- GPUs are like buses: slower than sports cars, but much better at shifting a lot of numbers in parallel.
- GPUs are used in machine learning, medicine, image processing, and games.
- Intel’s Iris Xe Max is designed to make laptops more powerful for creators and AI.
Intel’s new Iris Xe Max Graphics Processor Unit is now showing up in laptops, and by all accounts . But what is a GPU, and why is it important? Spoiler: It’s not about games, or even graphics.
The CPU in your computer, the one that does the day-to-day work, is expensive, and highly specialized. A GPU, on the other hand, is really, really good at math. Specifically, they can multiply big numbers, and they can perform many, many operations in parallel. This makes them good for generating complex 3D graphics, but they are used for much more.
“GPUs are great for big data, machine learning, and image processing,” 3D animator David Rivera told Lifewire via instant message. “I have many colleagues who use it in medicine to get MRI results.”
Big Math, Big Pictures
Anything that requires a lot of complicated math is perfect for offloading to the GPU.
“Graphics are usually very powerful because calculating 3D video stuff is very complex,” Barcelona-based computer engineer told Lifewire via instant message. But soon, computer boffins realized these math machines could be pressed into use for all kinds of math-intensive tasks.
“Now, supercomputing clusters are also being made with GPUs. They are used for scientific calculations, engineering, etc,” says Bonastre. Another advantage of the GPU is that it’s easy to scale up. It’s built to run identical operations in parallel, so adding more chips (or just more cores to the chip design, making it bigger) makes everything faster.
A GPU is also great for processing photographs. For example, Adobe’s Lightroom photo-editing suite to “provide significant speed improvements on high-resolution displays,” which includes 4K and 5K monitors.
“CPUs are optimized for latency: to finish a task as fast as possible,” . “GPUs are optimized for throughput: they are slow, but they operate on bulks of data at once.” Serpa compares a CPU to a sports car, and a GPU to a bus. The bus is a lot slower, but it can shift a lot more people.
What About Your Phone?
The GPU in your phone is used to drive its super high-resolution display, and to run the graphics. That’s why the phone gets hot when you play a game—the GPU kicks in, and your phone has no fan to cool it down.
On the iPhone, the GPU is used for image recognition, natural language learning, and motion analysis. That is, it processes images and video as you shoot them, and more.
“GPUs are great for big data, machine learning, and image processing.”
But that’s not all. Apple’s recent iPhones and iPads contain a “Neural Engine.” This is a big chip, specially designed to carry out machine-learning tasks. It’s not a GPU, but it’s GPU-like in concept, in that it crunches hard math problems in no time at all. The latest version is, , “capable of performing up to 11 trillion operations per second.”
Perhaps the biggest buzzword in computing right now is “machine learning.” This involves showing the computer a lot of examples, and letting the computer work out the similarities and differences. GPUs are perfect for this because they can view more examples per second. However, once that training is done, the GPU is no longer needed. Any learned algorithms can be run faster by the CPU.
Now, let’s go back to Intel’s new Iris Xe Max GPU. This is designed for running in “thin-and-light laptops and [to] address a growing segment of creators who want more portability,” said Intel Vice President Roger Chandler . That is, it’s meant to make power-constrained laptops better for editing video, photos, and any other GPU-intensive activity. Yes, including AI.
The Iris Xe Max is designed for machine learning. Perhaps its first task will be to learn how to pronounce its own name.