Groq LPU architecture page - on-chip SRAM bandwidth upwards of 80 TB/s, 14 nm, no off-chip HBM

Verbatim: 'Groq on-chip SRAM has memory bandwidth upwards of 80 terabytes/second, while GPU off-chip HBM clocks in at about eight terabytes/second.' and 'Our current chip set is built on a 14 nanometer process. As we move towards a 4 nanometer process, the performance advantages of LPU architecture will only increase.' Confirms accelerators/groq-lpu.json memory_bandwidth_gbps 80000 (= 80 TB/s) and process_nm 14, and the no-off-chip-HBM claim. This page publishes NO per-chip compute (TFLOPS/TOPS) figure and NO SRAM capacity figure. The 80 TB/s vs ~8 TB/s HBM comparison is Groq's own and is presented by Groq as a 10x advantage.