Intel to Manufacture ARM CPUs for Altera

[Nothingness]

ARM servers didnt move anywhere either. And x86 low power parts beat ARM in performance/watt there.

And Intel took over the mobile market in no time

ARM just started targeting servers, it's way too early to claim it's a failure. Let's wait to see how it evolves.

 

[ShintaiDK]

And Intel took over the mobile market in no time

ARM just started targeting servers, it's way too early to claim it's a failure. Let's wait to see how it evolves.

So with ARM we are supposed to wait and see, give it some years. But with Intel moving into the smartphone/tablet segment, then there is no patience? :whiste:

 

[Nothingness]

So with ARM we are supposed to wait and see, give it some years. But with Intel moving into the smartphone/tablet segment, then there is no patience? :whiste:

Who shows no patience for Intel? Not me. That being said they've been trying for 5 years now and at last they have something almost competitive. So let's give ARM 3-4 more years and let's see how it goes for them

 

[ancientarcher]

Who shows no patience for Intel? Not me. That being said they've been trying for 5 years now and at last they have something almost competitive. So let's give ARM 3-4 more years and let's see how it goes for them

To be fair, ARM 64bit chips aren't out yet. Without 64bit servers were not going to be a viable target anyways.

You can't say the same thing about Intel as Nothingness pointed out - they have been at it in the mobile space for 5 years now. And what is their marketshare??

 

[MisterMac]

To be fair, ARM 64bit chips aren't out yet. Without 64bit servers were not going to be a viable target anyways.

You can't say the same thing about Intel as Nothingness pointed out - they have been at it in the mobile space for 5 years now. And what is their marketshare??

Besides the A7 of course - right?


What's the bloody difference?
Intel hasn't gotten perf\watt within the required envelope until now.
ARM has not developed 64bit capabilities until now - because they've failed to do so in time.
(And still require perf\watt scaling - we've yet to see that in effect - short of a few specific light mass-parallel workloads IN THEORY).

Just like Intel has failed to tweak x86 into a mobile\tablet factor in time.

Same bloody metric.
It's been ATLEAST a few years since ARM said "OMFG SERBAR".

So...apples to apples please

 

[Nothingness]

Just like Intel has failed to tweak x86 into a mobile\tablet factor in time.

Same bloody metric.
It's been ATLEAST a few years since ARM said "OMFG SERBAR".

So...apples to apples please

Yep apples to apples!

Here's some evidence Intel said they'd take over the ultra mobile market back in early 2008: http://blogs.intel.com/technology/2008/04/video_moorestown_the_heart_of/

Now please provide some ARM material from the same period that claim they'll take over server

 

[theeedude]

And Intel took over the mobile market in no time

ARM just started targeting servers, it's way too early to claim it's a failure. Let's wait to see how it evolves.

What's Intel's share in the mobile market?

 

[Arachnotronic]

What's Intel's share in the mobile market?

In tablets, about 5%, in smartphones about 0%.

 

[Nothingness]

In tablets, about 5%, in smartphones about 0%.

And in both cases, it's much higher than ARM market share in servers, be they micro or not

 

[Homeles]

In tablets, about 5%, in smartphones about 0%.

And those numbers will begin to look very favorable for them in the coming months. ARM in regards to servers? Not a chance in hell.

 

[Nothingness]

And those numbers will begin to look very favorable for them in the coming months.

I doubt Intel have any chance in smartphones in the coming months as we've yet to hear of big design win. So it's likely it will be similar to what happened with previous core during 2012. Perhaps with the 14nm part?

For tablets, it seems different.

ARM in regards to servers? Not a chance in hell.

What makes you think so?

 

[Ajay]

ARM servers didnt move anywhere either. And x86 low power parts beat ARM in performance/watt there.

Well, they have some high profile wins in Google, Facebook and Yahoo. I don't see them making a dent in traditional industries, but they could become big in Internet companies, especially when ARMv8 solutions are available. The OSes are already there (Linux variants), but the tools are still developing and changing (for example - Node.js). Of course, these new platforms can run on either architecture.

If ARM doesn't gain a significant foothold in the next five years, it will likely stay a niche product. I don't know what follows Avoton @ 14nm, but it will be interesting to see.

 

[Homeles]

I doubt Intel have any chance in smartphones in the coming months as we've yet to hear of big design win. So it's likely it will be similar to what happened with previous core during 2012. Perhaps with the 14nm part?

Merrifield isn't ready yet. When it is, it will be quite competitive. Probably not the best on the market, but enough to get their landers on the shore. Once they finish porting their transceivers to their own process(es), it's game over.

What makes you think so?

Atom's simply better. There's certainly potential for ARMv8 to spoil Intel's party, but it's only a matter of time before Cherry Trail steals back the spotlight.

 

[Nothingness]

Merrifield isn't ready yet. When it is, it will be quite competitive. Probably not the best on the market, but enough to get their landers on the shore. Once they finish porting their transceivers to their own process(es), it's game over.
Atom's simply better. There's certainly potential for ARMv8 to spoil Intel's party, but it's only a matter of time before Cherry Trail steals back the spotlight.

Oh I see.

 

[theeedude]

It's just around the corner, as usual.

 

[Exophase]

Merrifield isn't ready yet. When it is, it will be quite competitive. Probably not the best on the market, but enough to get their landers on the shore. Once they finish porting their transceivers to their own process(es), it's game over.

Game over means what exactly? They'll own nearly 100% of the market? Or just be the largest single SoC supplier? Are you one of the people who thinks Apple will switch to using them? What's your time scale here? Wanna place some wagers on anything?

Atom's simply better. There's certainly potential for ARMv8 to spoil Intel's party, but it's only a matter of time before Cherry Trail steals back the spotlight.

What does Cherry Trail have to do with servers??

The way I see it, the server market isn't really about Intel vs ARM, it's about Intel vs other companies who build server stuff. Those companies are moving from supplying hardware systems built around third party CPUs and chipsets to full custom/integrated solutions. Given that ARM is just the reasonable choice for CPU in most of these cases.

Intel's Atom-based server parts will be formidable but I don't see them displacing everyone else's market viability (for the low power/cloud based microservers) just because they won't satisfy as many niches and they won't offer the best implementation for everything that they throw on there vs every other specialist with a lot of experience. Having a manufacturing advantage alone doesn't make you the best and most suitable at everything in the world.

 

[Homeles]

It's just around the corner, as usual.

This has to be one of the most overused and baseless arguments commonly thrown around in the tech industry as a whole.

Game over means what exactly? They'll own nearly 100% of the market? Or just be the largest single SoC supplier? Are you one of the people who thinks Apple will switch to using them? What's your time scale here? Wanna place some wagers on anything?

I think Apple switching to them is very unlikely -- they have the cash in hand to make more... effective, more... radical investments. Perhaps they'll buy a country.

I think Intel will become a big player within the next two years. I think Nvidia is going to suffer the most out of the current big contenders, but it's hard to say. Tegra's been underwhelming thus far, and they've been relying on being the first of the big players to market. That didn't work out with Tegra 4, and I think that things will only get worse as time goes on.

What does Cherry Trail have to do with servers??

I should have said Airmont.

The way I see it, the server market isn't really about Intel vs ARM, it's about Intel vs other companies who build server stuff. Those companies are moving from supplying hardware systems built around third party CPUs and chipsets to full custom/integrated solutions. Given that ARM is just the reasonable choice for CPU in most of these cases.

The closed-off nature of x86 is certainly its biggest weakness, and it will be interesting to see what Intel does, if anything (other than what they've already announced to support the industry's desire for custom solutions, which is of questionable use), to address that and keep ARM from infiltrating the low power server market.

Intel's Atom-based server parts will be formidable but I don't see them displacing everyone else's market viability (for the low power/cloud based microservers) just because they won't satisfy as many niches and they won't offer the best implementation for everything that they throw on there vs every other specialist with a lot of experience. Having a manufacturing advantage alone doesn't make you the best and most suitable at everything in the world.

I expect Intel to infiltrate more markets as time goes on. Intel's growing influence on the DRAM market will be fun to watch.

 

[Ajay]

Atom's simply better. There's certainly potential for ARMv8 to spoil Intel's party, but it's only a matter of time before Cherry Trail steals back the spotlight.

Oh come on, Intel is still going with IMG Tech because they are having trouble getting their own iGPU IP integrated into their SoCs comparable performance within their power budget. Intel isn't god. They will succeed in this regard, but how much better will their competition be by then?

Fortunately, most Android apps are written in Google's Java, so no 'golden handcuffs' for ARM on most smart phone apps. This fact, and Intel's process advantage are the only reason Intel has a reasonable chance to get into the majority of high-end phones and tablets. Everyone will stick with ARM for the cheap stuff for some time to come.

 

[theeedude]

This has to be one of the most overused and baseless arguments commonly thrown around in the tech industry as a whole.

Not as overused and baseless as the argument that Intel deciding to enter a market means they are going to succeed and their competitors are finished.

 

[Exophase]

Fortunately, most Android apps are written in Google's Java, so no 'golden handcuffs' for ARM on most smart phone apps. This fact, and Intel's process advantage are the only reason Intel has a reasonable chance to get into the majority of high-end phones and tablets. Everyone will stick with ARM for the cheap stuff for some time to come.

A lot of important software uses NDK, including most apps that are available on both Android and iOS. Fortunately for Intel, most of that would have been targeting x86 in addition to ARM for a while now.

 

[Homeles]

Not as overused and baseless as the argument that Intel deciding to enter a market means they are going to succeed and their competitors are finished.

No one here's actually making that argument, so your quip isn't relevant.

 

[Ajay]

A lot of important software uses NDK, including most apps that are available on both Android and iOS. Fortunately for Intel, most of that would have been targeting x86 in addition to ARM for a while now.

Well, I've been lead to believe that the vast majority of apps are Java and not NDK (Google even says "Before downloading the NDK, you should understand that the NDK will not benefit most apps." - their emphasis).

As far as targeting x86 - what x86 android devices???

 

[Exophase]

Well, I've been lead to believe that the vast majority of apps are Java and not NDK (Google even says "Before downloading the NDK, you should understand that the NDK will not benefit most apps." - their emphasis).

Think about it. You can't do Java on iOS. Having an NDK core is the only real way to share much of a code base between the two.

It's not odd that Google would be dismissive of the NDK because it goes against their portability philosophy and they resisted over doing it in the first place.. but the real truth is that if your app does any heavy lifting anywhere at all it'll probably benefit from NDK. You can see for yourself in the AndEBench comparison how inefficient Dalvik can be. Even if you're not causing a noticeable slowdown on any Android devices at all you're still wasting energy with inefficient software.

As far as targeting x86 - what x86 android devices???

There are plenty of Android devices (RAZR i is perhaps one of the more common) but that isn't really the point; it's the NDK which targets x86. The NDK has had x86 compilation built in for a long time now.

 

[Ajay]

Think about it. You can't do Java on iOS. Having an NDK core is the only real way to share much of a code base between the two.

It's not odd that Google would be dismissive of the NDK because it goes against their portability philosophy and they resisted over doing it in the first place.. but the real truth is that if your app does any heavy lifting anywhere at all it'll probably benefit from NDK. You can see for yourself in the AndEBench comparison how inefficient Dalvik can be. Even if you're not causing a noticeable slowdown on any Android devices at all you're still wasting energy with inefficient software.



There are plenty of Android devices (RAZR i is perhaps one of the more common) but that isn't really the point; it's the NDK which targets x86. The NDK has had x86 compilation built in for a long time now.

OK, so all your generic C/C++ code will work on IOS and Android and then you just write IOS specific stuff in Obj-C? I don't follow smartphone in much detail, so all the phones I've seen have been ARM based. I've read of the woes of Dalvik in developer blogs, seems to work well enough for most consumer apps. My wife owns and iPhone 4S and my brother-in-law totally geeked out on Android from the start. So I've used them enough to be both amazed and totally annoyed with them. I should learn to code for one or the other, as an exercise at the very least, but I've been a bit lazy lately.

I still own a dumb phone, while SoCs are very interesting to me, the phones themselves are less interesting to me, ATM. I'd have to get out my reading glasses to do anything besides making a phone call and I hate using my reading glasses unless I'm reading a book