Gadgetory

often Moore's law is incorrectly interpreted as meaning that technology is going to double in terms of performance and power each year as we'll see that's not exactly what Moore said but it's still the yardstick that many people use for a while it did seem that way every new handset came with a massive hike in power and speed but our phones still getting faster does Moore's law still apply to our modern mobile devices let's find out the problem here is that it can be quite tricky to compare two devices or CPUs seeing as they often aren't directly comparable there's only so much we can glean from the numbers so let's start off by looking at how clock speeds and benchmark scores have improved before taking a deeper look at some of the other factors that might be affecting the differences we see I'm a Samsung guy I've got a few Samsung devices lying around so let's take a look at how the flagships have changed over the years so looking at this table you can see that there's not really that much change between the SI in the s7 both are packing octa-core processors and there's only a small difference in terms of the gigahertz the Geekbench scores are pretty different though and that's our first clue that there is something else going on here but it's certainly not double the power we have to go all the way back to the s5 before we see a quad-core with any t gigabytes of RAM and a benchmark score that's under half of what we see on the s8 those of you who like to have the most powerful and fastest devices possible might be a little bit disappointed by that but it is a little more complex than that as well so as gary has explained in a few of his previous videos gigahertz does not tell you the full story in terms of the performance of a CPU so gigahertz tells you the clock speed and that's how quickly a CPU can carry out a sequence of instructions these are stacked up in order in what is known as the pipeline so back in the day doubling the clock speed would mean roughly double the power because it meant that the CPU could carry out those actions twice as quickly but modern advancements in efficiency have gotten around this in some clever ways so for example an execution engine can now carry out more than one instruction at the same time through something called instruction level parallelism this is sometimes referred to as the pipeline getting wider likewise CPUs can now begin fetching the next instruction before the current one is complete by splitting those instructions down into smaller blocks this is sometimes described as making the pipeline longer so basically what methods like this mean is that although the clock speed might be the same the number of revolutions times around might be the same the CPU is now carrying out more instructions on each go-around therefore it's more efficient and faster despite the gigahertz being the exact same then you have things like the advantages and slight disadvantages of having multiple cause you have the RAM you have the GPU you have the size of the cache all these factors can influence those benchmark scores as well as the performance you see every day even when the gigahertz doesn't actually change but what Moore was talking about more specifically was the number of transistors that we'd see on a chip he said that year-on-year the size and efficiency of transistors should improve such that we'd see double the amount on the same sized chip that would mean in theory that the number of transistors in your CPU should be doubling year-on-year so when you see a CPU described as 10 M that actually means 10 nanometers that means that half the length of a transistor cell is 10 nanometers on that chip 10 nanometers is currently the smallest we've got and the smaller the size of the individual cells the more you can fit onto the chip currently samsung and tsmc are racing to bring us the first seven nanometer chip and tsmc are also looking at building factories to create three and five nanometer chips so this is something that's increasing rapidly currently but the precise number of transistors is of course related to the size of those transistors as well as the size of the chip so of course the density is only part of the story we've got a very dense chip but it's tiny could still have fewer transistors than a less dense chip that is massive so how many transistors are there in the modern flagship well the Samsung Galaxy s8 has got the Snapdragon 835 and that has a whopping three billion transistors just to put that in perspective the human brain has only got 100 billion neurons and there's only just over seven point four billion people on the planet so there's a lot of transistors how does this compare to previous generations well unfortunately OMS are a little less candid when it comes to transistor counts versus gigahertz and I can't get that information for the s6 and the s7 however Apple are a bit more forthright we know that the iPhone 6 has and a8 chip in it and that has two billion transistors and we can compare this to the iPhone 5s which had the a7 chip and that had one billion so that is quite a big hike year-on-year even if it's not doubling we also know that the iPhone 8 and 10 will sport the a11 chip with an incredible 4.3 billion transistors and if that wasn't enough the ki-rin 970 is gonna have 5.5 billion transistors so these are massive jumps in terms of speed and performance it's over nine I'm sensing a table coming on so in this regard were definitely seeing some pretty big jumps each year and it almost seems to be speeding up but then if we look at the difference and the Geekbench scores between the iPhone 5s and the iPhone 6 we can see that they're not actually all that different and this might seem quite strange because the a8 has got double the number of transistors and they're otherwise pretty comparable in terms of their specs so you might have guessed there and transistor counts also can't tell us the whole story in terms of performance that used to be the case when things are a bit simpler but today once again things are more complicated so another law worth learning is Dennard scaling Dennard scaling tells us that as transistors get smaller their power density should remain the same and this tells us that the power usage of a chip should be more to do with the size of the chip and not to do with a number of transistors and this is pretty important because otherwise as we doubled the transistor count we'd end up with hardware that required huge amounts of power to run and got extremely hot so really Moore's law doesn't work without Dennard scaling and actually Dennard scaling stopped holding up around 2000 so today there's no guarantee that as density increases stators efficiency and that's one reason why doubling the number of transistors doesn't necessarily mean double the performance and it sort of down to OEM to decide how they want to utilize all those transistors not all of them will necessarily be used for pure computing power some might be used for power efficiency options etc so if we're being really strict and down our scaling is kind of broken and Moore's law doesn't apply it's somewhat redundant these days and Moore himself even revised his theory in 1995 to say that it would double every two years and even then he said it was approximate so no your phone isn't doubling in power year-on-year but you shouldn't be too disappointed about that it's kind of moot to think about that way anyways your phone is becoming more efficient and smarter and cleverer at the way it uses the power it's got it is getting faster and it's also improving in terms of build quality the screen resolution the camera the water resistance the software optimization if you can pair a phone today to a couple of years ago it's still pretty remarkable how far it's come even in that short amount of time and as we go forwards virtual reality augmented reality artificial intelligence and 4k screens they're all going to push the need for faster and even more efficient performance even further there are some beastly phones on the horizon thanks ton for watching guys I hope you found this interesting if you did then please consider leaving a like maybe leave a comment down below I'll be down there to answer any questions you have subscribe to the channel for more like this and hit the bell button to get notified and of course check out and related com4 we are your source for all things Android