HPE Cray XD670 vs QCT QuantaGrid D75F-7U
Both are built on the NVIDIA HGX H200 baseboard, so their accelerator performance is identical rather than merely similar. The choice between them is chassis, cooling, expansion and supplier, not FLOPS.
What actually differs
| Specification | HPE Cray XD670 | QCT QuantaGrid D75F-7U |
|---|---|---|
| Vendor | HPE | QCT |
| Part number | — | D75F-7U |
| Form factor | rack | rack |
| Released | 2024 | — |
| Accelerator | 8x H200 | 8x H200 |
| Aggregate memory | 1.1 TB | 1.1 TB |
| Aggregate bandwidth | 38.4 TB/s | — |
| System power | — | — |
| Interconnect | NVIDIA NVLink with NVSwitch across the eight SXM modules | One NVIDIA HGX H200 baseboard connecting the eight GPUs at 900 GB/s over NVLink, which QCT notes is seven times the bandwidth of PCIe Gen 5; two BlueField-3 DPUs handle networking |
Rows shown in grey are identical between the two.
HPE Cray XD670
A 5U server carrying eight NVIDIA H200 SXM GPUs at 700 W each with 141 GB of HBM3e apiece, driven by a single two-socket node of 5th generation Intel Xeon Scalable processors at up to 400 W, with 32 DDR5 DIMMs across eight channels per socket at up to 5,600 MHz. It is air cooled with a direct liquid cooling option. HPE has taken six first-place results with it in MLPerf Inference v5.1, including computer vision and LLM chat and text generation. Two things this row deliberately does not claim: HPE describes the GPUs only as "8x NVIDIA H200 SXM" and never names the HGX baseboard, though eight H200 SXM at 700 W is precisely the HGX H200 8-GPU configuration, and the compute figures shown are NVIDIA's for that baseboard rather than any HPE claim; and HPE publishes component TDPs but no system power figure, so total power is left blank rather than summing 8 x 700 W and 2 x 400 W, which would count silicon only and understate the machine. Part numbers are also not recorded, since none appears on HPE's page.
Vendor product pageQCT QuantaGrid D75F-7U
QCT's 7U Hopper machine, with two fifth or fourth generation Intel Xeon Scalable processors and eight NVIDIA HGX H200 GPUs linked at 900 GB/s over NVLink. Storage is eighteen small-form-factor all-NVMe bays for GPU-Direct Storage plus a boot drive, and two NVIDIA BlueField-3 DPUs offload networking. QCT gives 1.1 TB of aggregate high-bandwidth memory. It also quotes over 32 petaflops of FP8, which is not shown here: QCT does not say whether that is dense or sparse, it is close to double the dense figure NVIDIA published for this baseboard, and NVIDIA no longer lists the H200 on its HGX page, so there is no way to settle which column it belongs in without guessing. Power is four-plus-two 4,000 W Titanium supplies, which is capacity rather than draw.
Vendor product page