Inference, not training, may be AMD’s golden ticket

AMD might be down 50 per cent in six months, but investor Yiannis Zourmpanos (pictured) is doubling down, insisting the market’s missing the point. While the chipmaker continues to trail Nvidia, he says the prize isn’t in training gargantuan AI models—it’s in running them.

Based on Stable Diffusion AI model

UL has announced its newest AI Inference benchmark, the UL Procyon AI Image Generation Benchmark, which is based on the Stable Diffusion AI model and promises to provide a consistent, accurate and understandable workload for measuring the AI performance of high-end hardware.

Out next year

Alibaba will set up a dedicated chip subsidiary and aims to launch its first self-developed AI inference chip in the second half of 2019 that could be used for autonomous driving, smart cities and logistics.