XNews

OpenAI Models Deliver 14% PPA Gain on AMD Flagship GPU IP

OpenAI Models Deliver 14% PPA Gain on AMD Flagship GPU IP

OpenAI's models identified up to 14% PPA improvement on AMD's flagship GPU IP in under five months, building on designs AMD engineers had optimized for years. The same approach improved Jalapeño, OpenAI's first custom inference chip, with gains verified on real silicon, and some results were included in a recently taped-out design.

Event version 1.

Key insights

PPA gain on AMD GPU IP

OpenAI's models identified up to 14% PPA improvement on AMD's flagship GPU IP in less than five months. The work built on designs AMD engineers had already spent years optimizing.

Jalapeño inference chip gains

The same approach helped improve Jalapeño, OpenAI's first custom inference chip. The gains were verified on real silicon.

Taped-out design includes results

Some of the results were included in a design that has recently been taped out. OpenAI described the outcome as roughly an additional year of progress on AMD's chip achieved in less than five months.

Synopsys collaboration

OpenAI is working with Synopsys to bring its model capabilities and Synopsys' EDA design and validation tools to more chip designers.

Source posts