Cost of Manufacturing CPU and GPU at Equal Size

As a general rule, processors are smaller in size than graphics card GPUs : for example, a Core i9-9900K processor is 177 mm 2 in size while the die of a GTX 1660 Ti is 284 mm 2 . Even at equal size, why do CPUs cost more than GPUs?

Leaving aside the difference in the lithographic process of one and the other, why if a CPU is much smaller it has a much higher cost than a complete board that includes GPUs, memory chips, power phases, fans, etc.?

Cost of Manufacturing CPU and GPU at Equal Size

The cost of manufacturing the CPUs and GPUs

For starters we must keep one thing in mind, and that is that while a CPU is designed for general purpose computing, the internal components of a GPU are designed to perform a single task (or group of tasks rather), which is not another than graphics processing. In other words, a CPU would be able to perform the tasks of a GPU (much slower, yes) but a GPU cannot perform the tasks of a CPU. This means that both instruction games and the design of a CPU is much more complicated than that of a GPU, thus increasing its cost.

CPU

The manufacturing cost is therefore not determined by the size of the CPU or the GPU, but by the complexity of manufacturing and its design. This also implies the “waste” when manufacturing: when we talk about the wafers (wafers) of the processors, the small defects in the outer areas of these make the processors unusable, so their waste is greater than in GPUs (which, as you well know, “defective” GPUs are frequently reused in lower models, with disabled cores and others.

Speaking of complexity, we might think that a CPU typically has only a handful of cores, while a GPU is on the order of thousands in many cases, so it should be more complex. The reality is that, in the end, all of these GPU cores are identical and designed for parallel computation, so they don’t add up in this regard.

Therefore, the cost of manufacturing the CPUs is greater than that of the GPUs mainly due to two factors: first, the complexity, and second, due to the waste, and it does not have much to do with the size of the die because, in the end and at the Out, the difference in raw materials is almost negligible.

The price is dictated by the market

This does not mean that, for example, an RTX 2080 Ti is considered extremely expensive, while a 32- core AMD Threadripper is considered a good value, despite the fact that it also has a high price. It all comes from the perception that we have and the state of the market, but of course the manufacturing costs of the devices also influence since any manufacturer has the primary objective of making money with their products after all.

Thus, there is no relationship regarding die size and cost . If for example you were sold a CPU with a die of 150 mm 2 for 300 euros, one with a die of 300 mm 2 will not necessarily cost twice as much, and the same happens with GPUs. The price of products is dictated by manufacturing costs, of course, but also by the complexity of the design and, of course, the state of the market.