NVLink-C2C coherent host-GPU bandwidth (Grace–Blackwell), bidirectional aggregate
~900 GB/sobserved
| Value kind | observed — Reported measurements, counts, and specifications keep the precision and scope stated by their source; an exact specification is not treated as a range. |
|---|---|
| Scope | 900 GB/s is the bidirectional aggregate. Compare only against bidirectional PCIe figures (~128 GB/s Gen5 x16, ~256 GB/s Gen6 x16), giving ~7x and ~3.5x — not 7–14x. |
| As of | 2025 |
| Source | NVIDIA; arXiv 2411.02814 (coherent fabrics) |
| Review | checking…review by 2026-09-13 · standard cadence |
| Recorded changes | last 2026-06-29 |
| Claim id | nvlink-c2c-coherent-host-gpu-bandwidth-grace |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.