B200
NVIDIA's Blackwell-generation GPU as shipped in HGX B200 air- or liquid-cooled nodes — the tier below the flagship NVL72 rack systems.
Current numbers
~8 → 19.2 TB/sPer-GPU HBM bandwidth, HBM3E (B200/B300) to HBM4 (shipping Vera Rubin, 8 stacks): ~2.4x
80 → 288 GBPer-GPU HBM on shipping parts: H100 80 GB, HGX B200 180 GB, B300 288 GB, Vera Rubin 288 GB
A4/B200: 2× FP8; 4× FP4Google A4/B200 dense tensor peak ratios on one published SKU; not application speedup