XLA does for TensorFlow/JAX: compiles computation graphs for TPU/GPU execution

How can you speed up your favorite video game?

Imagine you're playing a fast-paced racing game on your console, and you want to hit the track as quickly as possible. You notice that sometimes the game lags, making it hard to keep up with your friends.

XLA helps by optimizing the game's code so it can run faster on your console's powerful graphics card, making the game smoother and quicker to respond to your commands.

Example

Without XLA, your game might take 10 seconds to load a new track, but with XLA, it loads in just 5 seconds.

Remember this

XLA compiles computation graphs for TPU/GPU execution, speeding up your game's performance.

Text adapted from Wikipedia, licensed under CC BY-SA 4.0.

Related concepts

Nvidia

Ever wondered how video games run so smoothly on your computer?

tensor cores are

Why can computers crunch numbers faster than humans?

List of free and open-source software packages

Can you run AI on your phone?

torch.compile does in PyTorch 2.0: traces and optimizes the computation graph

Ever wondered how computers speed up learning?

CPU cache

L1/L2 cache hierarchy reduces global memory latency

a Triton kernel is

Can your phone run faster with a different brain?

Swipe through 100 ML concepts daily

Open Pocket Polymath