Live updates
OpenAI says its models improved AMD's flagship GPU design by up to 14%, 'roughly a year of progress' in under five months
First seen on X 20 hours ago@OpenAI ♥ 1,92718 views
OpenAI says its AI models improved the design of AMD's flagship GPU by up to 14%, beating a baseline AMD engineers had spent years optimizing, in less than five months. AMD has already put some of the changes into a chip design it recently sent for manufacturing.
According to an OpenAI blog post on Nov 18, the model rewrote the code that describes the chip's circuits (SystemVerilog RTL), improving power, performance and area (PPA) by up to 14% over heavily optimized baselines. OpenAI called it "roughly a year of progress achieved in less than 5 months" and "unprecedented results at the frontier of chip design." AMD added other changes to production design files for future versions and verified single-digit PPA gains across key blocks of its next-generation flagship CPU. The same approach was used on Jalapeño, OpenAI's first custom inference chip: model-proposed changes cut standard-cell area by about 13% in the vector unit and 10% in the matrix unit, proven in silicon. One multiplier redesign used about 15% less area while cutting latency from two clock cycles to one. In 9,881 scored runs on open-source RISC-V cores, GPT-6.2 Astra scored highest. OpenAI is also training GPT-Synopsys with chipmaking-software firm Synopsys to run its design tools the way expert engineers do.
Comments 0
Select a sentence in the story (long-press on phones) to comment on it.
“Working with @AMD, our models identified up to 14% PPA improvement on flagship GPU IP in less than five months—building on designs AMD’s engineers had already spent year…”
