NVIDIA HGX B300 8-GPU vs QCT QuantaGrid D75L-3U
Accelerator configuration, aggregate memory, bandwidth and power, side by side.
Specifications
| Specification | NVIDIA HGX B300 8-GPU | QCT QuantaGrid D75L-3U |
|---|---|---|
| Vendor | NVIDIA | QCT |
| Part number | — | D75L-3U |
| Form factor | baseboard | rack |
| Released | 2025 | — |
| Accelerator | 8x B300 | 8x B300 |
| Aggregate memory | 2.1 TB | 2.3 TB |
| Aggregate bandwidth | 64.0 TB/s | — |
| System power | — | — |
| Interconnect | NVLink 5.0 / NVSwitch 4.0 | One NVIDIA HGX B300 8-GPU baseboard with fifth-generation NVLink at 1.8 TB/s; eight onboard ConnectX-8 SuperNICs at up to 800Gb/s for east-west traffic, with up to three further NICs for north-south |
Rows shown in grey are identical between the two.
NVIDIA HGX B300 8-GPU
The eight-GPU Blackwell Ultra baseboard OEMs build around. NVIDIA publishes both a dense and a sparse figure for every tensor precision here, and for FP4 the two are not a factor of two apart: dense rose by half over the B200, from 72 to 108 PFLOPS, while sparse stayed at 144. Total memory is NVIDIA's own 2.1 TB rather than the 2.304 TB that eight 288 GB GPUs would suggest. Memory bandwidth of 64 TB/s is the one derived number, being eight times the B300's published 8 TB/s, because NVIDIA gives no board-level HBM figure. No INT8 row is recorded: NVIDIA's table gives 3 POPS sparse for this board while our B300 chip entry implies 36 POPS dense, the two cannot both be right, and the question is open rather than guessed at. NVIDIA publishes no baseboard power figure, so total power is blank.
Vendor product pageQCT QuantaGrid D75L-3U
QCT's density play: the same NVIDIA HGX B300 8-GPU baseboard as its 10U sibling, compressed into 3U by liquid cooling and designed for the NVIDIA MGX rack rather than a standalone chassis. It takes power from a 1400A busbar and rack power shelves instead of its own supplies, which is why no power figure appears here at all, and supports DC-SCM. Fifth-generation NVLink runs at 1.8 TB/s, and eight onboard ConnectX-8 SuperNICs handle east-west traffic at up to 800Gb/s. Hosting is two Intel Xeon 6700 P-core processors at up to 350 W. QCT states 2.3 TB of HBM3e per system, which is eight B300 modules at 288 GB apiece; note NVIDIA quotes 2.1 TB for the same baseboard, and this entry keeps QCT's figure since QCT is the one publishing it. No throughput figure is given for this model.
Vendor product page