Coffee with Developers - Stephen Jones - NVIDIA

What happens when a chip gets as hot as the sun? An NVIDIA architect explains how CUDA is solving the power wall problem in modern computing.

#1about 2 minutes

Gaining perspective by using the products you build

Transitioning from a creator to a user of CUDA provides critical insights and humility by revealing the incorrect assumptions made during development.

#2about 3 minutes

Understanding CUDA as a complete computing platform

CUDA has evolved from a low-level language into a comprehensive platform of compilers, libraries, and SDKs that enable GPU access for multiple languages.

#3about 2 minutes

Supporting legacy languages like Fortran for scientific computing

CUDA supports languages like Fortran to accelerate existing codebases in supercomputing for fields such as physics and weather forecasting.

#4about 4 minutes

Why Python became the dominant language for AI

Python's large ecosystem, developer productivity, and vast talent pool made it the de facto language for AI, creating new challenges for parallel computing platforms.

#5about 3 minutes

The challenge of aligning long hardware and short software cycles

Developing new chips takes years of predictive work, creating a challenge to meet the rapidly changing demands of software, especially in the AI space.

#6about 3 minutes

How unexpected user adoption drives technological evolution

Technology evolves organically as users find novel applications for existing tools, such as using gaming GPUs for scientific computing and AI.

#7about 3 minutes

Why AI optimizations increase the demand for compute

Advances that make AI models cheaper or more efficient don't reduce overall compute demand; instead, they enable the creation of even larger and more powerful models.

#8about 3 minutes

The end of Moore's Law is a power consumption problem

While transistor density still doubles, the power per transistor is not halving, creating a thermal and power delivery bottleneck for chip performance.

#9about 6 minutes

The future of computing requires scaling out to data centers

Overcoming power limitations requires moving from single-chip optimization to building large, networked, data-center-scale systems with specialized hardware.

#10about 4 minutes

The rise of neural and quantum computing paradigms

The future of computing will be a hybrid model combining classical, neural, and quantum approaches to solve complex problems using the best tool for each task.

#11about 3 minutes

How developers can contribute to the open source CUDA ecosystem

While low-level drivers are proprietary, the vast majority of CUDA's higher-level libraries like Rapids and Cutlass are open source and welcome community contributions.