HPE ProLiant Compute XD685 (HGX B200) vs Penguin Solutions Altus XE4318GTS-DTC

Both are built on the NVIDIA HGX B200 8-GPU 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

SpecificationHPE ProLiant Compute XD685 (HGX B200)Penguin Solutions Altus XE4318GTS-DTC
VendorHPEPenguin Solutions
Part numberS4R96AAltus XE4318GTS-DTC
Form factorrackrack
Released2025
Accelerator8x B2008x B200
Aggregate memory1.4 TB
Aggregate bandwidth64.0 TB/s
System power
InterconnectNVLink 5.0 / NVSwitch 4.0 across the HGX B200 baseboard; 4x FHHL PCIe Gen5 x16 slots for fabric, NIC or DPUPenguin publishes no interconnect figure; expansion is 8x PCIe Gen5 x16 FHHL and 4x PCIe Gen5 x16 FHHL double-width

Rows shown in grey are identical between the two.

HPE ProLiant Compute XD685 (HGX B200)

The eight-way XD685 in its NVIDIA HGX B200 configuration, liquid cooled only, in the 5U DLC chassis S4R96A. HPE names B200 support twice in the QuickSpecs, in the supported-accelerator list and in the rule that the H200 and B200 liquid-cooled accelerators must be paired with this chassis, so the configuration is not in doubt. The accelerator kit part number is not recorded: the accelerator table lists two entries both described as HGX B300 with different part numbers, one of which is likely the B200 mislabelled, and guessing which would put an unverified number on a row buyers may quote. Two fifth-generation AMD EPYC processors, 24 DIMM slots and up to 6 TB of DDR5-6400. HPE publishes no system power draw, only twelve 3,000 W supplies, so nothing is summed; compute is NVIDIA's for the baseboard.

Vendor product page

Penguin Solutions Altus XE4318GTS-DTC

The Penguin Solutions Altus XE4318GTS-DTC is a 4U GPU server carrying eight NVIDIA HGX B200 SXM modules of 180 GB each, hosted by AMD EPYC 9005/9004 Series processors with 8x PCIe Gen5 x16 FHHL and 4x PCIe Gen5 x16 FHHL double-width. The DTC suffix marks Penguin's direct-to-chip liquid cooling, which is what folds an eight-accelerator Blackwell baseboard into 4U where the air-cooled equivalent needs 8U. Penguin publishes only rack height, host processor, expansion slots and the accelerator for its servers, so no power draw, total memory, interconnect bandwidth or throughput figure is shown here; its full datasheets are hosted off its own domain and were not used as a source.

Vendor product page