And I certainly can't imagine how a language model would be of any use here, in a problem which doesn't involve language.
They are "okay" at generating RTL, but are likely never going to be able to generate actual bitstreams without some classical implementation flow in there.
I think in theory, given terabytes of bitstreams, you might be able to get an LLM to output valid designs. Excepting hardened IP blocks, a bitstream is literally a sequence of sram configuration bits to set the routing tables and LUTs. Given the right type of positional encoding I think you could maybe get simple designs working at a small scale.
Re:The FPGA Turns 40!
by CamperBob2 on Monday June 23, 2025 @09:08PM
(#44361898)
I'd expect a diffusion model to outperform autoregressive LLMs dramatically.
Certainly possible! Or perhaps a block diffusion+autoregressive model or something like GPT 4o's image gen.
Re:The FPGA Turns 40!
by checker659 on Tuesday June 24, 2025 @09:05AM
(#44365795)
AI could use EDA tools
Placement and routing is an NP-Complete problem.