As to why there's not a Android "iPod"? Good question. Apple has sold just as many iPod Touch units as iPhones, which is why combined sales of iPhone and Touch and iPad combind beat out Android combined devices by 59%.
Not really. Even if you count "other related music services", iPods are still half the revenue of iPhones. There were more iPods sold but the profit margin isn't as fat on those as iPhones.
That being said, I think you can clearly say tablets and mobile media players are dominated by Apple. And even if you could dismiss Android's success as "carriers who don't have iPhones provide it", there's still no question that in their most profitable product space (smartphones), Apple is losing out.
That being said, with as much profit and mind-share that Apple gets -- and how huge the smartphone market will grow to be -- even if they own 20% of the market, that's still enough to be a huge wad of cash.
What you're arguing here is that all observations and internal consistency of the human brain is up for question. Which may be technically true but hardly equivalent to belief that scripture describes accurate non-observable phenomenon and history.
Are they both leaps of faith? Sure. The former leap of faith (believing in observable reality) causes me to get up in the morning and eat instead of going "fuck it, none of it is real anyway".
Of course, if that's your definition of "faith", what you're talking about here is the same as faith of existentialism. If you want to really expand faith to that level, the everything except self-existence is a leap of faith. And in that respect, there's no difference between me believing that I'm typing on an actual computer -- interacting with physical matter -- rather than just dreaming all of this.
Do I commit absolute certainty to the fact that others exist in the world? What about history? Do I believe that historical documents describing the actions of Einstein is true? Do I believe that mathematics is consistent and axiomatic? Do I believe that the logic and reason my brain is capable of in any way corresponds with the workings of the universe? Do I believe that any empirical observations made by anyone ever isn't just the result of some invisible, intangible flying spaghetti monster mucking with the results?
We've decided to narrow down the definition of faith with respect to religion to not encompass such a broad topic. Simply: religious faith is narrowed down to absolute certainty of the teachings of some organized scripture.
Said organized scripture may even have scientific or archeological evidence associated with it. But the definition of faith is that, regardless of whether or not anyone can produce any observable, repeatable evidence, one accepts something as truth. And in that respect, science is never like that because every theory comes with the caveat: "this is simply what is consistent with recorded observations". Science relies on faith in the general case the same way that your belief that the reality you observe is real relies on faith in the general case.
Coding? A basic flier requires a simple text editor like Wordpad. Hell, even if you include Office basic, you're still only up ~$80 over the sticker price.
What default software does a Mac come with that makes flyer making so much easier? I don't think it's any of the iStuff.
I'm sure the Apple fans comprise a core part of the sales but look around at an Apple store, most people aren't hardcore iPeople. They walk in, use the product, ask questions and get somewhat informed answers. Even if they don't buy, this gets you into the consumer's brain as a solid buy because you're informed about what you're getting.
Contrast this with a Xoom sitting around at a Best Buy. Nobody's around who knows anything about it. If you ask questions like "how do I purchase music" they can't just do a quick demo using iTunes.
HP might have the clout to fix this as they are used to large-scale sales but they'll have to step up their game of forcing retailers to educate their staff; or perhaps just having dedicated HP salespeople onsite at various retail outlets.
Android has been popular because of both price and the fact that they had the carriers to do the selling for them. This isn't true of tablets.
Why are they considered private property? It seems to me that they resemble real estate far more than they resemble furniture. They are publicly granted exclusive ownership of something that isn't inherently exclusive; similar to the land.
This is actually a brilliant idea. If you want to profit off of the exclusive rights granted to you by public laws, you should pay a tax to support said infrastructure (government) that enforces and protects said exclusive rights.
Treating all IP in every industry the same way is flawed to begin with and the barrier to entry to stake a claim on an idea is way too easy.
The whole point of IP is to encourage innovation by providing incentive for the inventor. Today's IP laws are a flimsy shadow of that. Studios and IP troll companies collect the rewards and inventors are relegated to idea-generating grunts.
There is no inherent morality to ownership of an idea; it is something granted by the public to individual holders of IP for the benefit of the public. If at any time, said laws are detrimental to the public, it should be repealed.
Of course, that would require a government that isn't bought and paid for and a populace that's at least decently educated and informed.
My guess is you're not an EE either. Interference doesn't quite work like that; orthogonal frequencies, for instance, do not interfere at all even when one is extremely high-powered.
Most electronic designers are competent enough to put a choke at their power line and a bandpass filter at their cabling. It's not "easy" but it's done in just about any military grade electronics. I guess Boeing engineers didn't think it was necessary.
I understand the need to build up hype but as others have said, doing it 6 months ahead of time will just kill your momentum. Announce things when they're ready.
This is a problem with or without graphene. Chip design usually doesn't have signals travel all the way across the chip. In fact, good place and route engineers will keep signal nodes as close as possible and logic designers will try to keep high-connection nodes to a minimum, separating out logical clusters to be as isolated as possible.
We're not talking "a bit high". ECL logic is *insanely* power hungry. If we implemented all of a processor's logic in source-coupled logic, it would consume 10-100 times more power than it does currently.
Some decades ago, people thought the GUI was insufficient for "serious" productivity as well. It still may be true today for a small subset of people (programmers and engineers mainly). But for the most part, computers are GUI based.
The problem with your argument is your assumption that "serious" gaming, productivity, etc. is the end-all, be-all of use-cases for the computer. It isn't even the majority. The vast majority of people use their PC for some light document editing, spreadsheets maybe, casual games (more Angry Birds, less WoW) and zero programming.
"The Death of the PC" doesn't mean there will be no PC's around. It just means they'll be relegated to what mainframes, unix terminals and shellscripts are today: tools only used by people who know what they are and how to use them.
The market will shift from selling PC's to selling tablets and netbooks. With the modern PC taking on the role of a home server hub. You will definitely see less families with 3+ laptops -- one for each person. Likely the average home will have one desktop box for storage on the network and multiple tablets with docking stations that enable mouse+keyboard input.
Catholicism also uses 14 bit signed numbers to keep track of years in history. Consequently, they think that 8192 B.C. is the beginning of time and refuse to believe otherwise.
That is my argument exactly, but not in the way you envision it. If you live in New York City, other than a cool factor, what exactly is all of that extra horsepower in a Ferrari going to get you? There is no place to use the extra HP. So, it may be super efficient at 100mph compared to the pinto, but there is nowhere to use it (legally) as such. So, how is it's efficiency at the 25mph to 40mph (or less) that is common in the big city.
For any given performance level, a dual-core solution -- within the vicinity of today's smartphone performance level -- will be more power efficient than a single-core.
Your argument isn't "a pinto is more efficient at 25mph", it's "a pinto going 25mph uses less gas than a Ferrari at 100mph". Again, no shit.
In this analogy, the Ferrari going at 25mph will use less gas than the Pinto.
Likewise, a dual core is more efficient in raw computing power than a single core.
No, a dual-core is more efficient at any level of computing power -- barring extreme cases such as a wristwatch. If you have to reach performance level x, and you have two choices: either make a dual-core solution that can achieve x or a single-core solution that can achieve x -- again, we're not talking about a case where the dual-core is more powerful, just the same level of performance -- you are much better off power-wise to go dual-core.
Or, to put it simply, if you take today's 1GHz SoC's and instead, put in two 500MHz processors -- again, with the caveat that this will only apply to parallel processes -- the two 500MHz processors will be more power efficient even if it's not any faster than the single core 1GHz CPU.
If again you want to use circular logic and say "well a 500MHz single-core will use less power", a dual 250MHz dual-core would have been a better choice for power efficiency.
The point I am trying to make is that just about any power saving measure you take with a dual core can also be implemented on a single core.
Your point completely ignores performance. Of course if you lower the performance expectation, you can get lower power.
If you lower the voltage to a single core and it will use less power. Now, at some point, you get to the point were the single processor can't get any real work done (becomes less responsive from the user perspective) running at lower voltage, whereas two cores running at such low voltage still could (of course chances are they'd both be "on" most the time if the voltage was low enough).
And if you stuck a Cortex M3 in there instead of a Cortex A8, you can save even more power! But your device will be slow as balls.
You're essentially arguing that if you're willing to accept lower performance, you can get lower power. Well no shit.
The point is, for the same processing power -- to accomplish the same task in an equal amount of time -- a dual-core solution is more power efficient than a single-core solution. This of course, has the caveat that the task must be parallelizable and that the software is smart enough to control the hardware efficiently.
What exactly would we be adding to a smart phone that would need the extra computing power of multiple cores? It's not like many people are going to be doing cad/cam on them.
As I've pointed out, even if you wanted to just maintain the performance level of today's single-core chips, a dual-core implementation is more efficient; you can accomplish the same performance using lower voltage cores.
We've already seen this in the netbook market. While desktops and high end laptops kept getting faster and faster processors, the processor in most netbooks is lower voltage and frequency to a) extend battery life and b) their intended use does not warrant higher processing power.
You forgot c) run annoyingly slow for more complex websites and anything with flash or high-def video.
Netbooks are a bad example. But again, we're not talking about the ramifications of increasing the performance wall. We're talking about maintaining the same performance at lower power, which dual-core solutions provide over single-core.
Using the car analogy, again, you can make an eight cylinder engine which is pretty fuel efficient by doing things that turn off cylinders, change compression, etc. But, no manufacture would do so and stick one in, something like a prius or even a smart car. Instead, they use a smaller, less powerful engine that is optimized for the task at hand.
Using the car analogy, if you can get the same torque and horsepower out of an 8-cylinder but use less gas compared to the 4-cylinder, why wouldn't you.
Your argument essentially boils down to "a pinto would use less gas than a Ferrari". Well no shit.
I'd say quite the opposite really. The CPU's are good at integer and branchy type stuff that the GPU can't do anyway. But the A9 implementation they chose is a really really really stripped down FPU.
The GPU is what's pulling the weight in graphics related tasks.
And it was destroyed 10-to-1 by Atom on CPU-bound floating point (Whetstone, LINPACK). Of course A8 will be a simpler more power efficient design when it leaves off floating point.
The A8 and A9, yes. However, you can look at Scorpion and see that it's possible to match and/or beat Atom in FP as well as integer at far lower power on ARM.
Atom will be old news this year... ARM will need to be able to compete with AMD CPUs.
Unless the screen technology changes (one of the biggest power draws) I really don't see ARM's traditional strength of low power being a big deal. And with transistor sizes shrinking, having a simple design is of little benefit. I think they are in for a lot more competition than they are used to. The question is really how well RISC instructions scale to very complicated chips, and the history (all high-performance CPUs left are CISC) and parallels point to CISC scaling better the more complicated the chip is.
Three things here:
1. A15 is a tier higher (both in performance and power) than the current A9's. Looking at the pipeline and DMIPS numbers, it looks on-par with the K8 as far as computational performance. 2. Atom will remain the only possible competitor to smartphones. AMD's new offering is nowhere near the power envelope needed to fit inside a smartphone. Atom is about 4x what it needs to be currently. I think Intel can pull off a 25% power reduction in the coming years. 3. If we're looking at ultra-high performance, the biggest, baddest processors today are Power from IBM. There really isn't some inherent scalability between CISC or RISC ISA's. It's just a matter of who wants to implement it.
In the past, RISC ISA's that try to go up against x86 have been from one or two vendors at best. ARM has literally every chip vendor in the world except AMD and Intel implementing it to some degree or other. Hell, from-scratch microarchitectures for ARM even outnumber x86 designs based on the number of architectural licenses.
I don't think there have really been any benchmarks that measure SoC power; nobody's taken these phones apart and put a current meter on them. As for overall performance of the phone, there are far bigger factors (power management on the board, software configuration, type of display, etc.) that simply dwarf the power of the SoC.
Transistor count is just one (and possibly a minority) factor in chip power. The single-biggest impact on chip power is operating voltage. For the same task/compute power, 2 cores running at a lower voltage (which correlates with lower frequency) will use less power than a single core running at a higher voltage (and thus higher frequency).
This of course, assumes the software is smart enough to downclock the two cores if it's a relatively low-priority task that can be threaded.
"Most" SoC's don't. IIRC the only chipsets that integrate the baseband are from Qualcomm. All of the others (Tegra, OMAP, A4) are CPU+GPU+MCU combos with some dedicated DSP functions here and there.
But then again, considering just how much of the SoC marketshare Qualcomm has, I guess "most" might be correct....
Not that I disagree that better software is needed, but Nokia is probably a bad example. The reviews of the N8 (their current flagship device) have all come to the conclusion that web browsing is annoyingly slow. Not intolerable, but certainly not what you'd expect from a smartphone in 2011.
A better example would be the iPhone 3GS. 600MHz Cortex A8 and it largely provided better browsing experience than most of the 1GHz Android devices out there. At least, before Android 2.3.
Slashdot Mirror
As to why there's not a Android "iPod"? Good question. Apple has sold just as many iPod Touch units as iPhones, which is why combined sales of iPhone and Touch and iPad combind beat out Android combined devices by 59%.
Not really. Even if you count "other related music services", iPods are still half the revenue of iPhones. There were more iPods sold but the profit margin isn't as fat on those as iPhones.
That being said, I think you can clearly say tablets and mobile media players are dominated by Apple. And even if you could dismiss Android's success as "carriers who don't have iPhones provide it", there's still no question that in their most profitable product space (smartphones), Apple is losing out.
That being said, with as much profit and mind-share that Apple gets -- and how huge the smartphone market will grow to be -- even if they own 20% of the market, that's still enough to be a huge wad of cash.
"I think, therefore I am"
What you're arguing here is that all observations and internal consistency of the human brain is up for question. Which may be technically true but hardly equivalent to belief that scripture describes accurate non-observable phenomenon and history.
Are they both leaps of faith? Sure. The former leap of faith (believing in observable reality) causes me to get up in the morning and eat instead of going "fuck it, none of it is real anyway".
People made planes without physics? Many modern day integrated circuits wouldn't work without understanding quantum mechanics....
Of course, if that's your definition of "faith", what you're talking about here is the same as faith of existentialism. If you want to really expand faith to that level, the everything except self-existence is a leap of faith. And in that respect, there's no difference between me believing that I'm typing on an actual computer -- interacting with physical matter -- rather than just dreaming all of this.
Do I commit absolute certainty to the fact that others exist in the world? What about history? Do I believe that historical documents describing the actions of Einstein is true? Do I believe that mathematics is consistent and axiomatic? Do I believe that the logic and reason my brain is capable of in any way corresponds with the workings of the universe? Do I believe that any empirical observations made by anyone ever isn't just the result of some invisible, intangible flying spaghetti monster mucking with the results?
We've decided to narrow down the definition of faith with respect to religion to not encompass such a broad topic. Simply: religious faith is narrowed down to absolute certainty of the teachings of some organized scripture.
Said organized scripture may even have scientific or archeological evidence associated with it. But the definition of faith is that, regardless of whether or not anyone can produce any observable, repeatable evidence, one accepts something as truth. And in that respect, science is never like that because every theory comes with the caveat: "this is simply what is consistent with recorded observations". Science relies on faith in the general case the same way that your belief that the reality you observe is real relies on faith in the general case.
Coding? A basic flier requires a simple text editor like Wordpad. Hell, even if you include Office basic, you're still only up ~$80 over the sticker price.
What default software does a Mac come with that makes flyer making so much easier? I don't think it's any of the iStuff.
I'm sure the Apple fans comprise a core part of the sales but look around at an Apple store, most people aren't hardcore iPeople. They walk in, use the product, ask questions and get somewhat informed answers. Even if they don't buy, this gets you into the consumer's brain as a solid buy because you're informed about what you're getting.
Contrast this with a Xoom sitting around at a Best Buy. Nobody's around who knows anything about it. If you ask questions like "how do I purchase music" they can't just do a quick demo using iTunes.
HP might have the clout to fix this as they are used to large-scale sales but they'll have to step up their game of forcing retailers to educate their staff; or perhaps just having dedicated HP salespeople onsite at various retail outlets.
Android has been popular because of both price and the fact that they had the carriers to do the selling for them. This isn't true of tablets.
Why are they considered private property? It seems to me that they resemble real estate far more than they resemble furniture. They are publicly granted exclusive ownership of something that isn't inherently exclusive; similar to the land.
This is actually a brilliant idea. If you want to profit off of the exclusive rights granted to you by public laws, you should pay a tax to support said infrastructure (government) that enforces and protects said exclusive rights.
Treating all IP in every industry the same way is flawed to begin with and the barrier to entry to stake a claim on an idea is way too easy.
The whole point of IP is to encourage innovation by providing incentive for the inventor. Today's IP laws are a flimsy shadow of that. Studios and IP troll companies collect the rewards and inventors are relegated to idea-generating grunts.
There is no inherent morality to ownership of an idea; it is something granted by the public to individual holders of IP for the benefit of the public. If at any time, said laws are detrimental to the public, it should be repealed.
Of course, that would require a government that isn't bought and paid for and a populace that's at least decently educated and informed.
To smartphones! The cause of and solution to all of live's cancer.
My guess is you're not an EE either. Interference doesn't quite work like that; orthogonal frequencies, for instance, do not interfere at all even when one is extremely high-powered.
Most electronic designers are competent enough to put a choke at their power line and a bandpass filter at their cabling. It's not "easy" but it's done in just about any military grade electronics. I guess Boeing engineers didn't think it was necessary.
I understand the need to build up hype but as others have said, doing it 6 months ahead of time will just kill your momentum. Announce things when they're ready.
This is a problem with or without graphene. Chip design usually doesn't have signals travel all the way across the chip. In fact, good place and route engineers will keep signal nodes as close as possible and logic designers will try to keep high-connection nodes to a minimum, separating out logical clusters to be as isolated as possible.
We're not talking "a bit high". ECL logic is *insanely* power hungry. If we implemented all of a processor's logic in source-coupled logic, it would consume 10-100 times more power than it does currently.
Some decades ago, people thought the GUI was insufficient for "serious" productivity as well. It still may be true today for a small subset of people (programmers and engineers mainly). But for the most part, computers are GUI based.
The problem with your argument is your assumption that "serious" gaming, productivity, etc. is the end-all, be-all of use-cases for the computer. It isn't even the majority. The vast majority of people use their PC for some light document editing, spreadsheets maybe, casual games (more Angry Birds, less WoW) and zero programming.
"The Death of the PC" doesn't mean there will be no PC's around. It just means they'll be relegated to what mainframes, unix terminals and shellscripts are today: tools only used by people who know what they are and how to use them.
The market will shift from selling PC's to selling tablets and netbooks. With the modern PC taking on the role of a home server hub. You will definitely see less families with 3+ laptops -- one for each person. Likely the average home will have one desktop box for storage on the network and multiple tablets with docking stations that enable mouse+keyboard input.
Catholicism also uses 14 bit signed numbers to keep track of years in history. Consequently, they think that 8192 B.C. is the beginning of time and refuse to believe otherwise.
That is my argument exactly, but not in the way you envision it. If you live in New York City, other than a cool factor, what exactly is all of that extra horsepower in a Ferrari going to get you? There is no place to use the extra HP. So, it may be super efficient at 100mph compared to the pinto, but there is nowhere to use it (legally) as such. So, how is it's efficiency at the 25mph to 40mph (or less) that is common in the big city.
For any given performance level, a dual-core solution -- within the vicinity of today's smartphone performance level -- will be more power efficient than a single-core.
Your argument isn't "a pinto is more efficient at 25mph", it's "a pinto going 25mph uses less gas than a Ferrari at 100mph". Again, no shit.
In this analogy, the Ferrari going at 25mph will use less gas than the Pinto.
Likewise, a dual core is more efficient in raw computing power than a single core.
No, a dual-core is more efficient at any level of computing power -- barring extreme cases such as a wristwatch. If you have to reach performance level x, and you have two choices: either make a dual-core solution that can achieve x or a single-core solution that can achieve x -- again, we're not talking about a case where the dual-core is more powerful, just the same level of performance -- you are much better off power-wise to go dual-core.
Or, to put it simply, if you take today's 1GHz SoC's and instead, put in two 500MHz processors -- again, with the caveat that this will only apply to parallel processes -- the two 500MHz processors will be more power efficient even if it's not any faster than the single core 1GHz CPU.
If again you want to use circular logic and say "well a 500MHz single-core will use less power", a dual 250MHz dual-core would have been a better choice for power efficiency.
The point I am trying to make is that just about any power saving measure you take with a dual core can also be implemented on a single core.
Your point completely ignores performance. Of course if you lower the performance expectation, you can get lower power.
If you lower the voltage to a single core and it will use less power. Now, at some point, you get to the point were the single processor can't get any real work done (becomes less responsive from the user perspective) running at lower voltage, whereas two cores running at such low voltage still could (of course chances are they'd both be "on" most the time if the voltage was low enough).
And if you stuck a Cortex M3 in there instead of a Cortex A8, you can save even more power! But your device will be slow as balls.
You're essentially arguing that if you're willing to accept lower performance, you can get lower power. Well no shit.
The point is, for the same processing power -- to accomplish the same task in an equal amount of time -- a dual-core solution is more power efficient than a single-core solution. This of course, has the caveat that the task must be parallelizable and that the software is smart enough to control the hardware efficiently.
What exactly would we be adding to a smart phone that would need the extra computing power of multiple cores? It's not like many people are going to be doing cad/cam on them.
As I've pointed out, even if you wanted to just maintain the performance level of today's single-core chips, a dual-core implementation is more efficient; you can accomplish the same performance using lower voltage cores.
We've already seen this in the netbook market. While desktops and high end laptops kept getting faster and faster processors, the processor in most netbooks is lower voltage and frequency to a) extend battery life and b) their intended use does not warrant higher processing power.
You forgot c) run annoyingly slow for more complex websites and anything with flash or high-def video.
Netbooks are a bad example. But again, we're not talking about the ramifications of increasing the performance wall. We're talking about maintaining the same performance at lower power, which dual-core solutions provide over single-core.
Using the car analogy, again, you can make an eight cylinder engine which is pretty fuel efficient by doing things that turn off cylinders, change compression, etc. But, no manufacture would do so and stick one in, something like a prius or even a smart car. Instead, they use a smaller, less powerful engine that is optimized for the task at hand.
Using the car analogy, if you can get the same torque and horsepower out of an 8-cylinder but use less gas compared to the 4-cylinder, why wouldn't you.
Your argument essentially boils down to "a pinto would use less gas than a Ferrari". Well no shit.
I'd say quite the opposite really. The CPU's are good at integer and branchy type stuff that the GPU can't do anyway. But the A9 implementation they chose is a really really really stripped down FPU.
The GPU is what's pulling the weight in graphics related tasks.
And it was destroyed 10-to-1 by Atom on CPU-bound floating point (Whetstone, LINPACK). Of course A8 will be a simpler more power efficient design when it leaves off floating point.
The A8 and A9, yes. However, you can look at Scorpion and see that it's possible to match and/or beat Atom in FP as well as integer at far lower power on ARM.
Atom will be old news this year... ARM will need to be able to compete with AMD CPUs.
Unless the screen technology changes (one of the biggest power draws) I really don't see ARM's traditional strength of low power being a big deal. And with transistor sizes shrinking, having a simple design is of little benefit. I think they are in for a lot more competition than they are used to. The question is really how well RISC instructions scale to very complicated chips, and the history (all high-performance CPUs left are CISC) and parallels point to CISC scaling better the more complicated the chip is.
Three things here:
1. A15 is a tier higher (both in performance and power) than the current A9's. Looking at the pipeline and DMIPS numbers, it looks on-par with the K8 as far as computational performance.
2. Atom will remain the only possible competitor to smartphones. AMD's new offering is nowhere near the power envelope needed to fit inside a smartphone. Atom is about 4x what it needs to be currently. I think Intel can pull off a 25% power reduction in the coming years.
3. If we're looking at ultra-high performance, the biggest, baddest processors today are Power from IBM. There really isn't some inherent scalability between CISC or RISC ISA's. It's just a matter of who wants to implement it.
In the past, RISC ISA's that try to go up against x86 have been from one or two vendors at best. ARM has literally every chip vendor in the world except AMD and Intel implementing it to some degree or other. Hell, from-scratch microarchitectures for ARM even outnumber x86 designs based on the number of architectural licenses.
I don't think there have really been any benchmarks that measure SoC power; nobody's taken these phones apart and put a current meter on them. As for overall performance of the phone, there are far bigger factors (power management on the board, software configuration, type of display, etc.) that simply dwarf the power of the SoC.
Transistor count is just one (and possibly a minority) factor in chip power. The single-biggest impact on chip power is operating voltage. For the same task/compute power, 2 cores running at a lower voltage (which correlates with lower frequency) will use less power than a single core running at a higher voltage (and thus higher frequency).
This of course, assumes the software is smart enough to downclock the two cores if it's a relatively low-priority task that can be threaded.
"Most" SoC's don't. IIRC the only chipsets that integrate the baseband are from Qualcomm. All of the others (Tegra, OMAP, A4) are CPU+GPU+MCU combos with some dedicated DSP functions here and there.
But then again, considering just how much of the SoC marketshare Qualcomm has, I guess "most" might be correct....
Not that I disagree that better software is needed, but Nokia is probably a bad example. The reviews of the N8 (their current flagship device) have all come to the conclusion that web browsing is annoyingly slow. Not intolerable, but certainly not what you'd expect from a smartphone in 2011.
A better example would be the iPhone 3GS. 600MHz Cortex A8 and it largely provided better browsing experience than most of the 1GHz Android devices out there. At least, before Android 2.3.