> My tablet today is significantly more powerful than my laptop in 2006 but still not quite fast enough for compiling stuff etc.
What kind of laptop did you own? If you've got an iPad 3, any Tegra 3 or Exynos 4412 powered Android Tablet known to my knowledge you have in no way the processing power of a 2006 Laptop. To remind you, a reasonable Laptop in 2006 had a Core (2) Duo CPU, discreet graphics aside. I've used such a Laptop until 2011 to work flawlessy with Eclipse, Photoshop and even older versions of Premiere.
Performance-wise the current ARM CPUs still are lagging behind anything the Desktop had roughly ten years ago.
It's only their power consumption that made them interesting for Apple and only that only with the uprising of very power efficient gpus aswell so the iPhone 1 could have that level of responsiveness at basically no cpu power.
And it's not a fault of X86, which is shown by the newest Intel Medfield CPUs which are starting to outperform the ARMs at the same level of power consumption. At these levels of engineering it boils down to physics. ARMs CPUs aren't just more "cleverly" designed, they had a different scope, different instruction sets and therefor different power specifications.
You can only get so much CPU Speed from a CPU at a certain NM scale with a certain intructionset at a certain power level, and if it's arm or x86 doesn't change that.
ARM should start to feel Intel's breath in their neck, since ARM doesn't have nearly the kind of spending power Intel has to strongarm(no pun intended) the technology beyond the reach of ARMs financial possibilities(The NM race is becoming exponentially more expensive with each iteration...).
Motorola has started to build the RazrI which will be very interesting to watch in the market.
What is the Razrl? Google search isn't being very helpful.
I haven't done much testing of ARM versus Intel, but the experience that I do have with ARM, compiling software wise, using an i.MX6Q (Freescale), it takes roughly 3 1/2 hours (softfloat) to build Firefox 15.
While the current market of ARM devices won't win any speed tests, there are more features to it than just power consumption. My mx51 netbook has no moving parts and doesn't burn my legs even when its been running a long compile (such as Firefox) versus the Core i7 I have that can build it much faster however it cuts the battery life almost in half. There is also the fan noise when those cores start churning, not to mention the heat.
With your mx51 netbook you've traded power consumption, noise and heat production against performance. If you value the former more than the latter, there is no discussion.
Ah. Okay. I'm reading/responding on my phone and I thought the i was an L.
The intel chip is definitely interesting.
Indeed it depends on what your needs/wants are. There are still quite a few people out who do expect an ARM device to have the same performance as an x86 one.
There are actually 2 different ARM cores that can be licensed, one is power efficient, and one is for performance. As far as I know, "everyone" licenses the power efficient one. I'd be interested in the performance one, but I'm not sure who licenses it. I could see Calxeda licensing it since they are in the server space, but I don't believe they are.
The laptop in question was a 1.6 Ghz Core Solo with 2 Gb of memory and integrated graphics. Even with x86 outperforming ARM at similar clockings etc., a current generation quad core 1.5 Ghz tablet will probably be in a similar performance range as that laptop.
But you are right in that I am comparing a mid range laptop from 2006 with a top of the line tablet in 2012, which is unfair. However, it does not alter the argument much: Tablet/phone processors are getting fast enough for desktop work.
As an aside, it does not really matter if future tablets will run ARM or x86. What matters is that we are talking about slim, light devices that can not afford laptop caliber batteries or active cooling.
Even a 1.6 Ghz Core Solo with 2Gb of memory will outperform a 1.5 ghz quad-core tegra 3 tablet with 1gb of memory(low power ddr3 memory is very expensive and unused ram still consumes power) for about any day-to-day task, software very specifically crafted for the tegra 3 soc aside.
The instruction sets and core design are just not as performance oriented on those ARM chips.
To put insult to injury, especially on mobile there are not many apps that use the multi-core capabilitys of those quad-core cpu's very well, even the Android OS is not very optimised in the sense that modern Desktop OSes are. That is one of the major reasons why a dual-core A15 SOC is faster than a Quad-Core A9... There is no true multi-threading available in browsers, thats why Intels Medfield chips win handsdown all browser benchmarks. Having a fast single-core with hyper threading is obviously going to kick a lot of ass in the browser department.
As far as performance goes, the iPhone 5's A6 is only 10% slower than the June 2004-released [1] Pentium 4 540! High end ARM processors (high end for phones at least) should be more powerful than midrange desktop processors from 2003, or 9 years ago. Unless Geekbench changed since then, which I didn't check for. Laptops with Core Duo processors (circa 2006) were indeed faster than the A6 though.
Geekbench numbers for each:
Pentium 4 540 (eyeballed average from [2]: 1800
iPhone 5's A6 [3]: 1623
What kind of laptop did you own? If you've got an iPad 3, any Tegra 3 or Exynos 4412 powered Android Tablet known to my knowledge you have in no way the processing power of a 2006 Laptop. To remind you, a reasonable Laptop in 2006 had a Core (2) Duo CPU, discreet graphics aside. I've used such a Laptop until 2011 to work flawlessy with Eclipse, Photoshop and even older versions of Premiere.
Performance-wise the current ARM CPUs still are lagging behind anything the Desktop had roughly ten years ago.
It's only their power consumption that made them interesting for Apple and only that only with the uprising of very power efficient gpus aswell so the iPhone 1 could have that level of responsiveness at basically no cpu power.
And it's not a fault of X86, which is shown by the newest Intel Medfield CPUs which are starting to outperform the ARMs at the same level of power consumption. At these levels of engineering it boils down to physics. ARMs CPUs aren't just more "cleverly" designed, they had a different scope, different instruction sets and therefor different power specifications.
You can only get so much CPU Speed from a CPU at a certain NM scale with a certain intructionset at a certain power level, and if it's arm or x86 doesn't change that.
ARM should start to feel Intel's breath in their neck, since ARM doesn't have nearly the kind of spending power Intel has to strongarm(no pun intended) the technology beyond the reach of ARMs financial possibilities(The NM race is becoming exponentially more expensive with each iteration...).
Motorola has started to build the RazrI which will be very interesting to watch in the market.