Why can’t companies like Apple (AppleSilicon) or other ARM chip making companies just make really big chips that would outperform anything and everything we currently have? Disregarding battery life and heat issues, what about a laptop that is only one mega chip that takes up all the inside?
In: Technology
People haven’t covered yet the other side of why you can’t just have one laptop-filling chip (besides die yield itself). Chips require support hardware. For example on a motherboard around the CPU slot you’ll see all sorts of capacitors and chokes and what have you, because part of a motherboard’s job is to provide the power chips need — and that includes the right voltage, and very stable.
Plus, modularity. In theory sure you could bake the CPU, GPU, RAM, Storage, all that into some ultra-hyper megachip (we already do that to an extent sometimes, it’s called System On a Chip, shortened to SoC). But what if you wanted to add RAM or switch to a bigger hard drive because they’re inadequate? You either have to buy a whole new humongous chip or a whole new laptop.
But even then it wouldn’t fill the entire laptop, because it still would need support hardware and connectors, and because a Chip with all that stuff wouldn’t need to *be* that big. In fact, if you research CPUs alone across different generations, even for having more and more stuff packed in them and being immensely more powerful than say 20 years ago, they’re getting *smaller,* and as counterintuitive as it may seem that shrinking actually *aids* in becoming more efficient and more powerful.
You can only fit so many transistors on a die of any specific size. The larger you make the die, the lower the yield will be on silicon wafers.
Have you ever seen those silicon wafer discs? [pic](https://static2.bigstockphoto.com/5/1/3/large1500/315929284.jpg)
The larger the die the fewer you get per disc, and the more expensive it is. Also, the larger the die the more chance of flaws. For every disc made sometimes portions of it are flawed.
One thing I would point out is chip yield. The larger and more complex a chip is, they higher the likely failure rate during manufacturing and the lower the yield.
I.e. a CPU today might have a 70% yield. This means that 30% of the chips you make aren’t usable. The larger and more complex you make a chip, the lower the yield.
People wouldn’t buy it.
The average consumer wouldn’t need that much power and the price of manufacturing such a chip would skyrocket.
There is an explanation for that and it has caused a slow down of chip technology recently. It is called Moore’s Law.
One aspect is modularity. A single CPU that only functions as a CPU can be paired with multiple and various periphery chips to build out a system for specific purposes.
Case in point, a laptop doesn’t need 8 SATA connections, so there is no reason to have periphery chips on the motherboard to support that. They can just incorporate the needed amount. If all that was on the motherboard was in the CPU instead, you wouldn’t have that modularity and there would be those unused resources being powered.
This is possible, but a bad design. Microchips are the most expensive component in a computer. Making a computer as one giant chip would work, but would cost a lot more. So that’s why people don’t do it. Not because they can’t, but because they can’t sell it.
Signal times. The physically larger the chip, the longer the distance a signal has to travel between components on one end of the chip to the other end. This is only a few millimeters, and takes a small fraction of a second, but modern computers work so fast that the signal travel time is significant. This is part of why it is always better to make a chip with smaller components than increase the number of components by putting more of them on a bigger chip.
Latest Answers