Intel vs phones vs ARM vs laptops

Deja vu all over again once again

Intel - logoThe following is a modern re-interpretation of an older article on a completely different subject, call it deja vu. Once again, the concept is the same, and no animals were harmed in the filming of either one.

The laptop business is the subject of much debate of late, with the most heated arguments on the subject of x86 vs ARM, and whether or not ARM chips can invade this traditional x86 market. Those who profess to have a clue about this subject invariably don’t for numerous reasons. Unfortunately, these people tend to be the loudest, and are picked up by the vast echo chamber that passes for technology news. To make matters worse, many companies with a vested interest will promote these ignoramuses for their own selfish reasons.

Lets look at the case against ARM in notebooks as Intel is so keen to promote. ARM comes from a traditionally low power environment where high processing performance is secondary to frugal battery use. This space is of course cell phones. In laptops, batteries are more than an order of magnitude larger, and CPU power is more than proportionately larger. The gap between x86 and ARM CPU performance has been shrinking, as has the energy use gap, but ARM is without question the energy use champ. x86 is also without question the CPU performance champ.

One big problem for ARM in laptops is that there is no shipping device with Windows for ARM (WARM), and there won’t be for quite a while. This will change before the year is out, but for now, in the eyes of many, ARM is shut out by the lack of an appropriate OS. To make matters worse, when WARM arrives, ARM CPUs would need to offer similar CPU performance to x86 chips for users to not turn their noses up at the experience. ARM claims to be in the right performance ballpark, but until WARM devices ship, it is just talk.

Physics is a bitch. To attain a certain level of performance, you need to flip a certain number of transistors. Higher performance means more flips, and that means more energy use. Some ISAs may be more efficient than others, but across a SoC as a whole, with memory, fixed function accelerators, and other things taken in to account, the difference is pretty small. If ARM is going to increase performance to the point where users are accepting of the performance levels, its power advantage will largely evaporate. More problematic is that lingering software incompatibilities will remain regardless of the CPU performance.

This is the long way of saying that ARM is invading x86 space, and it will have to meet x86 demands. Unfortunately, it will also run in to the same problems that x86 has already worked through, and will have trouble reaching that goal. There are many more barriers than people expect for ARM to play in the traditional laptop form factor, and the road will be anything but easy. Performance at a given power level, software compatibility overhead, and general suitability for the task will stop ARM from being successful in this space.

Now, lets look at it from the ARM point of view, with a word replace of x86 for ARM, and a little more nuanced replace of phone/low power/lower performance for laptop/high power/high performance.

The cell phone business is the subject of much debate of late, with the most heated arguments on the subject of ARM vs x86, and whether or not x86 chips can invade this traditional ARM market. Those who profess to have a clue about this subject invariably don’t for numerous reasons. Unfortunately, these people tend to be the loudest, and are picked up by the vast echo chamber that passes for technology news. To make matters worse, many companies with a vested interest will promote these ignoramuses for their own selfish reasons.

Lets look at the case against x86 in phones as ARM is so keen to promote. x86 comes from a traditionally high power environment where frugal battery use is secondary to high processing performance. This space is of course laptops. In cell phones, batteries are more than an order of magnitude smaller, and CPU power is more than proportionately lower. The gap between x86 and ARM CPU energy use has been shrinking, as has the performance gap, but x86 is without question the CPU performance champ. ARM is also without question the energy use champ.

One big problem for x86 in cell phones is that there is no shipping device with Android for x86, and there won’t be for quite a while. This will change before the year is out, but for now, in the eyes of many, x86 is shut out by the lack of an appropriate OS. To make matters worse, when Android for x86 arrives, x86 CPUs would need to offer similar energy usage to ARM chips for users to not turn their noses up at the experience. x86 claims to be in the right energy use ballpark, but until Android/x86 devices ship, it is just talk.

Physics is a bitch. To attain a certain level of performance, you need to flip a certain number of transistors. Higher performance means more flips, and that means more energy use. Some ISAs may be more efficient than others, but across a SoC as a whole, with memory, fixed function accelerators, and other things taken in to account, the difference is pretty small. If x86 is going to decrease energy use to the point where users are accepting of the battery drain, its performance advantage will largely evaporate. More problematic is that lingering software incompatibilities will remain regardless of the CPU performance.

This is the long way of saying that x86 is invading ARM space, and it will have to meet ARM demands. Unfortunately, it will also run in to the same problems that ARM has already worked through, and will have trouble reaching that goal. There are many more barriers than people expect for x86 to play in the traditional cell phone form factor, and the road will be anything but easy. Performance at a given power level, software compatibility overhead, and general suitability for the task will stop x86 from being successful in this space.

Eerie isn’t it? The arguments against one ISA in a non-native form factor are pretty much exactly the same as the ones against the other ISA in non-native form factor. Unfortunately for x86 or ARM, feel free to search and replace at will, all the arguments both for and against either apply to both equally well. In fact, if one can be successful invading the other’s turf, so can the other.

x86 vs ARM is much more complex and nuanced than the above ‘PR flack’ level arguments, but this is the tip of the iceberg. Interestingly, most of the deeper and more nuanced arguments also are almost totally search and replaceable. If x86 can be successful in phones, ARM can be successful in laptops. Neither has any magic barriers to exclude the other, and both have learning curves that cross over at about the same point.

Both ARM and x86 will climb their respective hills, and both are entirely capable of doing what the other does. Don’t let anyone tell you it can’t be done. Whether or not either side will do what is necessary, for technical, financial, or sheer persistence reasons is another question entirely. Both can and will try. Both can be successful, and maybe both will, but the outcome remains to be seen. It is deja vu all over again.S|A

The following two tabs change content below.

Charlie Demerjian

Roving engine of chaos and snide remarks at SemiAccurate
Charlie Demerjian is the founder of Stone Arch Networking Services and SemiAccurate.com. SemiAccurate.com is a technology news site; addressing hardware design, software selection, customization, securing and maintenance, with over one million views per month. He is a technologist and analyst specializing in semiconductors, system and network architecture. As head writer of SemiAccurate.com, he regularly advises writers, analysts, and industry executives on technical matters and long lead industry trends. Charlie is also available through Guidepoint and Mosaic. FullyAccurate