ARM CPUs reaching laptop performance levels?

[Arachnotronic]

ARM is touting a Specint2k score of 1250 for their vaunted A57. Wow. That's as fast as my Pentium M 1.6GHz which gets about 1400.

There's someone in this thread who should be worried about his job...

 

[Ferzerp]

That's as fast as my Pentium M 1.6GHz which gets about 1400

A 9 year old part.

 

[ShintaiDK]

A 9 year old part.

It will be 11 years old when it comes out

 

[Arachnotronic]

My post is obviously dripping with sarcasm.

 

[pelov]

ARM is touting a Specint2k score of 1250 for their vaunted A57. Wow. That's as fast as my Pentium M 1.6GHz which gets about 1400.

There's someone in this thread who should be worried about his job...

The out-of-order design will deliver performance measured at 1250 SpecInt2000 at 1.7 GHz, or about 25-30 percent more umph than today’s 32-bit A15 cores. It can be clocked at up to 2.5 GHz in a 20 nm process, said Mike Filippo, principal design architect for the part.

Under the same TDP envelope of current ARM SoCs on 28nm. Your Pentium M had a high 20s TDP for the CPU alone.

 

[Ferzerp]

My post is obviously dripping with sarcasm.

It's hard to tell these days when there are people out there who genuinely believe that's acceptable (or at least pretend to).

 

[Keysplayr]

The problems with a tablet are no keyboard/mouse, lack of software, poor security because of lack of software, slow response. Personally, I refuse to use google docs, so no spreadsheet or word processing either. I do have a tablet, and I like it for web surfing and watching netflix/low res videos. That is all it does well. It is not even good for e-mail, because the touch interface is so frustrating. Facebook games, you have to be kidding. It wont even play bejewelled blitz, the game is a slideshow.

What about netbooks? Basically a tablet with a keyboard attached.

 

[ShintaiDK]

Under the same TDP envelope of current ARM SoCs on 28nm. Your Pentium M had a high 20s TDP for the CPU alone.

24.5W TDP on 130nm.

 

[pelov]

ARM is dead.

(Reuters) - Apple Inc sold 3 million of its new iPads in the first three days the tablet computers were available, driving optimism for a strong holiday quarter despite intensifying competition.

Samsung's Galaxy S3 has hit another milestone: 30 million sales worldwide, The Verge reports.

Samsung Poland made the announcement via its Twitter and Facebook pages. This comes just two months after the Korean giant announced the S3 had sold 20 million units in its first 100 days, less than six months since the phone went on sale. Not bad going at all.

But the S3 isn't the only Samsung device to be flying off the shelves of late. This week Samsung announced its Galaxy Note 2 has already hit 3 million sales, just weeks after launching. That's equal to Google's Nexus 7. It should be noted though, that this figure represents how many Note 2s Samsung has sold to retailers, rather than to customers. Still, it's impressive nonetheless.

Yep. It's dead.

No reworked Atom architecture to compete with the A15 until Q1 2014 as well.

 

[ShintaiDK]

Yep. It's dead.

ARM is just replacing ARM in your numbers. So not sure what your great point is. Several of your ARM supporters are dying fast. And none of those companies you list cares the slightest about ARM. Both of them would drop ARM in a heartbeat for something better.

Your numbers basicly show how fast ARM products gets obsolete due to its lackbuster performance.

No reworked Atom architecture to compete with the A15 until Q1 2014 as well.

Its gonna go downhill for ARM fast already with Silvermont, Airmont will just seal it for good. ARM is running out of time fast.

 

[pelov]

Its gonna go downhill for ARM fast already with Silvermont, Airmont will just seal it for good.

That's a lot of ifs for 2 years away. Got any performance estimates for Silvermont? Hell, got any performance figures for the current Clover Trail Atoms? Intel has been pretty reluctant to release any benchmarks.

ARM is just replacing ARM in your numbers. So not sure what your great point is. Several of your ARM supporters are dying fast. And none of those companies you list cares the slightest about ARM. Both of them would drop ARM in a heartbeat for something better.

Over the past several years, Apple has spent a sizable chunk of change to bring iDevice chip development in-house. We've long heard rumblings that Apple will use that newfound expertise to design chips for its Mac computers, as well, but the speculation has been largely restricted to rumor sites—until now.

Quoting not one, not two, but three sources with knowledge of Apple's plans, Bloomberg says Apple engineers "have grown confident that the chip designs used for [Apple's] mobile devices will one day be powerful enough to run its desktops and laptops." The transition may not happen for "a few years," Bloomberg adds, but some of Apple's engineers reportedly consider the switch "inevitable."

Ooh.

Unification between Macs and iDevices may be in the cards on the software front, too. One of Bloomberg's sources notes that Craig Federighi, who's been overseeing both OS X and iOS since Scott Forstall's ouster, is "considered likely to push for this more integrated experience [between OS X and iOS]." One of the sources is also paraphrased as saying, "If [Apple CEO] Tim Cook wants to offer the consumer of 2017 and beyond a seamless experience on laptops, phones, tablets and televisions, it will be easier to build if all the devices have a consistent underlying chip architecture."

Meanwhile, Microsoft already offers a version of Windows for ARM, which powers the Surface and other tablets, and it's working on support for ARM's upcoming 64-bit architecture. And of course, AMD is cooking up ARM-based Opterons. It sure seems that Intel's dominance in the PC market will be challenged more and more strongly over the coming years. I think this is going to be an interesting decade.

http://techreport.com/news/23849/bloomberg-apple-eyes-homebrewed-chips-for-macs

Apple seems confident as does Qualcomm. Amazon too. Don't forget Google.

 

[Ferzerp]

If ARM seriously looked to be a threat, barring being blocked by regulatory bodies, Intel could purchase ARM with something like a year of its profits.

Gives you perspective, huh?

 

[Ferzerp]

That's a lot of ifs for 2 years away. Got any performance estimates for Silvermont? Hell, got any performance figures for the current Clover Trail Atoms? Intel has been pretty reluctant to release any benchmarks.





http://techreport.com/news/23849/bloomberg-apple-eyes-homebrewed-chips-for-macs

Apple seems confident as does Qualcomm. Amazon too. Don't forget Google.

"Will one day" is very vague.

Apple had to switch to Intel partially due to performance in 2006. State of the art arm is somewhere like Intel was in 2000-2002 currently. In what world is a 4 year step back from what Apple needed 6 years ago suddenly even worth discussion?

Oh yeah, magical arm pixie dust. I keep forgetting.

 

[pelov]

If ARM seriously looked to be a threat, barring being blocked by regulatory bodies, Intel could purchase ARM with something like a year of its profits.

Gives you perspective, huh?

The same can be said for Apple buying Intel. Or Samsung. But, no, Intel wouldn't be able to purchase ARM and nor would any other corporation. They're small because that's their business model.

Apple had to switch to Intel partially due to performance in 2006. State of the art arm is somewhere like Intel was in 2000-2002 currently. In what world is a 4 year step back from what Apple needed 6 years ago suddenly even worth discussion?

It also signifies that Apple isn't tied to any particular ISA.

I'm not delusional. I'm well aware ARM is nowhere near in the performance segment for laptops/desktops, but they're also well ahead in smartphones and tablets. It isn't the laptop/desktop sales that are booming, and ultimately it's money that talks, not the enthusiast. It's going to take several years for them to step into big boy shoes but that doesn't mean they won't be able to. Like I said, Apple and plenty of other corporations seem pretty confident that they will. Pricing still favors them and so does performance as well as perf-per-watt.

 

[Ferzerp]

But, no, Intel wouldn't be able to purchase ARM

ARM Holdings is a publicly traded company and as such is, by definition, always vulnerable to take-over.

 

[pelov]

ARM Holdings is a publicly traded company and as such is, by definition, always vulnerable to take-over.

If you honestly think it's that simple then you're the one that's delusional.

 

[Ferzerp]

Oh, I forgot that the pixie dust bars that.

 

[pelov]

Oh, I forgot that the pixie dust bars that.

Yep. It wouldn't harm the entire ARM ecosystem. I'm pretty sure their business model wouldn't be affected either. Samsung and Apple and all of the other ARM licensees would absolutely LOVE that to happen and would throw even more money at them.

You're either trolling or you have absolutely no idea how ARM works.

 

[Ferzerp]

While I am taking your bizarre claims to an absurd end, this will illustrate, potentially, the absurdity of your claims.

What you seem to envision, is similar to you having a say in another company purchasing your grocery store. Pixie dust doesn't change that.

Though yes, and aquisition would have to be defensive, because that's an awful lot of money for under $200M/quarter revenue.

edit: perhaps you are confused because there are different ARM implementations, that arm is some open standard. It is not, it is proprietary and licensed from ARM Holdings. The IP licensing terms are pretty permissive, and allow for customization.

 

[ShintaiDK]

That's a lot of ifs for 2 years away. Got any performance estimates for Silvermont? Hell, got any performance figures for the current Clover Trail Atoms? Intel has been pretty reluctant to release any benchmarks.





http://techreport.com/news/23849/bloomberg-apple-eyes-homebrewed-chips-for-macs

Apple seems confident as does Qualcomm. Amazon too. Don't forget Google.

Already performance numbers in another thread. Airmont=Phenom X6 performance roughly. ARM=Pentium-M? LoL....

And the Apple will use ARM in laptops rumour is seen before:

http://forums.anandtech.com/showpost.php?p=34199246&postcount=25

ARM, AMD, always some random rumour.

 

[pelov]

While I am taking your bizarre claims to an absurd end, this will illustrate, potentially, the absurdity of your claims.

That's called a straw man. Sort of like the way both you and ShintaiDK imply that I said ARM performance will trump x86 in a short period of time -- something that I've always said would take years but because it helps you win an argument by creating something I've never said and attributing it to me, it gives you something to write. Unfortunately for the both of you, I'm not anywhere near as stupid as you think I am.

Though yes, and aquisition would have to be defensive, because that's an awful lot of money for under $200M/quarter revenue.

Throw in nearly all of Apple's revenue along with that then ask yourself again how much we're talking per quarter. How many Intels is that with just Apple and Samsung? Then consider Qualcomm, TI, Broadcom, Marvell, and on and on and on...

They wouldn't license IP from ARM if they were bought out. The mere fact that you're implying that's a possibility means you don't understand how ARM works as a business model. Though that's pretty obvious by the $200mil quarter revenue you just posted.

Already performance numbers in another thread. And the Apple will use ARM in laptops rumour is seen before:

http://forums.anandtech.com/showpost...6&postcount=25

ARM, AMD, always some random rumour.

I was referring to the performance estimates for future Atoms and even the current Clover Trail atoms. Intel has been very un-Intel-like in that they've been extremely tight-lipped about how it performs. You're throwing around performance estimates as if you've got figures, so let's see them.

 

[Ferzerp]

There are certain topics that people are such fanboys of that any conversation on that topic with them is worthless.

pelov, you are an arm fanboy to the point of utter irrationality.

 

[ShintaiDK]

I was referring to the performance estimates for future Atoms and even the current Clover Trail atoms. Intel has been very un-Intel-like in that they've been extremely tight-lipped about how it performs. You're throwing around performance estimates as if you've got figures, so let's see them.

So quickly you forget in threads you post in.

http://forums.anandtech.com/showthread.php?t=2279950&page=2

 

[pelov]

pelov, you are an arm fanboy to the point of utter irrationality.

I'm not an ARM fanboy. I just want competition. Whether that's ARM, MIPS, x86 with AMD and VIA or PowerPC, I don't care. What I won't do, though, is deny reality.

http://www.google.com/finance?client=ob&q=NASDAQ:ARMH

 

[Fox5]

Under the same TDP envelope of current ARM SoCs on 28nm. Your Pentium M had a high 20s TDP for the CPU alone.

1.3Ghz Pentium M ULV had a TDP of 5.5W. 1.5Ghz LV Pentium M had a TDP of 7.5W.

Atom has just reached those TDP levels with that level of performance only recently.