any news or speculation on Steamroller right, maybe we should ask Rory Reed to come here and tell us how all that is going. What do you say, Mr. Reed, are there any engineers working on that still?

Yeah, and while you're checking on that Mr. Reed, make sure to tell the engineers that it has to be faster per core than at least Sandy Bridge (because it will be priced a little lower than Haswell, it doesn't have to equal or surpass that generation, but that would be a very pleasant surprise), stock clock at 4.2, turbo to 4.7, TDP can stay at 125W but under load it shouldn't increase the power draw 5-fold; and better execution of the "modules" working in harmony so that it is nearly the same as a true octocore.

In all seriousness, I am actually hoping they somehow pull something good out of their hats; if AMD had a serious competitor to SB-E or whatever IB-E is going to be, in the "budget" (i.e. - non-2011 socket Xeon) workstation arena, I would consider opting for it, even if it was slightly less performance, but the price was right.

 

[de5_Roy]

AMD exec defects to Samsung
http://www.fudzilla.com/home/item/29935-amd-exec-defects-to-samsung
Top Kabini 28nm APU is Kabini X4 5110
http://www.fudzilla.com/home/item/29934-top-kabini-28nm-apu-is-kabini-x4-5110
Top Richland 28nm APU is A10 6800K
http://www.fudzilla.com/home/item/29933-top-richland-28nm-apu-is-a10-6800k
28nm :O. mind. boggled. glofo has their 28nm node ready?!?!

 

[-Fran-]

"The expected launch date is June, but the production should be ready by late March 2013".

That's for Richland, so GF should have working ES around, right? Maybe we'll see them at CES.

Cheers!

 

[amdfangirl]

28nm :O. mind. boggled. glofo has their 28nm node ready?!?!

Better hope so!
Otherwise AMD will be so behind I will just facepalm.
Also why is is there no 28nm AM3+ part?
Saddens me to wait even longer for a CPU upgrade.

 

[Cazalan]

Better hope so!
Otherwise AMD will be so behind I will just facepalm.
Also why is is there no 28nm AM3+ part?
Saddens me to wait even longer for a CPU upgrade.

Steamroller got delayed to 2014 I thought. That's why no AM3+.

They're focusing on mobile now.

 

[gamerk316]

Well we can read about how cold fusion, although it has already been done, cannot be done because modern science doesn't understand how it can be done and therefore can't be done

ALLEGEDLY done, in one experiment, that was never reproduced by anyone, let alone the people who actually RAN it the first time.

or we can let gamerk316 tell us about how a computer program cannot use more than 4 cores because programmers don't understand how it can be done, yet.

A+B=C;
C+D=E;
E+F=G;

Guess what? I can't make that parallel, ever.

Secondly, programs typically do make use of all processor assets, but most of those threads don't do any meaningful work, so they don't show up in Task Manager. You still end up with 2-3 threads that do the majority of the work, and all the smaller tasks get offloaded to other cores. So you get the typical two cores >50%, and the rest hovering well below that. You are NEVER going to see equal work across cores, except in very rare cases where the algorithm in question scales naturally.

OpenMP will help optimize the low hanging fruit (FOR loops being a prime example) but you won't see any significant change in core usage going forward. Though I'm sure the next generation of consoles will be blamed for that too...