Gigabyte G593-SD1-LAX3 vs HPE Cray XD670
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 | Gigabyte G593-SD1-LAX3 | HPE Cray XD670 |
|---|---|---|
| Vendor | Gigabyte | HPE |
| Part number | 6NG593SD1DR000LBX3 | — |
| Form factor | rack | rack |
| Released | 2024 | 2024 |
| Accelerator | 8x H200 | 8x H200 |
| Aggregate memory | 1.1 TB | 1.1 TB |
| Aggregate bandwidth | 38.4 TB/s | 38.4 TB/s |
| System power | — | — |
| Interconnect | NVIDIA NVLink with NVSwitch, 900 GB/s GPU-to-GPU | NVIDIA NVLink with NVSwitch across the eight SXM modules |
Rows shown in grey are identical between the two.
Gigabyte G593-SD1-LAX3
A 5U direct-liquid-cooled server carrying the NVIDIA HGX H200 baseboard with eight SXM GPUs, connected at 900 GB/s by NVLink and NVSwitch. Both CPUs and GPUs are liquid cooled, which is what lets Gigabyte fit the platform into 5U rather than the 8U an air-cooled equivalent needs. Two LGA 4677 sockets take 5th or 4th generation Intel Xeon Scalable processors or the Xeon CPU Max series, with 32 DIMM slots across eight DDR5 channels, eight Gen5 NVMe bays, twelve PCIe Gen5 x16 slots and BlueField-3 DPU compatibility. Power comes from four plus two redundant 3000 W Titanium supplies. It weighs 91 kg net and orders as 6NG593SD1DR000LBX3. Note that Gigabyte's page quotes "over 32 petaflops of FP8 deep learning compute" for an eight-way HGX H200; that is NVIDIA's own marketing figure for the baseboard rather than a Gigabyte measurement, and it is the sparse number. The dense figures shown here are NVIDIA's for the same baseboard, and they are what this site publishes throughout.
Vendor product pageHPE 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 page