AMD CPU speculation... and expert conjecture

Page 376 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.


Interesting to note they actually removed HTT from their "small" cores aka Atom-Silvermont. Citing that the gains weren't worth the added logic. They just doubled the cores instead like Jaguar.
 


So, let me get this straight...

You are going to tell me that because AMD used a tiny ARM core for a separate security protocol on Kaveri, Beema, and Mullins...you're going to tell me that it is a "sign" that AMD is migrating entirely to ARM?

LOL!!!

...

LOL!!!

Ok...to offer a quote from Shakespeare: "Lord, what fools these mortals be..."

EDIT: I would like to point something out before your crazy illogical wandering becomes contagious...

The FTC would have a bone to pick with you about AMD abandoning x86 CPUs. That would create what is called a "monopoly". So if AMD were even considering stopping production of x86 in it's entirety...the government would step in and prevent them from doing so purely on the basis of another term called "free market competition integrity".

That would be the same entity that you, yourself, cited earlier in the article talking about when the FTC sued Intel for monopolistic practices.

I hate to say it, but if AMD were to even consider that course...they would be prevented from doing so. Purely because they are the only competitor with enough market share to even force competition. VIA has less than 3% of the market share and could not plausibly supplant them as a large player in any sense in the consumer x86 market.

/silliness
 


You cannot answer because you don't know which is the efficiency. Ok.
 


From the second link:

AMD makes more profit than Sony on the PS4
Sony makes $18, AMD makes close to $100
 


But...I am pretty sure I do. Because I told you what it was, or is it that you do not know what the efficiency is? There are benchmarks to show GCN > VLIW4 in every facet.
 


I always thought that Nvidia and VIA would be the perfect merger for competing with AMD.
 


The article is talking about revenue, gamerk316 was talking about profit. The article title is completely wrong as the cost is not $0 to AMD. Journalism these days...

If the $100 revenue is accurate and AMD's "mid teens" (quote from Q3 earnings conference) margin is 15% then that would be $13 profit. ($86.9565 x 1.15 = $100)
 



Wow, that really puts the brakes on "if Sony isn't making that much money off of PS4 then neither can AMD, or TSMC, or Foxconn, or anyone else that is a part of making a PS4"
 

The issue is, why would AMD or any others invest in the PS4, there has to be something deeper in..
 



As someone mentioned earlier, AMD feels they are growing with regard to gaming on PC. Games ported from PS4 and Xbox One will be well optimized to run on x86 AMD hardware. Many software developers are lazy, so I don't see them changing the code much for the PC ports. Probably support so they will run properly on a Windows platform and little else. This could give AMD an edge in future gaming.
 


The real issue is why have we spent time discussing the fact that if Sony isn't making massive profits then neither is anyone else related to PS4.

AMD will not stop making x86 CPUs until their big GPUs and PC gaming stops being profitable to them. If AMD drops x86 CPUs, they will leave the only platform for AMD GPUs to run in in the sole hands of Intel.

You know, the same Intel that S|A wrote about who wants to kill PCIe to remove AMD and Nvidia from competing with their GPUs.

AMD needs to keep making big x86 CPUs to support their dGPU and gaming business. If they don't, Intel could easily show up with a new platform with one PCIe 3.0 4x port max and watch as AMD burns.

The only way AMD could go ARM only would be if they convinced members of the HSA Foundation as well as game studios to support other platforms like ARM. They would have to get major companies to also port over their existing solutions (Visual Studio, Photoshop and other Adobe products, Microsoft Office, etc).

I do see it as feasible at some point in time but I don't see it happening any time soon. AMD is in control of the tools for HSA and Mantle game development but it's already a hard enough sell to get people to optimize just for GCN chips, imagine if you had to convince people to make Mantle + ARM ports.

Also, take a look at Via in the last JPR. They are the biggest losers so far and they are doing what some of you are suggesting in this thread, focusing on mobile and lower end products.
 


That nobody is porting Photoshop to run in a phone has little to see with the ARM ISA. Just by curiosity how many x86 phones you know that run full Photoshop? Do you conclude that the problem is the x86 ISA? It is not.

There is a version of Microsoft Office for ARM tablets. It lacks some stuff of the PC (x86) version and adds other (touch support) because it is for tablets, but according to Microsoft "differences between the two products are subtle."

About AMD future plans, AMD chief executive already said:

AMD will no longer focus on delivering high-performance products built on the latest process technology.

http://www.extremetech.com/computing/116977-goliath-wins-amd-retreats-retrenches-and-seeks-to-reinvent-itself

http://www.pcper.com/news/Editorial/AMD-will-not-chase-Intel-making-needlessly-powerful-CPUs

http://www.anandtech.com/show/5492/amds-rory-read-outlines-amds-future-strategy

I expect he did mean x86 only... at least the server roadmap seems to indicate so
 


That is, you took the first part of my post, ignored completely the second part, and just used this as excuse to post your really lovely idea what if AMD want to abandon the market, the FTC would obligate them to continue to work

<facepalm>
 
Uhm... The console business model is not the same as a regular product supplier like AMD or Qualcomm. I'm getting the vibe you guys think Sony is going to have a bad revenue stream because of the close-to-zero profit margin on each console sold, but the reality is that consoles make money out of the games they sell and the content they provide (this one is new). Nintendo is the only console maker that has a different business approach to them.

And I agree with AMD not making a $100 "profit" out of every PS4 sold. Profit = sum(Revenue) - sum(Costs), remember that. It's the basis for companies, haha. I will bet the guys from the article used in a loose manner the term "profit"; they shouldn't have used any specific term to refer to AMDs cut; it's just a "cost" to Sony, but by no means "profit" to AMD.

"This sees AMD making quite a healthy profit per console.", from the first link. If we go by the definition of profit, it's an inaccurate statement (i.e. false), so please drop the economics discussion in regards to how much money AMD makes, since although the links provide good info, it's not accurate in the terms used and incomplete, as they don't say how much it costs to AMD make each APU sold to Sony.

Cheers!

PS: http://en.wikipedia.org/wiki/Profit_(economics)

EDIT: Lol, after reading the comments, they all point out that same mistake in the article.
 


From the second link:

AMD makes more profit than Sony on the PS4
Sony makes $18, AMD makes close to $100
 


Still waiting for the answer
 



If AMD has 18-20% x86 market share, and VIA 0.5% then how is that even close to a monopoly? A monopoly would be if no one else can produce x86 parts period. The ARM CEO is tossing around a term that he knows scares people.

"A monopoly (from Greek monos μόνος (alone or single) + polein πωλεῖν (to sell)) exists when a specific person or enterprise is the only supplier of a particular commodity."

Anyway, the point I was making is that AMDs troubles are from the resources the competition has. Lack of capital to demand cutting edge process nodes being one of the major ones. Lack of R&D budget to produce and test multiple die configurations. Lack of budget to hire key engineers away from Apple/Qualcomm/Intel.

If AMD (2.48B market cap) has trouble competing with Intel (122B mc). How much easier will it be to compete with say Qualcomm (120B mc) or Samsung (210B mc)? Excuse the rudimentary MC as a gross valuation but it shows ball park resources.

With so many ARM competitors a company like Qualcomm is free to undercut any particular product they want to. When AMD employees start adding ARM core development skills to their resume they become ripe for poaching.
 


And to add hurt to injury... Right after the ATI buyout, AMD sold the low power technology for GPUs to Qualcomm, haha. Man, what a bad irony now.

I don't think AMD will step away from X86 for the simple reason that even though Intel has the upper hand in mostly everything right now, AMD is not incapable of actually making a profit out of that market. Intel has managed to keep them in the support line (ICU?) for a good while, but they've had their success stories against Chipzilla. Now, the APUs are getting a great spotlight thanks to being inside the consoles and that I'm sure will translate to some more sales in the lower power segment. Not a big market, but profit is profit I guess.

Cheers!
 
In other news, ASUS has finally got their ThunderboltEX card certified and updated, which is a decent advancement http://rog.asus.com/280892013/news/asus-thunderboltex-ii-what-you-need-to-know/. This technically should be able to work on any board since the original Crosshair V and P8P67 series, but I bet Intel will go "ERMEHGAWD NO BOARD BESIADE LIAK SEEZEXTEHHATE THRU ZEEAITYSEVEN CAN LIAK GIVE TEH PROPHURR VULTAGEZ OR ELSE DAT STAWFFF BE BRICKED"
 


If you really wanted to you could break it all the way down to the person who shoveled the sand for the silicon wafer.
 


Indeed. To expand more on what I said earlier, and to just blatantly say it: I think that as time goes on piledriver CPUs will begin winning benchmarks against haswell more and more often. Optimization is not a small thing at all.

This is somewhat unique, because back when the original xbox was out that was the only other time x86 was used in a console. At that time it was an intel chip and software was generally optimized in that direction anyway. Even if it wasn't, everyone only had one single core, so there wasn't any fancy breakdown of resources on the chip. At least nothing like how it is now.

It was also really during the 360 era that console ports to PC started happening alot, and at that time the consoles were generally some powerpc type, which doesn't directly translate to anything. I expect this behavior to continue because it is well known, however the landscape is suddenly very different because of Jaguar.
 


I explained what I mean by monopoly in the paragraphs that you snip from my message: from the Nvidia example to the final two links. I did mean monopoly of the x86 architecture.

There is a problem with your numbers and it is that one cannot isolate x86 and ARM as disjoint markets and then pretend that AMD either fights Qualcomm or fights Intel. It is not so anymore. There is an ARM vs x86 fight (from phones to servers/HPC) and AMD is convinced that ARM will win

why-arm-will-win.jpg


That is why AMD plans to recover server market share to intel using ARM, not x86.

If AMD had stuck to x86, released a 16-core Steamroller Opteron, it would waste lots of money and recover that at best 1% of market share? If AMD had accelerated plans and released a superb 16-core Excavator Opteron that destroy anything Intel, they had gained market share to Intel just to lost it against future ARM servers.

The move to ARM is a wise decision.
 
About console profits. AMD is not only earning money from selling silicon but also from licensing graphics technology for custom designs.

Nvidia reaction to AMD win of console business was to license also

http://www.mcvuk.com/news/read/nvidia-to-licence-graphics-tech-to-third-parties/0117365


Due to x86 monopoly, AMD cannot license x86 technology to third parties, only sell the x86 silicon they make. Nvidia couldn't even obtain a x86 license.
 
Status
Not open for further replies.