ARM CPUs reaching laptop performance levels?

[Cerb]

Considering that old computers are quite capable of running recent OSes and software, implying that hardware requirement for OS and software is kind of slowing down

Really, Vista was the one and only aberration, that required nice hardware to run. Most prior OSes, and Windows 7, worked great on slower computers right out of the box. XP, FI, ran just fine on <500MHz PII, PIII, and K6 CPUs, IME, and ran OK with as little as 192MB RAM for the first couple years. Windows 7 runs fine on P4 and Athlon XP machines, made 7-8 years prior to it.

Take all the OEM crapware off, and Vista is the only MS OS of any note that runs poorly out of the box on slower systems that can run it. MS has really gotten a lot of flak over the years, that has mostly deserved to directed towards the big PC vendors, their software bundling deals, and their own resident support software.

Considering the convertible tablet devices that we see appearing on the market, like the Asus transformer and the Windows Surface

Convertible tablets have been around for a long time...they've just been expensive. Fujitsu's Lifebooks were $1800+, FI, just a few years ago.

How much time do you think, until ARM processors reach a performance level high enough that we will see phones and tablets being dockable and turn into full-fledged computers, if it ever happens?

When what they're plugging into doesn't suck, and they can actually handle it. For a notebook replacement, we'd need not only a good CPU, but:
1. A GPU capable of an external output with reasonable performance (Apple, NV, and Qualcomm certainly understand this, Apple is there, and the other two will be shortly, I'm sure).
2. A memory bus of decent width. Right now, Apple is the only major one out there, but I'm sure others will follow. I mean, 32-bit at full speed with LP-DDR3 is still, at best, half of what it's going to take for a few cores and a GPU (Qualcomm has dual-channel, but with slow memory that's only a so-so option).
3. More RAM and/or a standard memory slot.
4. More/better external I/O support.

It's all coming, but the totality could still be a few years off, unless Apple makes one. Today, a separate dedicated device still makes much more sense, and I doubt that will change over the next 3-5 years (5-10 years, who knows). The CPU cores are very close, but there are still a few major integration hurdles yet to cross.

ARM just said they had a processor that's 3x more powerful than their current ones by 2014.

They're catching up quickly.
http://www.theverge.com/2012/10/30/3576560/arm-cortex-a57-cortex-a53-cpu-core

How good will it be at racing to idle? Intel will laugh all the way to the bank if ARM requires vendors to use their big.little scheme, instead of real low-power active idle states for the fast CPUs. I'll believe their dimension-free graph with lowering power requirements at much higher speeds when I see it (my bet: we won't actually see it, and they're taking advantage of the graph not stating what exactly the 'less energy' line is measuring).

Alot of stuff is moving to the cloud, so you gotta ask yourself what you really need those local horsepowers for ... gaming? ..

The video card is for gaming.
The rest? Several virtual machines, compiling stuff, and testing stuff. Ditto on my notebook. And, if I ever have to wait more than a second or two for an input response, process are going to start getting whacked.

My question is: what do you want to waste all that network bandwidth and data center power for? "The cloud," may have its uses, but come on! We're getting cell phones with the kind of power we had in nice PCs 8-10 years ago (and remember, people are still buying Atoms), all we need for storage speed is for UHS SD or mSATA to catch on more, and we're starting to get a decent amount of RAM in phone/tablet chips, too. Not, "wow," but it's not like the processors in portable devices are slowing down. They are becoming sufficiently powerful that cloud computing is going to be more useful to the data mining service providers than it will be to the consumer users of those services.

gaming on the PC .. has been predicted dead for a while now right? And judging from all the crappy console ports, well .... Not a future I like at all, but hey, dont shoot the messenger.

Judging from Steam and indie games, while the super-high-production-value games are only rarely gems, I'm not worried. Much like music, the control is shifting away from the big publishers. And, just like music and movies, not being able to afford to make a blockbuster/chart-topper doesn't mean somebody can't make a damn good piece of work.

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.

Not yet. I happen to have the luck of having some computers right here with P-M 1.2Ghz CPUs in them (convertible tablets, at that), and the new Atoms are still inferior in responsiveness, no matter what some loopy benchmarks have to say about it, even with more and faster RAM and larger and faster HDDs. That said, I have no doubt that the Atom refresh will be big leap in performance, and would be quite surprised if the new OOO Atoms don't reach at least 1st-gen Core performance.

 

[podspi]

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.

The Pentium-M was pretty awesome. I remember the first time I saw one, it was like magic compared to the Pentium-4M based behemoths and AthlonXP that were common at the time.

Even so, the inability to idle like modern chips today can, and the lack of power-efficiency in the rest of the platform led to battery life that wouldn't be impressive today, similar to my Atom machine.


Edit:

Really, Vista was the one and only aberration, that required nice hardware to run. Most prior OSes, and Windows 7, worked great on slower computers right out of the box. XP, FI, ran just fine on <500MHz PII, PIII, and K6 CPUs, IME, and ran OK with as little as 192MB RAM for the first couple years. Windows 7 runs fine on P4 and Athlon XP machines, made 7-8 years prior to it.

Take all the OEM crapware off, and Vista is the only MS OS of any note that runs poorly out of the box on slower systems that can run it. MS has really gotten a lot of flak over the years, that has mostly deserved to directed towards the big PC vendors, their software bundling deals, and their own resident support software.

Convertible tablets have been around for a long time...they've just been expensive. Fujitsu's Lifebooks were $1800+, FI, just a few years ago.

I own a convertible tablet WITH Vista, so I feel like an expert on this . Vista was not as bad as everybody makes it out to be (especially after a service pack or two), it got a lot of flak because drivers just weren't ready for it, and it represented a big change (many/most XP drivers didn't work). XP actually got a lot of similar flak back in the day, as IIRC 95/98/ME drivers didn't work at all with XP. It's just that XP lasted so long everybody forgot about its growing pains.

As for the convertible tablets, the tablets coming out today are NOT the same convertible tablets they came out with back in the day. These things get much better battery life, are much more portable, and way cooler. That being said, most of them are also much smaller and lack proper inking support, and so are also mostly useless (but cooler, so will probably do ok).

 

[frozentundra123456]

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

Truthfully, there are times I wish I had gotten a netbook or something like the HP dm1z instead of the tablet. And I recently saw some netbook with a celeron that looked like it might be nice and have better performance than atom. The tablet even has only about 4 hours of battery life, so I am sure a netbook could top that.

 

[Idontcare]

Welcome to the forums Bambew :thumbsup:

I think it is inevitable that someday the performance capabilities of ARM products will deliver enough computing power to be good enough for a significant portion of the market.

I don't think that time is right around the corner though, personally I think we are looking at a 7-10yr timeline for this transition to really take effect.

 

[Ferzerp]

good enough for a significant portion of the market

My question is: Is that signicant portion of the market the same portion that can (and already are) getting by with a tablet?

 

[Cerb]

I own a convertible tablet WITH Vista, so I feel like an expert on this . Vista was not as bad as everybody makes it out to be (especially after a service pack or two), it got a lot of flak because drivers just weren't ready for it, and it represented a big change (many/most XP drivers didn't work). XP actually got a lot of similar flak back in the day, as IIRC 95/98/ME drivers didn't work at all with XP. It's just that XP lasted so long everybody forgot about its growing pains.

Trying to make a DOS-based OS driver work in NT is what I'd call a PEBKAC.

ATI's driver problems were the same as before, and everybody else's drivers for pretty much everything but printers and scanners (which are still problematic, sometimes) worked great. Maybe not on release, but by the time it has been out for 6-12 months, everything pretty much just worked. Windows 7's wide swatch of beta users was very much an exception, so XP and Vista had to deal with final testing being done by early buyers, so a few configs ended up needing some bug fixes. C'est la vie. If you kept your PC up to date, all would have been well within months, usually. Even so, it wasn't very common, when it did happen.

I still get to deal with Vista. Out of the box, it is a pig, compared to 7. Windows 7 runs OK in 1GB, better with 2GB or more. Vista, today is just plain pokey with <2GB RAM, on a fresh install, compared to 7. The main problem was mismanagement by MS. Instead of getting more service packs to make it all better, we got Windows 7, which is really Vista done right . You can tune Vista to be leaner, but it's easier to just use 7, and 7 has a much more streamlined and useful GUI, anyway.

The driver problems were severely overblown, just like with XP. They were minor and rare, and so were XP's. Seriously, I was using Vista from the beta, and it was almost reliable enough for daily use, if it weren't for being annoying at every turn*. Some companies might have had a stick up their butt about it, having gotten complacent with ME/2K/XP combo drivers, but most were rock solid well before the release. Multiple displays were kind of iffy for a bit, but that's all I can recall having any issues with, and that was well prior to release versions. By release, it seemed to only be oddball peripherals that had real issues (as I'm giving the evil eye to a fairly new HP printer, giving me grief in Windows 7**). I didn't like it because of the numerous irritating UI bits, but it was reliable enough.

DOS-based Windows was never as reliable as Vista was out of the gate. IMO, that is, once again, big vendors choosing poor hardware and poor software, and MS getting blamed for it, since they didn't have the necessary control to stop it (I don't like Vista, but it's not like everything was terrible about it).

An example that I've come across more than once have been those evil Dell-branded Angel TV tuner/capture boxes, which would cause XP/MCE, and Vista, to BSOD, quite reliably. Even ATI was better than that, much less Aver, Pinnacle, Leadtek, Hauppauge, etc.. Heck, I've had no problems with no-name ones like Zyxel. The big problem is that the user goes and blames Vista, being the trendy thing to do, instead of being able to troubleshoot it, and then go get pissed at Dell, who should have either never used those things to begin with, or should have been replacing them with ones that worked with the latest drivers. Meanwhile, those of us who had already learned years ago, were figuring out which Aver, Hauppauge, or Pinnacle model would be the best one for us, and already knew better than to get some Dell/HP/Gateway/etc. branded capture device, from prior good and bad experiences . On the software side of things, the heavy HDD activity from McAfee or Norton security software has consistently been a case of designing software to corrupt a user's data, as soon as the power blips for a second.

My question is: Is that signicant portion of the market the same portion that can (and already are) getting by with a tablet?

If so, I would expect it to be a market that will shrivel in short order. How many people are getting by with just a tablet? Using a tablet, sure. Wondering how they ever got along without one, sure. But I don't know anyone who's getting rid of their computers for a tablet, nor who owns a tablet but not a PC.

Whoever these people supposedly doing that are, I haven't met any of them, yet, even among people who just use their computers for email, word processing, and managing family photos. Maybe there's some sampling bias, but my suspicion is that it's a fantasy that managers want to become true, to keep the market growing for more than just a few more years; and who look at slowing PC sales and hope the demand will just fall off a cliff, soon.



* Just today, I got annoyed to no end by the OK dialog box every time I did a safe removal. It's the little things like that that drove us nuts. And then, there are the little things, like right-clicking taskbar icons for a useful menu, Aero Snap, halfway-predictable Windows Explorer view behavior, more sensible network settings arrangements, and so on.

** It was free. If I'm going to spend money on one, it will be a Brother.

 

[Bambew]

Hey, thanks for the warm welcomes! The answers have been very enlightening so far.

The performance level I am referring to would be the kind that allows casual gaming (like indies or sim games like sims or spore), browsing with less than 50 tabs open, some light Photoshop, things like that. So basically, maybe something that would be around the level of a 2008 mainstream market laptop. In terms of portions of the market, it would include people who currently really can't get by with a tablet because of the obvious usability issue that comes with not having a keyboard, a trackpad or a mouse.

Side note : My step dad still uses a 2005 laptop with XP on it. The thing is a brick.

 

[Idontcare]

Couldn't help but think of this when I saw tablets coming up in the discussion.

 

[CTho9305]

http://www.anandtech.com/show/6422/samsung-chromebook-xe303-review-testing-arms-cortex-a15/5

Hiccups and lag is something that's SUPER noticable.



Really? Youtube? So yes, maybe 98% of the population will notice but not care because they're already used to working on clunky old machines.

I would be interested in seeing what happens to hardware like that when paired with the "polish" Apple brings to software. They had stuff running smoothly on a ~400MHz ARM11, and every iteration since then...

If all you're doing is surfing, using Word, email, facebook type games, then I don't see anyone having any sort of wants if using an ARM CPU.
Now add gaming, video editing, Server applications, databases, and ARM may not be what you want. At least at this moment. There are some pretty good designs using ARM CPU's for server use. And the low power usage is a plus.

Would anyone really want edit video on a CPU in a few years? You can do it faster and with less energy (i.e. battery) with a hardware accelerator. I'm always impressed by what can be done quickly on phones and tablets (I guess developers are forced to use acceleration there, whereas Intel & wall outlets have been letting them be lazy on the desktop).

"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.

I'll join the "call ferzerp delusional" party... in late 2001, Anand was reviewing 1.53GHz single-core Athlon XPs (which, for comparison, outperformed 2GHz P4's). The Chromebook has a dual-core 1.7GHz A15. Now, I know that 1MHz of Athlon XP isn't the same as 1MHz of A15, but I wouldn't be surprised if A15 performs respectably in that comparison (after all, like the Athlon, it's a wide out of order design, and I think A15's load-store unit is actually more capable of reordering memory accesses than Athlon's). And you get two A15s (it was 2005 before Anand reviewed the first dual-core Athlon 64's).

Plus, the A15 in the Chromebook is a power-constrained design - if they cranked up the voltage, there's probably more performance available. On top of that, desktop CPUs are binned, so if only 10% of parts can reach 1.53GHz, they'll sell some at that speed; mobile parts tend not be be binned, so they all sell at a low enough speed that 90% (or more) of the parts meet the target (i.e. they sell everything at what would be Intel / AMD's slowest bin, or 1.33GHz in that Anandtech article from late 2001). Oh, and the Chromebook costs as much for the whole laptop as the 1.53GHz CPU did.

If anybody tried to build a power-unconstrained ARM that could cost what standalone CPUs cost, I think the picture would look very different.

edit: Don't some of the AMD Llano parts run at just 1.9GHz? And there's a Trinity part at 1.9GHz (although with a much higher boost)?

 

[Ferzerp]

I know that 1MHz of Athlon XP isn't the same as 1MHz of A15

You concede that comparing clock speed is meaningless, and then you cite clock speed over and over as "proof" of performance?

No, ARM is a decade behind x86 in performance, regardless of what you want to believe. The way you feel that processors perform based on no proof (or even experience) really doesn't affect reality.

 

[dguy6789]

It's likely the Snapdragon S4 and Cortex A15 are at least as fast as the Athlon XP clock per clock

 

[AdamK47]

defn. doable, plus, allthough i am still hoping for the cloud as a concept to "pass", i guess i must brace myself for it to stay (shudder, hate it.).
Consider services like

http://www.cloudcompiling.com/

Alot of stuff is moving to the cloud, so you gotta ask yourself what you really need those local horsepowers for ... gaming? ..
gaming on the PC .. has been predicted dead for a while now right? And judging from all the crappy console ports, well .... Not a future I like at all, but hey, dont shoot the messenger.

The death of PC gaming has been predicted to be dead many times over the past decade or so. The PC is dead, long live the PC!

 

[Idontcare]

The death of PC gaming has been predicted to be dead many times over the past decade or so. The PC is dead, long live the PC!

I realized that PC gaming really was dead when I found myself updating Oblivion with unofficial bug patches and texture packs provided 100% by unaffiliated enthusiasts just to make the game enjoyably playable on my desktop PC.

That was when it hit me that the PC gaming model had devolved into being analogous to that of Linux, which in the consumer segment gets along pretty much because people volunteer their time to work on it.

Crowd-sourcing is ok, it works, but when an industry finds itself falling back to relying on it as plan A then that pretty much tells you the industry itself is receding into niche territory the likes of which will only persist provided random people pitch in and makes things enjoyable enough for the niche to not entirely die out. (like Linux on the desktop, so too is PC gaming headed IMO)

 

[AdamK47]

I really don't follow the Linux analogy. The majority of what I play on the PC is not created by individuals and funded through their own means. It's still the games made by large developers and funded by big name publishers. The real problem is the number of sales on the PC compared to that of the consoles which limits exclusive titles to the PC. This results in the majority being cross platform titles with little in advantage over their lesser console counterparts. Another problem is development focused on cheaply made casual titles that are viewed as a safe bet to any publisher. The Oblivion example was an exception and definitely not the norm. My choice was/is the opposite, to play it unmodded.

There are still a large amount of big budget games being released for the PC every year. As long as it keeps up I don't see it dying anytime soon. The ones who do see it dying are viewing PC gaming simply through an off-the-cuff surface analysis.

 

[pelov]

I realized that PC gaming really was dead when I found myself updating Oblivion with unofficial bug patches and texture packs provided 100% by unaffiliated enthusiasts just to make the game enjoyably playable on my desktop PC.

That was when it hit me that the PC gaming model had devolved into being analogous to that of Linux, which in the consumer segment gets along pretty much because people volunteer their time to work on it.

Crowd-sourcing is ok, it works, but when an industry finds itself falling back to relying on it as plan A then that pretty much tells you the industry itself is receding into niche territory the likes of which will only persist provided random people pitch in and makes things enjoyable enough for the niche to not entirely die out. (like Linux on the desktop, so too is PC gaming headed IMO)

Although you're right that game studios/developers really don't tend to PC gaming as much as they do to consoles, I think the Oblivion analogy may be a bit off. The Elder Scrolls series has always had a big cult underground modding community. It's a lot like Half-life in that sense.

I'm hoping AMD did manage to get their x86 chips in the next series of consoles. Having modern ISAs and hardware that at least somewhat resembles PC hardware would be a godsend for PC gamers and developers alike.

 

[Blandge]

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.

Intel earned higher profit than Samsung Electronics in 2011 .

 

[2is]

If it ever happens? It already has. The vast majority of consumers are perfectly happy with the computing power offered to them by their cell phone. Enthusiasts like us are the minority.

I would imagine most corporations are like the one I work for. Still running P4's with 512mb of ram and 20gb hard drives.

For simple content consumption maybe and even then, not always. so I wouldn't say "it already has" not by a long shot. A VERY small handful of my clients are running Pentium 4's.

I realized that PC gaming really was dead when I found myself updating Oblivion with unofficial bug patches and texture packs provided 100% by unaffiliated enthusiasts just to make the game enjoyably playable on my desktop PC.

That was when it hit me that the PC gaming model had devolved into being analogous to that of Linux, which in the consumer segment gets along pretty much because people volunteer their time to work on it.

Crowd-sourcing is ok, it works, but when an industry finds itself falling back to relying on it as plan A then that pretty much tells you the industry itself is receding into niche territory the likes of which will only persist provided random people pitch in and makes things enjoyable enough for the niche to not entirely die out. (like Linux on the desktop, so too is PC gaming headed IMO)

Community mods for PC games is hardly a recent thing. If anything, there used to be a LOT more of it. Today, they're often are called DLC's that you pay for.

 

[CTho9305]

I know that 1MHz of Athlon XP isn't the same as 1MHz of A15

You concede that comparing clock speed is meaningless, and then you cite clock speed over and over as "proof" of performance?

No, ARM is a decade behind x86 in performance, regardless of what you want to believe. The way you feel that processors perform based on no proof (or even experience) really doesn't affect reality.

I thought the rest of my post explained why I think what I think...did you read it? You seem to just take my initial statement and say "nope", completely ignoring my reasoning. Do you disagree with any of the points that lead me to think what I think? Do I need to explain any of them more clearly?

 

[wsw1982]

You mean they don't come close for you. The other 98% of the people in the world wouldn't be able to tell the difference.

All the tablets out there right now are basically giant cellphones, and they are replacing Laptop/Desktops for more and more people each day.

As long as you only run android/chrome os/IOS/windows metro apps, it doesn't show any difference. Just like the playstation 1 games show hardly any difference running on PS3 compare to running on PS1

The clover trail is comparable with most of the best ARM offers, and run the metro applications like lighting. However, it become so sluggish even just running chrome browser with more then several taps.

 

[wsw1982]

ARM just said they had a processor that's 3x more powerful than their current ones by 2014.

They're catching up quickly.
http://www.theverge.com/2012/10/30/3576560/arm-cortex-a57-cortex-a53-cpu-core

Yes, ARM claim that ARM also showed the A9 is two times faster than the current ATOM, and claimed ATOM is not possible for smartphone until the 14nm node last year :whiste:.

 

[ShintaiDK]

Talking about Intel and Smartphones.
http://www.4-traders.com/INTEL-CORP...-Mind-with-Motorola-MT788-Intel-Ins-15531639/

Another Atom smartphone for China.