HP Apollo 2000 System, Maintenance And Service Manual

Page: 97 / 178

8. Install the air baffle.
9. Install any removed PCI riser cage assemblies.
10. Install the node into the chassis.
11. Connect all peripheral cables to the node .
12. Power up the node .
Accelerators
Accelerator population rules
Observe the following population rules when installing an accelerator in the node:
• The FlexibleLOM 2U node riser cage assembly supports one accelerator option. Install the accelerator into
slot 2.
• A three-slot riser cage assembly supports a maximum of two accelerators.
◦ If installing a single NVIDIA GRID K2 RAF GPU, NVIDIA Tesla K80 GPU, NVIDIA Tesla M60 GPU,
NVIDIA Tesla M40 GPU, NVIDIA Tesla P40 GPU, or NVIDIA Tesla P100 GPU, install it into slot 3 and
leave slot 4 empty.
◦ For all other accelerator models, and to ensure proper thermal cooling, populate both slot 3 and slot 4
with accelerators.
◦ The mixing of different accelerator models is not supported.
• If installing an NVIDIA Tesla P100 GPU or NVIDIA Tesla P40 GPU, note the following guidelines and
limitations:
◦ The HPE Apollo r2800 Chassis does not support server nodes installed with NVIDIA Tesla P100 GPUs.
◦ If NVIDIA Tesla P40 GPUs are installed in the server node, and the server node is installed in the HPE
Apollo r2200 Chassis, the inlet ambient temperature must be maintained at or below 30°C (86°F).
◦ If NVIDIA Tesla P40 GPUs are installed in the server node, and the server node is installed in the HPE
Apollo r2600 Chassis or HPE Apollo r2800 Chassis, the inlet ambient temperature must be maintained
at or below 25°C (77°F).
◦ If NVIDIA Tesla P100 GPUs are installed in the server node, and the server node is installed in the HPE
Apollo r2200 Chassis, the inlet ambient temperature must be maintained at or below 24°C (75.2°F).
Accelerators 97