From the course: Video Production and Post Tips

What is a GPU?

From the course: Video Production and Post Tips

Start my 1-month free trial Buy for my team

What is a GPU?

“

- Hey there, I'm Robbie Carmine. - And I'm Rich Harrington. - And Rich, this week we're talking about something that, once again, is near and dear to my heart. - Yeah, performance. - Yes, performance indeed. And specifically we're going to be talking about GPUs. What's GPU? - Well, it's the graphic processing unit. And for years, computers were dependent upon the CPU. - Still are, to some degree, of course. - The central processing unit. But we really saw this huge innovation a few years back in video games, where we were having increased demand for incredible visuals, real-time performance. Things just took a huge leap with the gaming industry. - Yeah, the fact is that CPUs, while obviously still vital to the performance of any system, are tasked with doing a lot, running your operating system, managing memory, doing this, or doing that. GPUs, in a lot of cases, especially for professional creative apps, tend to be a little faster, and sometimes a lot faster, than the CPUs at processing things, things like effects in your NLE, or noise reduction, or things of that nature. And so, while it might seem that a fancy GPU is just something, oh, yeah, just gamers use that. Really, in reality, a lot of application designers are offloading very complex mathematical tasks to a GPU cause it can simply do it faster and do those tasks better than on a CPU. - [Rich] Yeah, we're seeing that both photo and video apps are being optimized to take advantage of this. - So, when it comes to the actual physical hardware of a GPU, there's two main companies producing these items, NVIDIA, as well as AMD. And Rich, no doubt you've probably heard of these companies. - Yeah, and AMD used to go by ATI before that. And so, what really happens is there's the two main companies, but then you'll see subsets where they're using one company's technology sold under a different brand. - Right, sure. And there's also sort of the underlying programming technology about how the machine is communicating with the GPU. And you'll hear things like CUDA. Well, CUDA is a programming language that NVIDIA developed. You'll also hear about OpenCL. - Or Metal. - Right, or Metal from Apple. These are all programming languages, and ways of the GPU communicating with different applications. And it's a little beyond the scope of this week to discuss the ins and outs of CUDA and OpenCL and Metal, but just know that those things don't refer to the actual hardware, they refer to sort of the programming languages that are used to allow the GPUs to communicate. - And we're going to explore both laptop and desktop workflows, but one of the things that's critically important these days, if you haven't bought a new computer in a few years, you really need to pay attention to the GPU options, because, particularly in laptops, and even some desktops these days, it's not like you can upgrade this. PCs make it easy. Apple, you kind of get what you get. - That's a really important factor, Rich. With a system like a laptop or an all-in-one, you might not be able to actually swap out that GPU. But with a tower-based system, one that has PCI slots in it, you can. And Rich, that's my biggest selling point of why I switched, for my professional work, from a Mac to a Windows system, because I could swap out the GPUs. And the rate of development of GPUs is so much faster than pretty much any other piece of hardware. - Every six months, you're telling me you got a new GPU. - Right, and it's because they come out, I can swap them in, and it really does feel, in a lot of circumstances, that I'm getting a new computer, because a lot of the apps that I'm using depend on the GPU. - Well, let's go ahead and take a detailed look at both laptop and desktop decisions that you'll need to make.

Contents