Intel's MWC announcements...disappointing

[blckgrffn]

Intel is Doomed.

Clearly.

They are getting serious in the Intel way. Slowly, but then with overwhelming force and engineering spend.

At least, that seems to be their past pattern.

Unless the margins aren't good enough/future of a segment looks dim. Then they sell it off or go into a different market.

A dumb way to run a business, right?

 

[LogOver]

With each iteration ARM cpus are getting hotter and hotter. Not a single A15 based phone was announced during MWC (and is not expected in the near future). Even SGS-IV is going to use Qualcomm 600 (which is slower then Z2580). Intel also announced LTE modem which can be used with CloverTrail+... Not to mention that right now LTE is actual for US only.

 

[meloz]

Intel won't be in serious play for another 12-18 months. 22 nm Atom should give them a keep-alive. It is what comes after Silvermont and Valleyview that should concern everyone.

With 14nm Airmont things should truly get interesting. A lot depends on how the migration to newer nodes works out at TSMC and Samsung. The foundry gap could increase, stay the same or decrease. We'll see.

After mistreating Atom like unloved stepchild Intel are finally throwing everything they have behind it. I think a lot of people will (re)learn in the next 24 months why Intel are called Chipzilla. It is going to be interesting, for sure. Great news for us consumers.

 

[IntelUser2000]

And not to forget: Next year nVidia will have Tegra 4i on the market in real devices. So there is no room for another cpu architecture until Intel gets there modem ready. Until then nVidia and Qualcom have the whole market for them.

In the tablet and low end market A15 will rule everything in the same power consumption room. Tegra 4 is nearly 3x faster than Clover Trail and 33% slower than the IB in the Surface Pro (Geekbench, Sunspider, Kraken).

I don't see a future for ATOM.

You aren't doing yourself a favor by comparing results from WIDELY different OSes and platforms.

Atom Z2460 Android Kraken: 14455.3
Atom Z2760 Windows 8: 33220.9

For Surface Pro, the Chrome result is the accurate one it gets 2492.4, which makes it 2.7x the speed of Tegra 4. Also, Tegra 4 is 2.1x the speed of the Atom Z2460. Surface Pro is more than 3x the speed of Tegra 4 in Sunspider as well. Geekbench can't be compared because Nvidia is claiming that for version 1.0, where the latest Geekbench is version 2. In Sunspider, Tegra 4 is 2.1x the speed of the Z2460, but latest version of Android on the Z2580 reduces that to 1.9x.

 

[krumme]

Intel won't be in serious play for another 12-18 months. 22 nm Atom should give them a keep-alive. It is what comes after Silvermont and Valleyview that should concern everyone.

With 14nm Airmont things should truly get interesting. A lot depends on how the migration to newer nodes works out at TSMC and Samsung. The foundry gap could increase, stay the same or decrease. We'll see.

After mistreating Atom like unloved stepchild Intel are finally throwing everything they have behind it. I think a lot of people will (re)learn in the next 24 months why Intel are called Chipzilla. It is going to be interesting, for sure. Great news for us consumers.

The competition will kick in, we will benefit for sure and that is great. For phones and tablets. I am pretty sure we will see some impressive numbers. The current Atom line is worthless on the market.

Where i have my doubts is, where will it benefit? - how is Atom positioned? What is the market here, and will Intel earn from it?

Intel might pull some extremely good improvements, but will it be enough benefit for the consumer to shift from the dirt cheap arm? Really what is the benefit, and marketing benefit to fx. a quad core A7 at only 2-3mm2 for the cores?
And what is the benefit to fx. a tablet with quad core temash with fx. docking and turbo facility providing speed that aproach IB territory?

I simply ask, who is going to buy the new Atoms?, secondly what is the profit for Intel here?

By far most of the arm stuff is still on 40nm, the need for process improvement does not seem like the most important, as 28nm is available and cheap already. I simply doubt the market value beyond a quad core a7 on 28nm. And when we see some improved A7 versions - think what apple can do - even a15 seem like a huge dog to me for the phones.

 

[PingSpike]

haha this is the same thing that comes into my mind when I see "Atom". Those stupid Dell netbooks that performed like a 486 without the turbo on.

It doesn't feel like Intel has done much to change this perception either. Newer atoms seem to focus on marginal speed improvements while neutering the chipset down to uselessness in an effort to avoid cannibalizing other product lines. The trouble is AMD has some options in that segment for the same price and their own celeron products use about the same amount of power while featuring full featured chipsets and a similar price. It feels like they're trying uninvent the Atom more than anything.

I can't comment on their newer plans for Atom since I'm not up to speed on that. I just know when shopping for a low power platform the Atom seems pointless at the price point it is offered.

 

[podspi]

The competition will kick in, we will benefit for sure and that is great. For phones and tablets. I am pretty sure we will see some impressive numbers. The current Atom line is worthless on the market.

[...]

I simply ask, who is going to buy the new Atoms?, secondly what is the profit for Intel here?.

I'm typing this on a current-gem Atom tablet. Unless Haswell (or competitive AMD chip, if it exists) ends up passively cooled, I will be buying a next-gen Atom tablet. And I bet a lot of other people will too.

Having a light, passively cooled x86 machine is really really awesome. The biggest issue I see so far with the 22nm Atoms is that the brand is pretty damaged.

 

[Maragark]

I'm typing this on a current-gem Atom tablet. Unless Haswell (or competitive AMD chip, if it exists) ends up passively cooled, I will be buying a next-gen Atom tablet. And I bet a lot of other people will too.

Having a light, passively cooled x86 machine is really really awesome. The biggest issue I see so far with the 22nm Atoms is that the brand is pretty damaged.

Why ignore Temash when it will be out in a couple of months and will destroy any Atom or ARM based tablet? That seems rather silly to me.

 

[p4ck3tl055]

486 without the turbo on

OMG TURBO MODE! 33MHz, that was crazy! LOL.

 

[podspi]

Why ignore Temash when it will be out in a couple of months and will destroy any Atom or ARM based tablet? That seems rather silly to me.

I specifically said I'd consider and AMD chip if it was competitive. But Temash does not exist yet, and I still haven't heard anything on whether it supports S0ix. Having used an x86 device that supports this, I honestly wouldn't want to give it up. The boost in battery life is amazing, and not having tto deal with the lags and idiosyncrasies of S3 is nice too.

 

[Gideon]

... whether it supports S0ix. Having used an x86 device that supports this ...

CURRENT gen Atom supports S0ix ?!? AFAIK It's only Haswell and presumably Silvermont.

 

[Maragark]

I specifically said I'd consider and AMD chip if it was competitive.

No, you specifically said, "Unless Haswell (or competitive AMD chip, if it exists) ends up passively cooled, I will be buying a next-gen Atom tablet. And I bet a lot of other people will too."

Temash isn't a Haswell competitor, it's an Atom competitor, both of which will cost a fraction of the price of Haswell or Kaveri.

But Temash does not exist yet, and I still haven't heard anything on whether it supports S0ix. Having used an x86 device that supports this, I honestly wouldn't want to give it up. The boost in battery life is amazing, and not having tto deal with the lags and idiosyncrasies of S3 is nice too.

Temash does exist, and will be released in a couple of months, before Haswell and Silvermont. Also S0ix is nowhere near important in the tablet segment as it is in the phone segment. Phones are primarily communication devices, tablets aren't and don't need to be always connected. As for the boost in battery life, there is no boost in battery life at all when compared to S3, just a couple of seconds saved from having to reconnect.

 

[podspi]

CURRENT gen Atom supports S0ix ?!? AFAIK It's only Haswell and presumably Silvermont.

It does, and it is awesome.

No, you specifically said, "Unless Haswell (or competitive AMD chip, if it exists) ends up passively cooled, I will be buying a next-gen Atom tablet. And I bet a lot of other people will too."

Temash isn't a Haswell competitor, it's an Atom competitor, both of which will cost a fraction of the price of Haswell or Kaveri.

Temash does exist, and will be released in a couple of months, before Haswell and Silvermont. Also S0ix is nowhere near important in the tablet segment as it is in the phone segment. Phones are primarily communication devices, tablets aren't and don't need to be always connected. As for the boost in battery life, there is no boost in battery life at all when compared to S3, just a couple of seconds saved from having to reconnect.

You are being pedantic, I didn't mention Temash by name because I couldn't remember the codename. I said if an AMD chip was competitive and was able to be passively cooled I would consider it.

I think S0ix is very important. First of all, S01x can be active WHILE you are actively using the machine (so there is a boost to battery life), and second, when you have the ability to quickly enter-and-exit low-power modes you tend to do it more often. I was playing with a Surface Pro and while its sleep/wake performance was admirable, you could still tell it was entering and exiting S3, as opposed to my ATIV and every ARM-based tablet/phone where it feels like you are just turning the screen on and off. I have toggled my tablet quickly to check the time before. This is just not something you could do without S0ix. Is it required? Of course not. We survived without it until recently. But in order to get my money it is a feature I want.

Finally, I would argue that Haswell and Atom, and Temash ARE competitors. Certainly they are not direct competitors (very different price brackets) but I've known people who got rid of their computers(!) after buying iPhones. It's hard to argue that computation devices are not competitors to each other, regardless of differences in price and performance. In my case, I want a passively cooled tablet that I can ink on (preferably using wacom technology, but it must be active). I'm not going to buy a Haswell, Temash, and Atom tablet. I am going to buy the one that best suits my needs.

Edit:
And according to my understanding of the current roadmaps, that'll probably be Silvermont.

 

[Maragark]

You are being pedantic, I didn't mention Temash by name because I couldn't remember the codename. I said if an AMD chip was competitive and was able to be passively cooled I would consider it.

I don't think I was being pedantic. You specifically said Haswell competitor which Temash isn't - that's Kaveri.

I think S0ix is very important. First of all, S01x can be active WHILE you are actively using the machine (so there is a boost to battery life), and second, when you have the ability to quickly enter-and-exit low-power modes you tend to do it more often. I was playing with a Surface Pro and while its sleep/wake performance was admirable, you could still tell it was entering and exiting S3, as opposed to my ATIV and every ARM-based tablet/phone where it feels like you are just turning the screen on and off. I have toggled my tablet quickly to check the time before. This is just not something you could do without S0ix. Is it required? Of course not. We survived without it until recently. But in order to get my money it is a feature I want.

S0ix is a nice feature that will probably become a standard feature on x86 eventually, but would you honesty trade greater performance for a reduced time to wake up? I wouldn't make that trade on anything but a smartphone.

Finally, I would argue that Haswell and Atom, and Temash ARE competitors. Certainly they are not direct competitors (very different price brackets) but I've known people who got rid of their computers(!) after buying iPhones. It's hard to argue that computation devices are not competitors to each other, regardless of differences in price and performance. In my case, I want a passively cooled tablet that I can ink on (preferably using wacom technology, but it must be active).

Most people are severely limited by a budget and would in no way see a $1000 tablet as competition to a $300 tablet. The $1000 tablet wouldn't even be considered by most people. Haswell only competes with Atom in the minds of those with more money than sense.

I'm not going to buy a Haswell, Temash, and Atom tablet. I am going to buy the one that best suits my needs.

Well, if your need is to wake the system as quickly as possible, then that rules Temash out. If that's not a need, then it comes down to performance and price. Temash can run Dirt Showdown in 1080p, will Silvermont have such performance? We'll know the price of Temash tablets in a couple of months. Are you expecting Silvermont tablets to be significantly cheaper?

 

[IntelUser2000]

Why ignore Temash when it will be out in a couple of months and will destroy any Atom or ARM based tablet? That seems rather silly to me.

Only time where Temash will "destroy" ARM or Atom is in HEAVY multi-threaded scenarios, or enthusiast gaming. But then if you want to do those, there are better options.

S0iX isn't Connected Standby. S0iX is what ENABLES Connected Standby to happen.

Go look at Anandtech review of the Acer W510 Tablet. 27WHr battery x86 device achieves 8 hour web browsing and 9 hour video playback battery life! Previous Atoms needed 50+Whr batteries to do the same, at lower brightness. That's what the new power states enable. Those that support the new power management and those that don't are in another league altogether.

S0ix is a nice feature that will probably become a standard feature on x86 eventually, but would you honesty trade greater performance for a reduced time to wake up? I wouldn't make that trade on anything but a smartphone.

You are conveniently ignoring the fact that it achieves greatly more battery life. And that's what the "average people" you are talking about cares more about than playing "Dirt Showdown" at 1080p or whatever.

 

[podspi]

I don't think I was being pedantic. You specifically said Haswell competitor which Temash isn't - that's Kaveri.

Ah, I see what you are saying now. I am thinking x86 win-based tablets. I doubt we will see x86 win-based tablets from anybody for $300 any time soon. That being said, in my mind they still do compete to some extent. I guess with Haswell (theoretically) you will no longer have to go low-performing for low-power, but we will see.

S0ix is a nice feature that will probably become a standard feature on x86 eventually, but would you honesty trade greater performance for a reduced time to wake up? I wouldn't make that trade on anything but a smartphone.

As others have already stated, S0ix also enables longer active use.

Most people are severely limited by a budget and would in no way see a $1000 tablet as competition to a $300 tablet. The $1000 tablet wouldn't even be considered by most people. Haswell only competes with Atom in the minds of those with more money than sense.

That's just silly. I think I already made my case for them being indirect competitors. Certainly I considered both the ATIV 500t and 700t when I was trying to decide which x86 tablet to get, and the 700t is much more expensive. Clearly if you are budget constrained (which many people are) Haswell will be out, but especially with convertible/dockable tablets pushing out traditional laptops, I imagine more people will be looking at these expensive tablets as an alternative to a cheap tablet + cheap laptop.

Well, if your need is to wake the system as quickly as possible, then that rules Temash out. If that's not a need, then it comes down to performance and price. Temash can run Dirt Showdown in 1080p, will Silvermont have such performance? We'll know the price of Temash tablets in a couple of months. Are you expecting Silvermont tablets to be significantly cheaper?

I've never played Dirt Showdown. We all know the GPU in the current-gen Atom is shameful, and it does not affect my usage of my tablet AT ALL because I don't use it for gaming. I am not saying it is not impressive, I am not saying it is useless (for some people I am sure it will be a useful feature), but I do not care, and I am sure I am not alone in that apathy. I care most about good enough Office performance, web browsing, and battery life. For everything else, I have a non-power constrained desktop.

 

[Maragark]

You are conveniently ignoring the fact that it achieves greatly more battery life. And that's what the "average people" you are talking about cares more about than playing "Dirt Showdown" at 1080p or whatever.

I'm not ignoring that. We simply don't know yet what Temash's battery life will be like. What we do know is that Temash will be better than Bobcat based APU's in this area according to AMD. Hopefully, we'll get some more details in a day or two.

Also, I didn't say anyone cared about playing Dirt Showdown at 1080p on their tablets, I used that as an example to show how powerful Temash is.

 

[Maragark]

I care most about good enough Office performance, web browsing, and battery life. For everything else, I have a non-power constrained desktop.

Do current tablets not provide that? Would a tablet that can run PC games in 1080p provide good enough performance? Would a 5W SoC not offer better battery life than a 20W APU?

It sounds to me like you're dismissing Temash as being too powerful and power hungry for a tablet, while at the same time considering Haswell as a possible match for your needs. Seems rather strange to me.

 

[podspi]

Do current tablets not provide that?

I would say that current Atom tablets do, but just barely. I'd like some more performance :twisted:. A15 might be even better, but WinRT tablets aren't much cheaper than Atom ones (if at all) so I don't really see the point in them.

Would a tablet that can run PC games in 1080p provide good enough performance?

I imagine that it would.

Would a 5W SoC not offer better battery life than a 20W APU?

Not sure what you are trying to say here. Clover Trail's TDP is less than 2W isn't it?

It sounds to me like you're dismissing Temash as being too powerful and power hungry for a tablet, while at the same time considering Haswell as a possible match for your needs. Seems rather strange to me.

I never said that at all. Here is what I said:

I'm typing this on a current-gem Atom tablet. Unless Haswell (or competitive AMD chip, if it exists) ends up passively cooled, I will be buying a next-gen Atom tablet. And I bet a lot of other people will too.

Having a light, passively cooled x86 machine is really really awesome. The biggest issue I see so far with the 22nm Atoms is that the brand is pretty damaged.

I realize that I didn't mention Temash by name, you're just going to have to get over that. I plainly stated that unless Haswell or whatever AMD ended up fielding (be it Hondo, Temash, Kaveri, a unicorn, whatever) end up passively cooled I'll be buying a next-gen Atom. S0ix came up later, and is also pretty important to me. Now, it sounds to me that Temash-based tablets will not be passively cooled, nor will they support S0ix. For me, that makes it a non-starter. If it helps you sleep at night, it sounds to me like the same could be said for Haswell too...(for tablets)

 

[Maragark]

Not sure what you are trying to say here. Clover Trail's TDP is less than 2W isn't it?

I'll get right to it then. I've been talking about the quad core Temash but there will be a dual core version as well. It seems to me that it would be more up your alley than the quad core.

 

[paulcheung]

Everyone is talking about lower power cpu. How about the battery? Is anyone look into the batteries? Can the quality and higher capacity be improve on the battery in smaller and lighter form factor?

 

[IntelUser2000]

I'm not ignoring that. We simply don't know yet what Temash's battery life will be like. What we do know is that Temash will be better than Bobcat based APU's in this area according to AMD. Hopefully, we'll get some more details in a day or two.

Also, I didn't say anyone cared about playing Dirt Showdown at 1080p on their tablets, I used that as an example to show how powerful Temash is.

Its irrelevant in the big picture. The "new power management scheme" I'm talking about addresses the whole platform, while Temash will do more of the same they did in the previous generation - cut power in the CPU side which is already very very small.

That's why you saw forgettable improvements every generation before(maybe offering 15-30 mins extra out of the current 6 hours for example). But with the "new scheme", Atom is able to get a HUGE improvement.

Again, forget everything you know about "power management advances". Those that support this feature and those that don't will have a unbridgeable gap. The 10-inch Atom Z2760(Clover Trail) Tablet we have in the house uses less power playing 1080p Youtube videos, than a 5-inch Atom MID I have in Idle, despite the fact that they have similar TDP levels.

 

[Sable]

Everyone is talking about lower power cpu. How about the battery? Is anyone look into the batteries? Can the quality and higher capacity be improve on the battery in smaller and lighter form factor?

Err. Intel make CPUs not batteries.

hth

 

[NTMBK]

Again, forget everything you know about "power management advances". Those that support this feature and those that don't will have a unbridgeable gap. The 10-inch Atom Z2760(Clover Trail) Tablet we have in the house uses less power playing 1080p Youtube videos, than a 5-inch Atom MID I have in Idle, despite the fact that they have similar TDP levels.

What was the Atom in that MID- and more importantly, what was its chipset? Those early Atoms are notorious for having desktop chipsets which use more power than the actual processor, whereas Clover Trail is an SoC.

 

[OBLAMA2009]

Err. Intel make CPUs not batteries.

hth

hes right though. in the short term, instead of spending billions designing more efficient cps/cellular chips, they could easily just put in batteries with 2-3 times the capacity, and for not that much higher cost. problem solved