Table of Contents
Advertisement
7. Install the access panel.
8. Install the server into the chassis.
9. Power up the server .
GPU accelerator options
This section provides instructions for installing GPU accelerator options into the server.
To accurately estimate the power consumption of the server and select the appropriate power configuration
and other system components, see the Hewlett Packard Enterprise Power Advisor website.
For more information about product features, specifications, options, configurations, and compatibility, see the
product QuickSpecs on the Hewlett Packard Enterprise website .
Supported GPU accelerator power cables
GPU accelerator power cables are installed in the server. Before installing GPU accelerators, ensure that the
supported power cables are installed.
•
To install NVIDIA Tesla K40 GPUs or AMD FirePro S9150 GPUs, four long GPU power cables (PN
857501-001) and four short GPU power cables (PN 853655-001) are required.
•
To install NVIDIA Tesla K80 GPUs, NVIDIA Tesla M40 GPUs, NVIDIA Tesla P40 GPUs, or NVIDIA Tesla
P100 GPUs, four long GPU power cables (PN 857500-001) and four short GPU power cables (PN
853650-001) are required.
For more information, see "GPU accelerator power cabling."
GPU accelerator population guidelines
Population guidelines vary depending on the type of PCI riser modules installed and the number of GPU
accelerators installed.
•
High-performance mode enables optimal communication between the GPU and the server processor.
•
Peer-to-peer mode enables optimal GPU-to-GPU communication.
•
The mixing of different GPU accelerator models is not supported.
•
If three or more NVIDIA Tesla K80 GPUs or NVIDIA Tesla P40 GPUs are installed on one side of the HPE
ProLiant XL270d Gen9 Accelerator Tray, the inlet ambient temperature must be maintained at or below
30°C (86°F).
GPU accelerator options
79
Hide quick links:
Advertisement
Table of Contents
