Accelerated Computing
FIELD GUIDE
§ 08
Timeline

Twenty-five years, in seven beats.

2001

Programmable shaders

DirectX 8 vertex and pixel shaders ship on consumer GPUs. Researchers begin smuggling general computation into pixel shaders.

2007

CUDA released

A C-like language to address the GPU directly. Scientific computing on GPUs becomes practical.

2012

AlexNet

A convolutional network trained on two GPUs wins ImageNet by a wide margin. Deep learning becomes the field's gravitational center.

2016

TPU v1

Google ships a custom matrix-multiply ASIC for inference, then training. The age of domain-specific silicon begins in earnest.

2020

Transformer scaling laws

Empirical work shows model quality scales predictably with compute. The accelerator becomes a strategic asset.

2023

Generative wave

Foundation models, agents, and image generation drive accelerator demand into a multi-year shortage.

2026

Heterogeneous everywhere

NPUs ship in nearly every new laptop and phone. The accelerator is no longer a server peripheral.