How will AMD answer the challenge posed by Haswell?

[pelov]

I doubt Apple or Samsung are considering Intel for their next smartphone and I doubt either of them are considering ARM for their next laptop.

They're both making their own products and using their own custom ARM architectures and SoCs.

I think we'll be seeing some ARM laptops soon. The A6 and Snapdragon SoCs are pretty good performers. With the software tailored and optimized at the ultramobile level the way it is, I think it's certainly a possibility in the near future. The issue would be some x86 applications requiring a bit more oomph -- photoshop and other workstation applications like, Maya, Blender, etc. I think part of the performance ailments that ARM does face can be remedied with GPU acceleration. Just take a look at how much faster certain filters in Photoshop are with openCL/CUDA. That sort of performance bump would be a godsend on ARM, and with developers who are making a killing on software with millions of devices sold, you'd have to figure they'll do anything they possibly can to get it increase their potential customer base.

 

[tipoo]

I'm saying at this point in time, I agree that this focus on igpu is silly.

Ah, got ya. I think decent iGPUs are still important for casual users, more PCs capable of running games decently is still a bigger pool for developers and better for everyone, I just wish we had the option not to buy something we aren't going to use, or use that die space for something we would find useful. The GPU is like half the die size or more now, I wish there were models with that dropped off for a lower price, or the same price with higher CPU performance.

 

[2is]

They're both making their own products and using their own custom ARM architectures and SoCs.

I think we'll be seeing some ARM laptops soon. The A6 and Snapdragon SoCs are pretty good performers. With the software tailored and optimized at the ultramobile level the way it is, I think it's certainly a possibility in the near future. The issue would be some x86 applications requiring a bit more oomph -- photoshop and other workstation applications like, Maya, Blender, etc. I think part of the performance ailments that ARM does face can be remedied with GPU acceleration. Just take a look at how much faster certain filters in Photoshop are with openCL/CUDA. That sort of performance bump would be a godsend on ARM, and with developers who are making a killing on software with millions of devices sold, you'd have to figure they'll do anything they possibly can to get it increase their potential customer base.

They are still a long way off from being able to perform common desktop level tasks with decent performance. That's before you take into account issues with running x86 emulation which is bound to happen if arm tries to sneak into the PC market.

 

[pelov]

They are still a long way off from being able to perform common desktop level tasks with decent performance. That's before you take into account issues with running x86 emulation which is bound to happen if arm tries to sneak into the PC market.

You don't exactly need x86 emulation. That would only be with backwards compatible legacy applications which receive little or no software support. What exactly does the average desktop/laptop/tablet user need from x86 that they can't find a viable alternative somewhere in Google Play or iOS?

If you think about how the ARM architecture and ecosystem work, it thrives off of regular updates, frequent recompiling/optimization and just very involved developers that pour through their code to make sure they can squeeze out every little bit of the number crunchers that they possibly can. That's the exact opposite of x86, where we see developers ignore architectural/ISA advancements for 4-5 years at a time, relying instead on the IPC and clock speed bumps that the two big chip makers have provided.

 

[2is]

You don't exactly need x86 emulation. That would only be with backwards compatible legacy applications which receive little or no software support. What exactly does the average desktop/laptop/tablet user need from x86 that they can't find a viable alternative somewhere in Google Play or iOS?

If you think about how the ARM architecture and ecosystem work, it thrives off of regular updates, frequent recompiling/optimization and just very involved developers that pour through their code to make sure they can squeeze out every little bit of the number crunchers that they possibly can. That's the exact opposite of x86, where we see developers ignore architectural/ISA advancements for 4-5 years at a time, relying instead on the IPC and clock speed bumps that the two big chip makers have provided.

Most people, particularly businesses are not going to give up their investment in software to find alternatives that are still in their infancy and essentially guaranteed not to work as well in an effort to move to arm, for what reason exactly? Cloud based office apps dont have the integration with the OS as well as 3rd party programs that many people rely on.

The benefits have to outweigh the costs. As of now, the only benefit is a lower power envelope. Would you buy a ARM based desktop? I know I wouldn't.

 

[Ferzerp]

If you think about how the ARM architecture and ecosystem work, it thrives off of regular updates, frequent recompiling/optimization and just very involved developers that pour through their code to make sure they can squeeze out every little bit of the number crunchers that they possibly can

This is more of that weird ARM pixie dust utopian fantasy. The only reason the speed is acceptible on these small devices is that they are doing substantially less than their desktop/laptop counterparts.

You might make the argument that maybe the normal person doesn't need to do as much, but that's not the argument you are making. Instead, you're pretending that almost all software for arm devices isn't written in a high level language and instead there are legions of assembly savants making fart apps. This just isn't the case, and most of the widely used arm applications are going to be on Andriod (and probably java, *not* known for its efficiency) or iOS (Objective-C, again, a high level language).

Making an email application or a calendaring application run with extremely limited resources, but that is sharing few resources with the rest of the system is simple. Doing the same thing with a full groupware client while also behaving such that every other application is not impacted is an entirely different matter all together.

Your phone or your arm tablet is "fast" because, well, honestly, they aren't doing an awful lot at any given time. I put fast in quotes, because after using various phones, tablets, etc, I tend to get frustrated at waiting on them. Arm has power efficiency at the *extreme* low end. That doesn't magically translate in to anything else. No matter how convenient it might be if it did.

You can make the argument that we try to do too much, and could get by with less, but that doesn't logically lead to arm either. That leads to any underpowered device. x86, arm, mips, whatever.

 

[pelov]

Most people, particularly businesses

Most people are buying ARM-based software, not x86.

particularly businesses are not going to give up their investment in software to find alternatives that are still in their infancy and essentially guaranteed not to work as well in an effort to move to arm, for what reason exactly?

I agree with this, but that's also the crowd that Microsoft has had an iron grip on. Remember that Microsoft, too, is transitioning to ARM with their Metro environment, or at least offering alternatives that were once x86 only -- cloud-based Office and ARM-based Office.

As of now, the only benefit is a lower power envelope. Would you buy a ARM based desktop? I know I wouldn't.

For most users, the benefits are numerous.

- mobility. Something x86 lacks almost entirely and is absent on the tablet/phone end
- massive amounts of applications. Over half a million between Android/iOS by my last count
- They've already bought in. The same "transitioning" that it would take for a business to move away from x86 is also there for the average user, except in the other direction.
- Ease of use. The current x86 environment is scattered and fragmented. Microsoft is attempting to address that by implementing their own "app store" (I'm sure Ballmer will thank Linux for this later) but it currently has a very limited selection of applications. Furthermore, those same applications are split into either x86 Win8/Metro or winRT/Metro.

Would you buy a ARM based desktop? I know I wouldn't.

But would the average person even notice if it was ARM at all? They'd be quite happy that it's the same as their phone and tablet, I think. Like I said before, people aren't complaining that the A6 and Snapdragon SoCs are too slow. In fact, it's just the opposite.

Your phone or your arm tablet is "fast" because, well, honestly, they aren't doing an awful lot at any given time.

There's a difference between content consumption and content creation. An overwhelming majority of people fall into the prior rather than the latter. Thus, most people are quite content with an ARM SoC for nearly all of their computing needs, even in its current state. So much so that they're willing to upgrade much more frequently and spend more money on these devices while skipping 2, 3, 4+ generations on their PCs. Hell, look around the forums and you'll see how many people haven't bothered to update their rigs in years. And why should they? Do you need even more computing power? The answer is "No, but I can use a new gadget."

 

[Ferzerp]

There's a difference between content consumption and content creation. An overwhelming majority of people fall into the prior rather than the latter. Thus, most people are quite content with an ARM SoC for nearly all of their computing needs, even in its current state. So much so that they're willing to upgrade much more frequently and spend more money on these devices while skipping 2, 3, 4+ generations on their PCs. Hell, look around the forums and you'll see how many people haven't bothered to update their rigs in years. And why should they? Do you need even more computing power? The answer is "No, but I can use a new gadget."

If the experience were good enough, why on earth would they need to upgrade all the time?

It seems to me that it's far more likely that the experience is poor, and they hope for improvements on incremental upgrades because the experience is lacking (but is so much better than older phones were, that it is accepted).

That doesn't seem like any utopian vision of computing to me. That looks a lot like throwing money at a bad experience.

 

[2is]

When the average person has Windows and associated applications run at a fraction of the speed, Yes, I think they'd care. If it was all gravy we wouldn't have this conversation, they'd already be out there. Theres nothing magical about ARM, By the time arm is powerful enough to do what intel does its going to consume more power than it does now, so even less of an incentive to move in that direction.

Fragmentation? Serious? ARM market is far more fragmented. You have Apple designing their own custom arm based cores, different manufacturing processes coexisting, muktiple different OS' running on them, multiple different versions of the same OS even (thank android for that) fragmentation is certainly not an arm advantage. The exact oposite actually.

We'll just have to agree to disagree on just about all points. I don't see what you're seeing.

Also, we are talking about ARM transitioning into Intels market. So when i say "most people" I'm talking about what they are doing on their desktops and laptops. The apps being purchased on tablets and smart phones are often times nothing more than a purpose built browser. ARM is great for that. It's not great for what you're proposing it can do. Which is exactly why its used for what its used for and isn't used for what it isn't used for.

 

[pelov]

If the experience were good enough, why on earth would they need to upgrade all the time?

It seems to me that it's far more likely that the experience is poor, and they hope for improvements on incremental upgrades because the experience is lacking (but is so much better than older phones were, that it is accepted).

That doesn't seem like any utopian vision of computing to me. That looks a lot like throwing money at a bad experience.

Except that they're buying them at much higher rates than PCs, where the experience is supposedly "better."

The upgrades are upgrades nearly all around, every single generation. Take for instance the iPhone5. Larger display, bigger battery (though equal battery life), snappier processor + LTE. Same with the Samsung Galaxy S3 or Note. Move up to tablets and you'll find the same thing.

You're thinking people buy phones and tablets for megahertz. They don't care. They don't even know what's inside it. They know it's faster, they know it looks nicer and they know that 4G LTE is faster than 3G.

Fragmentation? Serious? ARM market is far more fragmented. You have Apple designing their own custom arm based cores, muktiple different OS' running on them, multiple different versions of the same OS even (thank android for that) fragmentation ism certainly not an arm advantage.

The software environment. If you're on Win7 and need an application, where are you going to find it? Google? Because that's what most people go for. Do you know it's safe? How? How are you going to pay for it? It is secure? With both Android and iOS they've pulled a feature right from Linux, the "app store" or software center/repositories, and ran with it. Microsoft doesn't have this outside of Metro which is split into two: x86 and ARM.

 

[Ferzerp]

Except that they're buying them at much higher rates than PCs, where the experience is supposedly "better."

The upgrades are upgrades nearly all around, every single generation. Take for instance the iPhone5. Larger display, bigger battery (though equal battery life), snappier processor + LTE. Same with the Samsung Galaxy S3 or Note. Move up to tablets and you'll find the same thing.

You're thinking people buy phones and tablets for megahertz. They don't care. They don't even know what's inside it. They know it's faster, they know it looks nicer and they know that 4G LTE is faster than 3G.

You're even admitting my point without realizing it.

Your argument is that the PC is fast enough, but the phone isn't so people keep upgrading because "they know it's faster". Obviously, there is very much room for improvement there because they notice a speed difference. Then you turn around and say that everyone should discard their "fast enough" pc (again, your opinion, not mine) for something that even in its currently reduced functionality state is not good enough and so people feel the need to upgrade it every year?

I'm sorry, but that's a bit of a delusional fantasy. No, progress is not tossing out decades of advancement because a few tech journalists looking for page hits have talked up arm like it's the second coming if sliced bread. If we wanted to do that, we may as well all just trash our existing PCs and start using old Pentium 3s along with a new set of software. It would be a fundamentally similar experience as what you're advocating.

 

[pelov]

You're even admitting my point without realizing it.

Your argument is that the PC is fast enough, but the phone isn't so people keep upgrading because "they know it's faster". Obviously, there is very much room for improvement there because they notice a speed difference. Then you turn around and say that everyone should discard their "fast enough" pc (again, your opinion, not mine) for something that even in its currently reduced functionality state is not good enough and so people feel the need to upgrade it every year?

I'm sorry, but that's a bit of a delusional fantasy. No, progress is not tossing out decades of advancement because a few tech journalists looking for page hits have talked up arm like it's the second coming if sliced bread. If we wanted to do that, we may as well all just trash our existing PCs and start using old Pentium 3s along with a new set of software. It would be a fundamentally similar experience as what you're advocating.

ARM is going to keep bumping up the performance incrementally because it's looking to take over other emerging markets. Never mind the cutthroat competition at the architectural and performance level that exists there. Can you imagine what we'd be getting from Intel and AMD had they had another 5-6 competitors? and the prices?


They need to keep pushing up the clock speeds and improving IPC because that's what nets them sales.

You're ignoring the reasons I gave for people wanting upgrades, instead only honing in on CPU performance. Here's something else for you to ignore:

Tablet sales grow 250 per cent in 2011 to reach 73 million

In January 2011, Forrester Research released a study that said around 10.3 million tablets were sold in 2010. This number will continue to rise and by 2015, 195 million tablets would have been sold.

On the other hand, Harris Interactive reports that 20% of all American consumers will own a tablet in just 3 years or so.

Apps are going to be more popular than notebooks too, with app sales seen to exceed those of notebooks beginning in 2012.

This means that for companies who are going to develop their own apps, they will have a potential audience of close to 200 million in the next few years. But the good news does not stop there. It seems that there are more opportunities to either earn from creating your own app, or market more effectively using an app.

The reason for this is that Forrester Research sees that tablet users would be more likely to upgrade to the next new model when it becomes available. iPad users soon found themselves wanting an iPad2. This means that if you provide a great app with some cool functionality, there is a good chance that a tablet user would download your app twice just to keep your app in their lives.

Tablets are getting cheaper and cheaper by the minute. Forrester reports that Amazon’s Kindle e-book reader is now two-thirds cheaper than its debut price. Also, with the increased competition, companies that are manufacturing tablets are now offering a lower price than the earlier tablets. The falling prices is what drives the popularity of tablets.

And here's just the last quarter for smartphones

Worldwide, there were 419 million phones sold to end users, is down 2.3% compared to a year ago, Gartner says. Just over one-third (36.7%) of all devices sold were smartphones, which continued to grow well even as the wider market (which includes feature phones) declined. Sales of smartphones were up by 42.7% to 154 million units, with Apple and Samsung together accounting for 83% of all smartphone sales.

Apparently, every single one of them was slow

The PC isn't dying, it just looks a little different.

 

[2is]

The software environment. If you're on Win7 and need an application, where are you going to find it? Google? Because that's what most people go for. Do you know it's safe? How? How are you going to pay for it? It is secure? With both Android and iOS they've pulled a feature right from Linux, the "app store" or software center/repositories, and ran with it. Microsoft doesn't have this outside of Metro which is split into two: x86 and ARM.

The same way I and millions and others have been doing it for years without issue. Mobile platforms have their fair share of security issues, many of which quite severe like collecting location information, contact information, text messaging content. Heck, there is far more user data harvesting and privacy infringement on mobile "arm" platforms than anything x86 right now. I'm not sure were you're getting your info from or what you're basing these opinions on, but you seem to be quite misinformed.

You're all over the place with your arguments. We went from a performance argument to a security argument. Both of which I didn't see a case for. Like I said, you aren't going to convince me, not with these arguments where your opinion is the opposite of reality.

Besides, even if there was some merit to your app store argument, it would be a heck of a lot easier to simply add an "app store" which can be as simple as a Windows Update then to change your entire processing platform to something with a fraction of the power that won't run any of your current software and reinvent the wheel with a crappier wheel.

 

[ericloewe]

The same way I and millions and others have been doing it for years without issue. Mobile platforms have their fair share of security issues, many of which quite severe like collecting location information, contact information, text messaging content. Heck, there is far more user data harvesting and privacy infringement on mobile "arm" platforms than anything x86 right now. I'm not sure were you're getting your info from or what you're basing these opinions on, but you seem to be quite misinformed.

You're all over the place with your arguments. We went from a performance argument to a security argument. Both of which I didn't see a case for. Like I said, you aren't going to convince me, not with these arguments where your opinion is the opposite of reality.

Besides, even if there was some merit to your app store argument, it would be a heck of a lot easier to simply add an "app store" which can be as simple as a Windows Update then to change your entire processing platform to something with a fraction of the power that won't run any of your current software and reinvent the wheel with a crappier wheel.

Pelov seems to live in a fantasy world where ARM will magically take over the desktop segment because it's "less fragmented", which couldn't be farther from the truth.

What is really happening is that there hasn't been a killer app for the PC in while, so most people are happy with their current system. Since they have money to spend (or at least think they do) on gadgets, they buy tablets, partly because everyone has one, partly because they're cool, partly because Apple told them do so.
Anyone who expects to get any real work done on iOS or Android is going to be seriously disappointed, as not even a reputable office suite is available.

 

[NTMBK]

Intel isn't being squeezed by ARM at all, they don't even compete in the same markets. ARM needs to get MUCH more powerful to compete in Intels market and Intel needs to get much more efficient to compete in ARMs market. I doubt Apple or Samsung are considering Intel for their next smartphone and I doubt either of them are considering ARM for their next laptop.

The fear is that tablets and phones could squeeze Intel's laptop and desktop market, not really that ARM would successfully invade it.

I just wish we had the option not to buy something we aren't going to use, or use that die space for something we would find useful. The GPU is like half the die size or more now, I wish there were models with that dropped off for a lower price, or the same price with higher CPU performance.

The thing with that is the cost of creating a new, smaller die would be greater than the savings of fitting more dies onto a wafer, given the limited size of that market. The only people interested in that is the narrow window between everyday users and high end enthusiast/workstation/server chips, in desktops- and that's really quite a small market. For Intel it makes more economic sense to include the iGPU across the mainstream range, and only ditch it on the seriously high end.

 

[ShintaiDK]

If ARM CPUs are fast enough. Why cant newer software run on older? Why do people need a new expensive smartphone every year in a faster upgrade cycle than the PC? Why can my TV get a CPU board upgrade?

The answer is because ARM is so freaking slow. I think most others than pelov noticed

Even browsing websites are horrible on ARM devices. Not to mention ARM CPUs in smartphones and tablets could essentially be gone tomorrow due to no real dependencies. Its not for fun Google got x86 Android.

 

[NTMBK]

If ARM CPUs are fast enough. Why cant newer software run on older? Why do people need a new expensive smartphone every year in a faster upgrade cycle than the PC?

Frankly, that's more to do with phones being a highly visible status symbol, unlike desktops. (See also: Macbooks in Starbucks.)

 

[LogOver]

- mobility. Something x86 lacks almost entirely and is absent on the tablet/phone end

Samsung S3 vs. RAZR i video:
http://www.youtube.com/watch?v=SILUVIwiNF0

Also RAZR i review (battery life is especially interesting):
http://www.theverge.com/2012/10/4/3447984/motorola-razr-i-review-intel-smartphone

 

[pelov]

Pelov seems to live in a fantasy world where ARM will magically take over the desktop segment because it's "less fragmented", which couldn't be farther from the truth.

What is really happening is that there hasn't been a killer app for the PC in while, so most people are happy with their current system. Since they have money to spend (or at least think they do) on gadgets, they buy tablets, partly because everyone has one, partly because they're cool, partly because Apple told them do so.
Anyone who expects to get any real work done on iOS or Android is going to be seriously disappointed, as not even a reputable office suite is available.

Pulling the wraps off a tablet version of Office at the SharePoint conference suggests the company sees the software as an enterprise rather than a consumer product, although there are plenty of consumers interested in a slate edition of the suite. SharePoint is Microsoft's software for collaboration within organizations.

The November time frame for Office for the iPad emerged in May when the Boy Genius Report (BGR) reported that target date and revealed that Microsoft planned to release an Android version of Office at the same time it released the software for the iPad.

Microsoft already has an ARM version of Office on their winRT tablets.

And that's what I mean by uninformed opinions. Never mind that I mentioned it above in my posts that you chose to ignore, though you obviously weren't the only one. Quite a bit of that going on around here...

 

[Nemesis 1]

Intels Medfield has proven all the naysayers wrong. The OoO Smartphone intel has out in 2013 will be the best performing and have great battery life . So many said X86 will never go in smart phones . NOT true as medfield proves . AT 22nm in the smartphone Intel will be the performance and likely the efficiency leader also. Intel coming from top down seems to be working better than ARMS tring to enter notebook space from the bottom up. . Tablets are differant than phones Performance will steal win with tablets

 

[ShintaiDK]

Microsoft already has an ARM version of Office on their winRT tablets.

And that's what I mean by uninformed opinions. Never mind that I mentioned it above in my posts that you chose to ignore, though you obviously weren't the only one. Quite a bit of that going on around here...

You do know how limited that office version is, right? Something again going back to the "limited processing power" of ARM.

Not to mention WinRT is almost stillborn. Almost everyone picked the x86 version for tablets.

 

[pelov]

You do know how limited that office version is, right?

No... but there's also a cloud-based version. So it isn't limited and it's also not limited?

Not to mention WinRT is almost stillborn. Almost everyone picked the x86 version for tablets.

I don't think they'll pick either, frankly. Microsoft is going about it all wrong. The winRT version is cordoned off from the x86 world while the x86 version has nothing ARM-ified.

 

[ShintaiDK]

No... but there's also a cloud-based version. So it isn't limited and it's also not limited?



I don't think they'll pick either, frankly. Microsoft is going about it all wrong. The winRT version is cordoned off from the x86 world while the x86 version has nothing ARM-ified.

You never used cloud did you? Cloud version aka o365 is extremely limited. Not to mention unreliable. Even people just using it for cloud based mails aint exactly happy.

ARM fails because its like trying to sell you an 8-10 year old PC to do todays work. And without all the x86 software, then there is even less reasons to pick ARM.

More and more smartphones are also coming out with Atom now.

 

[NTMBK]

More and more smartphones are also coming out with Atom now.

Motorola Razr I, Intel's reference phone (rebranded by several different countries), and ZTE Grand X IN; 3 hardly counts as "more and more". And can you get any of them in the US yet?

 

[ShintaiDK]

Motorola Razr I, Intel's reference phone (rebranded by several different countries), and ZTE Grand X IN; 3 hardly counts as "more and more". And can you get any of them in the US yet?

I dont know what you can get in the US. You can get Atom phones in Europe and Asia. And Asia is bigger than Americas+Europe combined.