“There’s a great chip already on the market. For $999, you get a 123 TFLOP card with 24 GB of 960 GB/s RAM. This is the best FLOPS per dollar today, and yet…nobody in ML uses it.”
The card is AMD’s, and geohot says the driver kernel panics with two cards and ROCm never built cleanly in three tries. His $5.1M startup will write the framework, runtime and driver AMD never did. AMD has never appeared on MLPerf because of software, not silicon. NVIDIA keeps its 80% margins because the competitor’s own software is the moat.