I wrote a custom CUDA inference engine to run Qwen3.5-27B on $130 mining cards

I bought four NVIDIA CMP 100-210 cards off the secondhand market for about $80 each. They are ex-mining cards based on the In practice, NVIDIA had crippled them in hardware. The throttle The CMP 100-210 has its tensor cores throttled 64×. HMMA latency is stretched from 8 cycles to 512. cuBLAS WMMA c