Skip to content

The single-card ceiling

The single-card ceiling

rows train() wall peak device mem throughput train r2 (1M sample)
300M 328s 42.2 GiB 54.8M rows/s 0.8305
400M 414s 56.0 GiB 57.9M rows/s 0.8303
450M 465s 63.0 GiB 58.0M rows/s 0.8340
500M 513s 69.9 GiB 58.5M rows/s 0.8329

Source: single-card-ceiling-2026-07.jsonl. Single-pod ladder (2026-07-18, NVIDIA A100-SXM4-80GB): a 500M x 100 float32 matrix trains end to end on one 80GB card at 69.9 GiB peak, 60 rounds in 8.5 minutes, with the device-resident objective keeping the fit loop bus-free. Evidence: benchmarks/single-card-ceiling-2026-07.md.