Does AMD has some future?

[ferwindjacks]

That apple rumor is false. They did testing, but inside sources stated it would be too difficult to switch at this point in time (especially considering support needed in their Bootcamp utility, and other programs). Apple is only developing processors for their mobile devices, which turned out to be the best course of action considering they can pick and choose performance necessary for that specific device.

 

[-Fran-]

I don't think OEMs will like Intel as the only supplier for their core markets. Even more, I don't think OEMs will support that idea at all. There's a reason why IBM told Intel to let AMD license X86. It's not even a "monopoly" thing, it's OEMs not wanting Intel as the only supplier. I'm sure OEMs are scratching their heads now, because it's a period where there's no clear future and every path is a gamble: if you go ARM (notice how they're not putting 65W CPUs in regular PCs yet?) and start playing the tablet/phone game you already have Samsung, Nokia, Apple and LG/Sony already kicking it hard (you can also say Xiao-asdfasdf and Huawei); just ask Dell and Amazon. If they lock themselves with Intel, there IS a trend that PCs are being replaced by tablets because there is ZERO incentive to move from the old Pentium Dual or Athlon 64 X2 to a more recent PC/Lappy (I've seen a lot of Turion 64 CPUs around alive and kicking), making OEMs burn money on designs and other products that are bound to not meet sale targets, hence lose the investment. Google is the only one trying with a radically different mix, but it's slow (to say the best) in adoption rate.

All in all, OEM are not even trying to sell PCs and Lappies with the "this thing is more powerful than ever before" slogan anymore. They're trying to do the "this is a gorgeous PC! look how shiny it is!", trying to mimic Apple's way of selling new shiny stuff.

So, in short, until someone using ARM comes out of the closet and gets into PC territory for real, Intel will still charge 200 greenies for a mediocre i3 or i5 locked, plus PC sales will continue the downward trend unless more computing power is actually needed. I hope the VR movement makes a jump. I have high hopes with Windows 10 to up the stakes a bit in favor of actually making an upgrade a real necessity. Games as well, but they'll be tied to console level until next gen, so no great hopes there.

Cheers!

How many server OEMs chose AMD Opterons instead Intel Xeons?
How many boutique OEMs chose AMD Centurion instead Intel DC?
How many laptop OEMs chose AMD Jaguar/Steamroller instead Intel Silvermont/Haswell?
And so on.

Many of them don't have a choice in practice. Thus we are already in that situation where there is almost a single supplier of x86 chips. Do you believe Cray is happy by being forced to use Intel chips? Do you believe Apple is happy by being forced to use Intel chips?

What are doing Cray and Apple? Are waiting for an inexistent AMD comeback to the x86 market? Nope. The former has associated with Cavium and will start building clusters based in ARM SoCs. Apple is developing its own desktop class CPUs, and the last rumor is that Apple is preparing to replace Intel by its own CPUs after Broadwell.

Like you say, the key point is "choosing". At this point in time, the practical reason they're with Intel, is because AMD doesn't have anything realistic to compete with. If AMD was to disappear, then OEMs will be at the mercy of Intel and with no power to "choose" anything. Intel won't play nice in that position, they're a company and will want to take advantage of that position.

Plus, you're still thinking of servers only. Cray doesn't give a rats ass about desktop class CPUs, nor any other server-specific OEM. I'm thinking more of Dell or HP here. They cover a wider spectrum. Apple on the other hand has its own ecosystem; they can do pretty much anything they want. Plus, they don't own much of the desktop market AFAIK.

That apple rumor is false. They did testing, but inside sources stated it would be too difficult to switch at this point in time (especially considering support needed in their Bootcamp utility, and other programs). Apple is only developing processors for their mobile devices, which turned out to be the best course of action considering they can pick and choose performance necessary for that specific device.

That's interesting. Do you have a source by any chance for that info?

Cheers!

 

[gamerk316]

It does make sense though. If Apple goes away from X86, they need to essentially re-write their SW from scratch again. They could go from PPC to X86 because X86 was simply powerful enough to brute force PPC at the time, but that can't happen with ARM. Point being, Apple tied themselves to X86, at least as long as they want to keep their OS SKUs separate.

 

[ferwindjacks]

Here is a source that covers most of the Intel v ARM in Apples case. More of an opinion article, but has a strong case backed by evidence.
http://www.cultofmac.com/144942/why-youll-probably-never-own-a-mac-with-an-arm-processor-feature/

Gamerk hit the mark, it was much easier switching from Power PC to Intel back then than it is to make another switch now. They would have to rewrite everything.

Honestly haven't been keeping up with every post, but didn't AMD report a higher than expected loss of profit in 2014 even with the console sales? It seems I keep seeing Intel doing well and AMD keeps falling behind. Personally haven't owned any AMD products, but have built with them/used them. I'll stick with Intel/Nvidia.

 

[-Fran-]

It does make a lot of sense. The uplift of moving from one ISA to another can't be simple.

Now, there's a counterpoint; simple, but works as a counter point anyway: Apple has WADS of cash. Like, too much money. If they *really* want to move away from Intel (whatever reason they might have), they can, like any person would, throw money at it until OSX is fully ARM compatible.

Plus, don't they have the source code? It's not like they're working from closed binaries or anything. The Kernel from OSX is still FreeBSD-ish, right? That thing, just like the Linux one, run even on toasters, isn't it?

Cheers!

 

[8350rocks]

New rumor that AMD is for sale

http://channeleye.co.uk/amd-rumoured-to-be-up-for-sale/

AMD just needs to get out of the CPU market and stick with the graphics cards because then they will actually succeed more as a company I think in total profit. Of course Intel would have a monopoly in which, at least in the U.S., the government will have to use antitrust legislations to regulate the prices of the processors.

not so sure about that. their FX cpus still sell like hotcakes to budget gamers and the like. mabye they should just continue down that avenue?

That isn't helping them any. Its a very tinny market compared to even their APU sales. I'm sure K12+GCN could keep them alive in the dgpu market(few more years) and they might be able to get into mobile. Also they could make a lot of money one day on K12 servers.

The issue lies in there being no money in mobile.

Samsung, one of the biggest chip vendors for mobile devices, lost money 3rd quarter, mostly because of competition requiring them to cut prices to compete. Meaning that AMD entering the mobile market will not be a wise idea at this time. In essence...that ship sailed about 6 years ago.

 

[ferwindjacks]

You're mostly right. While OSX is almost completely open source, and they do have the framework to completely switch to ARM, it would take years for them to completely rewrite the software, which is something they don't want to commit to considering that they are all about competing with Windows and work on software performance improvements (Yosemite brought a basket full of these) and features to entice users to upgrade. Bit of a run on sentence, but hear me out. Why would they switch to ARM when sugar daddy Intel is making strides in keeping size and power consumption down, while keeping performance up (Broadwell M is an amazing example). If they switched to ARM, they would have to have a couple of little tiny processors just to keep up with the performance of the, say, Broadwell M. Granted they will most certainly reach that performance someday, but the tech just isn't there yet.

It's not just their software they have to worry about either. It's compatibility. Existing programs would need to be updated, Bootcamp would be entirely hosed up considering Windows has no support for such processors, and I'm sure many other factors weigh in. I think Apple will just hold their cash close to them and stay with Intel for a long while.

In my opinion, I think it would be cool to have the option of how many ARM processors you want in your Macbook, almost like a customization feature. I say processors plural, because they'd need more than one just to keep up with performance cost of OSX, which they might need to strip down for it anyway. But having the option of multiple processing chips to further increase power, sort of a choose your mileage type thing, would be sick.

A bit hard for me to get all my thoughts down here, but I hope you get the gist of my reasoning. I would like to point out that these posts are completely off topic.

 

[juanrga]

I did mention servers, HEDT, and laptops, not "only servers" Yuka.

The cultofmac article is completely outdated. In the first place Kanter's predictions have been proven wrong. Intel has not dominated mobile space by 2014 but abandoning it for 2015. Why? Because the ARMv8 ISA advantage over x86-64 is bigger than Kanter and others believed. ARMv8 didn't exit when he wrote what is quoted in the article. Moreover, the traditional 3 years advantage of Intel foundries has disappeared as well (another point that Kanter did not see). Today Samsung is already producing 14nm chips and full convergence with Intel foundries will be achieved at 10nm probably.

Everyone with some inside info knows that Apple is dissatisfied with Intel Haswell/Broadwell. There is a reason why Apple has developed a desktop class core (Cyclone) and then downclocked it to fit inside a phone. Apple want use the same hardware from phones to desktops. Several Apple engineers have confirmed the switch to ARM hardware, and one ex-apple executive confirmed he think the switch will be made soon, but I think will not be before the A10.

Main software will be rewritten, whereas legacy apps will run under emulator (an ARM processor can run x86 software with an overhead of 20% or so).

http://appleinsider.com/articles/15/01/14/apple-may-shun-intel-for-custom-a-series-chips-in-new-macs-within-1-2-years
http://www.theinquirer.net/inquirer/news/2390856/intel-ceo-dismisses-rumours-apple-will-look-to-arm-for-future-mac-chips

 

[ferwindjacks]

"Power between the Atom and Intel Core i3". That's pathetic. No one can justify spending 1k+ on a Macbook with performance like that. I said that the tech isn't there yet, and I still hold true to that. I'd say within 5 years or so you might see some articles with a possible switch. The Apple ARM processors just aren't at that stage of performance that could be acceptable.

 

[-Fran-]

Well, you always bring Craig to "desktop" discussions, so you got me confused.

In any case, if Apple is the only one bringing "desktop level ARM" CPU/SoC 's, then it won't make much of a difference. Now, to your side, Juan, I find it interesting that Windows 10 will be "free" for the Raspberry Pi 2: It's an MS OS with full blown ARMv7 (it's a Cortex A7 quad) ISA support (maybe not all, but meh). That could be a hint all by itself of things to come.

Cheers!

 

[gamerk316]

Main software will be rewritten, whereas legacy apps will run under emulator (an ARM processor can run x86 software with an overhead of 20% or so).

Worse then that. Core ISA implementation is actually quite trivial, since you can get 1:1 in most cases. X86 is hard to do this for though because of things like SSE extensions, which don't have 1:1 replacements in other CPU arches. As a result, that is where you lose almost all your performance. You also need to keep in mind ARM has a much less strict memory access model then X86 does, so there's a lot of performance loss there as well.

Point being, X86, being the most complicated/powerful CPU arch right now, is near impossible to emulate well. You can do it, but the performance is going to be BAD.

 

[juanrga]

Well, you always bring Craig to "desktop" discussions, so you got me confused.

In any case, if Apple is the only one bringing "desktop level ARM" CPU/SoC 's, then it won't make much of a difference. Now, to your side, Juan, I find it interesting that Windows 10 will be "free" for the Raspberry Pi 2: It's an MS OS with full blown ARMv7 (it's a Cortex A7 quad) ISA support (maybe not all, but meh). That could be a hint all by itself of things to come.

Cheers!

When discussing desktops I mentioned AMD Centurion and Intel Devyls Canyon.

Apple is not the only one releasing desktop-class ARM designs. Nvidia Denver provide Haswell-class performance and they have long-term plans to use ARM on the desktop

http://blogs.nvidia.com/blog/2011/01/05/project-denver-processor-to-usher-in-new-era-of-computing/

"Power between the Atom and Intel Core i3". That's pathetic.

I think he is underestimating performance of the future cores.

Now since this is an thread about AMD as a company in trouble. I will stop here.

 

[-Fran-]

When discussing desktops I mentioned AMD Centurion and Intel Devyls Canyon.

Apple is not the only one releasing desktop-class ARM designs. Nvidia Denver provide Haswell-class performance and they have long-term plans to use ARM on the desktop

http://blogs.nvidia.com/blog/2011/01/05/project-denver-processor-to-usher-in-new-era-of-computing/

Wait, Centurion and Devil's Canyon are ARM designs? Why mention them when talking about ARM? Are they you're "near ARM power envelope" CPUs from AMD and Intel?

And yeah, we already discussed about Denver and their comparison to a low clocked Celeron ULV.

I actually trust more in Apple than Qualcomm or Samsung making a Desktop class SoC / CPU using ARMv8.

In any case, what are your thoughts of Windows 10? Also the VR turn they're trying to get on with it. Think ARM will be able to cope with that?

Cheers!

 

[-Fran-]

Main software will be rewritten, whereas legacy apps will run under emulator (an ARM processor can run x86 software with an overhead of 20% or so).

Worse then that. Core ISA implementation is actually quite trivial, since you can get 1:1 in most cases. X86 is hard to do this for though because of things like SSE extensions, which don't have 1:1 replacements in other CPU arches. As a result, that is where you lose almost all your performance. You also need to keep in mind ARM has a much less strict memory access model then X86 does, so there's a lot of performance loss there as well.

Point being, X86, being the most complicated/powerful CPU arch right now, is near impossible to emulate well. You can do it, but the performance is going to be BAD.

Don't they have "NEON" in ARMv7 and ARMv8?

Cheers!

 

[ferwindjacks]

I as well have faith in Apple. They are the first to bring desktop class architecture to ARM, and will be leaders in that field.

I think ARM is perfect for VR gadgets in the sense that they could reduce overhead on the CPU that may be computing the high definition display and the circuit boards inside. More specifically, something used in Microsofts Hologram or whatever it is called is prime real estate for ARM.

Liking the way Windows 10 is turning out, DX12 has me excited as well. I'd like to put a lot more down on each subject but it has nothing to do with the OP

 

[juanrga]

AMD rated B minus for long-term scenario

http://www.reuters.com/article/2015/02/02/fitch-affirms-amds-idr-at-b-outlook-stab-idUSFit91263120150202

AMD's Recovery Ratings (RRs) reflect Fitch's belief that the company would be reorganized as a going concern rather than liquidated in a bankruptcy scenario. To arrive at a going concern value, Fitch believes AMD would: i) reorganize businesses serving target markets (Enterprise, Embedded, and Semi-Custom, graphics chips, and APUs), and ii) wind down the legacy PC business.

Agree with ii), only partially with i) The graphics division will be unsustainable in long-term.

 

[-Fran-]

AMD rated B minus for long-term scenario

http://www.reuters.com/article/2015/02/02/fitch-affirms-amds-idr-at-b-outlook-stab-idUSFit91263120150202

AMD's Recovery Ratings (RRs) reflect Fitch's belief that the company would be reorganized as a going concern rather than liquidated in a bankruptcy scenario. To arrive at a going concern value, Fitch believes AMD would: i) reorganize businesses serving target markets (Enterprise, Embedded, and Semi-Custom, graphics chips, and APUs), and ii) wind down the legacy PC business.

Agree with ii), only partially with i) The graphics division will be unsustainable in long-term.

Graphics division is more than Pro and Gaming cards; remember the Kiosk type of stuff. Also, they can't get rid of it now, since the APU business needs to continue growing in terms of GPU power. They can't externalize the design (a la Intel) of core components, specially since they own the IPs. They have to keep'em tight.

Cheers!

 

[juanrga]

AMD rated B minus for long-term scenario

http://www.reuters.com/article/2015/02/02/fitch-affirms-amds-idr-at-b-outlook-stab-idUSFit91263120150202

AMD's Recovery Ratings (RRs) reflect Fitch's belief that the company would be reorganized as a going concern rather than liquidated in a bankruptcy scenario. To arrive at a going concern value, Fitch believes AMD would: i) reorganize businesses serving target markets (Enterprise, Embedded, and Semi-Custom, graphics chips, and APUs), and ii) wind down the legacy PC business.

Agree with ii), only partially with i) The graphics division will be unsustainable in long-term.

Graphics division is more than Pro and Gaming cards; remember the Kiosk type of stuff. Also, they can't get rid of it now, since the APU business needs to continue growing in terms of GPU power. They can't externalize the design (a la Intel) of core components, specially since they own the IPs. They have to keep'em tight.

Cheers!

The kiosk life stuff belongs to the embedded division.

By graphics division we mean discrete cards. The APU division (and the integrated graphics) will have future, as has the Embedded, and Semi-Custom groups.

Which is the future of the AMD discrete cards?

1. GPGPU? Nope, That is dominated by Nvidia and soon will be taken by Intel Phi.
2. PC? Nope. Sales are dropping quarter after quarter at nearly constant rate. Whereas integrated graphics rise.
3. Gaming? Nope. Most gaming (in general sense) is made on integrated graphics. Only a small percentage uses discrete cards and Nvidia is dominating the market
http://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-could-dominate-market-of-desktop-graphics-cards/

AMD seems to want now to push 4K gaming to continue on the discrete card market but there is doubts that will have enough financial return. 4K is a niche of a niche market: about 0.03% of gamers

http://www.fool.com/investing/general/2015/02/02/will-advanced-micro-devices-amd-bet-on-super-fast.aspx

What I lack on AMD is a clever strategy for the future. I see them trying a bit here and another bit there but without any real plan.

 

[-Fran-]

The kiosk life stuff belongs to the embedded division.

By graphics division we mean discrete cards. The APU division (and the integrated graphics) will have future, as has the Embedded, and Semi-Custom groups.

Which is the future of the AMD discrete cards?

1. GPGPU? Nope, That is dominated by Nvidia and soon will be taken by Intel Phi.
2. PC? Nope. Sales are dropping quarter after quarter at nearly constant rate. Whereas integrated graphics rise.
3. Gaming? Nope. Most gaming (in general sense) is made on integrated graphics. Only a small percentage uses discrete cards and Nvidia is dominating the market
http://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-could-dominate-market-of-desktop-graphics-cards/

AMD seems to want now to push 4K gaming to continue on the discrete card market but there is doubts that will have enough financial return. 4K is a niche of a niche market: about 0.03% of gamers

http://www.fool.com/investing/general/2015/02/02/will-advanced-micro-devices-amd-bet-on-super-fast.aspx

What I lack on AMD is a clever strategy for the future. I see them trying a bit here and another bit there but without any real plan.

Right. I forgot about the word "embedded", lol.

I will only put point 2 in debate, since like I told you, I think we might see a jump in processing requirements thanks to Win10 and all the VR stuff. I'm pretty darn sure tablets won't be able to power a full VR experience.

Cheers!

 

[jdwii]

so Amd stock should be in the toilet based on your predictions right ??? looks to me the market has spoken

mid day report
amd= 2.79 up 4.49 %
intel =33.57 dwn 0.24%

Like in chess you can't base the whole game on one move. Look at the overall picture
Intel
http://s15.postimg.org/z3ugo5xwr/Intel.png

Amd
http://s27.postimg.org/40dtyznz7/Amd.png

 

[jdwii]

The kiosk life stuff belongs to the embedded division.

By graphics division we mean discrete cards. The APU division (and the integrated graphics) will have future, as has the Embedded, and Semi-Custom groups.

Which is the future of the AMD discrete cards?

1. GPGPU? Nope, That is dominated by Nvidia and soon will be taken by Intel Phi.
2. PC? Nope. Sales are dropping quarter after quarter at nearly constant rate. Whereas integrated graphics rise.
3. Gaming? Nope. Most gaming (in general sense) is made on integrated graphics. Only a small percentage uses discrete cards and Nvidia is dominating the market
http://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-could-dominate-market-of-desktop-graphics-cards/

AMD seems to want now to push 4K gaming to continue on the discrete card market but there is doubts that will have enough financial return. 4K is a niche of a niche market: about 0.03% of gamers

http://www.fool.com/investing/general/2015/02/02/will-advanced-micro-devices-amd-bet-on-super-fast.aspx

What I lack on AMD is a clever strategy for the future. I see them trying a bit here and another bit there but without any real plan.

Right. I forgot about the word "embedded", lol.

I will only put point 2 in debate, since like I told you, I think we might see a jump in processing requirements thanks to Win10 and all the VR stuff. I'm pretty darn sure tablets won't be able to power a full VR experience.

Cheers!

That VR stuff isn't going to be a non SOC base system. As for windows 10 the OS should have similar requirements as vista from 9 years ago. I do expect to see increases in sales in the laptop/desktop market this year however over it.

Since 10 looks actually good, i'll be getting it day one or sooner using dreamspark

 

[-Fran-]

Careful with the VR stuff we're talking about. In particular, remember the main reason for the Oculus Rift dev length has been the actual hardware behind it. In particular panels (screen) and processing power itself.

I'm not saying you cannot create a VR setup in a modern Smartphone, but a rich experience such as processing images and predicting (look at the video from the tech demo MS showed) can't be run off a Smartphone at all; not even a tablet's performance point. You need a backend for heavy lifting and a small front end (ARM SoC's) to handle image and light processing for display only. There are things the VR headsets themselves can do, sure, but they're not the "whole" experience I'm talking about. Well, I'm basing this on the tech demo from MS (if you haven't seen it, you should).

Cheers!

EDIT: Typo.

 

[-Fran-]

Careful with the VR stuff we're talking about. In particular, remember the main reason for the Oculus Rift dev length has been the actual hardware behind it. In particular panels (screen) and processing power itself.

I'm not saying you cannot create a VR setup in a modern Smartphone, but a rich experience such as processing images and predicting (look at the video from the tech demo MS showed) can't be run off a Smartphone at all; not even a tablet's performance point. You need a backend for heavy lifting and a small front end (ARM SoC's) to handle image and light processing for display only. There are things the VR headsets themselves can do, sure, but they're not the "whole" experience I'm talking about. Well, I'm basing this on the tech demo from MS (if you haven't seen it, you should).

Cheers!

EDIT: Typo.

are you referring to hololens ?

I don't want to engage on semantics, but the "holo" component MS talked about is "virtual reality", since you need a headset / headgear. Holograms are created as solid image with no need for glasses to see it. Marketing BS at it's finest.

In any case, yes.

Cheers!

 

[juanrga]

so Amd stock should be in the toilet based on your predictions right ??? looks to me the market has spoken

mid day report
amd= 2.79 up 4.49 %
intel =33.57 dwn 0.24%

That is irrelevant.

Analysts were waiting poor Q4 results and downgraded the stock days before the data was made public by AMD. The stock did rise latter because AMD numbers were not so bad as analysts expected (but still number were bad). Since then persistent rumors of a chinese company negotiating the buy of AMD has increased the stock artificially some few peanuts. It will return to normal values soon.

You can continue pretending that AMD is in good shape and Intel is full of trouble, it is funny!

 

[jdwii]

so Amd stock should be in the toilet based on your predictions right ??? looks to me the market has spoken

mid day report
amd= 2.79 up 4.49 %
intel =33.57 dwn 0.24%

Like in chess you can't base the whole game on one move. Look at the overall picture
Intel
http://s15.postimg.org/z3ugo5xwr/Intel.png

Amd
http://s27.postimg.org/40dtyznz7/Amd.png

I doubt he was predicating amds downfall back in 2011 but lets look at dec 8 when this thread was started

amd close dec 8 = 2.58 feb 3=2.76
intel close dec 8 =37.20 feb 3= 33.70

which one is the better performing stock ?

I honesty wouldn't recommend either, id rather put money in qualcomm or Nvidia both of which have went up well in the last few years.