Artificial intelligence requires manipulating large amounts of data and it must be kept in mind that the more data there is to process, the more power there is and this is where the division of labor becomes important and for this we need a large number of hearts. And what types of processors comply? Well, we have two candidates, on the one hand those which are designed for HEDT or workstations and on the other hand those which are designed for servers.
Which of the two types is better? Well, HEDT processors like the AMD Threadripper or Intel equivalents like the Xeon Workstation. The reason for this is that server CPUs, due to the need to always be on, run at lower clock speeds, they often lack Boost. Additionally, its add-on hardware is more expensive and often not ready in shape and size for desktop use.
Are processors good for AI?
At first glance, no, if we make a classification of the different types of processors to execute artificial intelligence algorithms. Although this has a nuance and it is the fact that today they are generally used to execute scalar parts of code. What does it mean? Well, it refers to arithmetic operations that work with a single number, while those that work with many are sent to coprocessors such as the powerful SIMD units of GPUs or the recently added systolic arrays or Tensor units.
CPUs for AI don’t really exist, but in recent years we’ve seen new instructions and execution units that make them much better at running such algorithms. For example, it is normal to work with data sets of low precision and therefore of length 16, 8 and even 4 bits. Well, these instructions were not common in a CPU, but they were added due to demand and especially in SIMD units.
Units or matrices of systolic tensor in CPU
Then we have the fact that many deep learning problems are solved with so-called Tensor units or Systolic Arrays, the operation of which we will not discuss in this article. What we’ll drop is that the neural processors, NVIDIA GPU tensor units, and new Intel Core 12 AMX units are systolic arrays, so these are the Intel processors released to the market from the end 2021 to convey those who are more ready to run artificial intelligence algorithms. As long as applications have been optimized to use these instructions.
However, these instructions when moving large volumes of data require large bandwidths and here we get into how Intel and AMD respectively solved this problem. In the case of the three-letter company, what they have done is increase the LLC through the V-Cache to improve its performance, the company with the blue mark, on the other hand, has certain models of its Xeon Workstation memory HBM in a complementary way to get the most out of your AMX unit.
Whichever processor you choose, we recommend pairing it with RAM with good bandwidth. It is for this reason that we recommend one of the HEDT type, since their motherboards generally have configurations with four and up to eight memory channels. This means two or even four times the transfer speed or bandwidth of a desktop PC.
Isn’t a GPU better for AI?
Yes, that statement is true, but nothing beats the combination of a powerful workstation processor and a powerful graphics card working together. Let’s not forget that for some time and especially in the case of NVIDIA, we have seen their graphics processors evolve towards optimizations to run artificial intelligence algorithms, based on different types of machine learning. However, the strength is in the union and since a system is only as powerful as its less powerful part, choosing a desktop processor not only means a reduction in processing performance but also in terms of memory. .
In the midst of a shortage of graphics cards, we have to look for alternative solutions and this is a problem that Intel and AMD are not the only ones to be aware of in the world of x86 processors. The ARM world is also focused on optimized high-core, high-bandwidth AI processor designs. We have examples of processors like Nuvia, now Qualcomm, Ampere or Fujitsu’s A64FX which mark a new paradigm in the face of new needs.
So if you are looking to buy a PC to run your AI algorithms and can’t find high caliber graphics cards available then a HEDT or workstation processor is the best option and even more so when they have specialized units for this.
