IBM Power Systems 8001-22C, Manual

Page: 88 / 138

Cables
Mellanox LinkX 10 Gb/s, 25 Gb/s, 40 Gb/s, 50 Gb/s and 100 Gb/s Direct Attach Copper cables
(DACs), Copper Splitter cables, Active Optical Cables (AOCs) and Transceivers. For information
regarding Mellanox LinkX cables and Transceivers, see LinkX
™
InfiniBand Direct Attach Copper
Cables. (http://www.mellanox.com/products/interconnect/infiniband-copper-cables.php)
Operating system or partition requirements
If you are installing a new feature, ensure that you have the software that is required to support the new
feature and that you determine whether any prerequisites must be met for this feature and attaching
devices. To check for the prerequisites, see IBM Prerequisite website (www-912.ibm.com/e_dir/
eServerPrereq.nsf).
The adapter is supported on the following versions of the operating systems:
Important:
For a list of known Linux OS issues that may affect this adapter, see the IBM Power LC
readme. (https://ibm.biz/BdrgTZ)
v Linux
– Red Hat Enterprise Linux 7.2, little-endian, for Power, or later
– Ubuntu 16.04.1, or later
– Ubuntu 14.04.5, or later
– For support details, see the Linux Alert website (www14.software.ibm.com/webapp/set2/sas/f/
lopdiags/info/LinuxAlerts.html).
– The latest version of Mellanox OFED can be downloaded from Mellanox. (http://
www.mellanox.com)
NVIDIA Tesla P100 16GB PCIe3.0 x16 Dual-slot FHFL with cables (FC EKAZ)
Learn about the specifications and operating system requirements for the feature code (FC) EKAZ
adapter.
Overview
The NVIDIA Tesla P100 16GB PCIe3.0 x16 Dual-slot FHFL with cables is a PCI Express (PCIe) generation
3 (Gen3) x16 adapter. The adapter has a double-wide, full-height form-factor and takes up two x16 PCIe
slots in the system. The NVIDIA Tesla P100 16 GB GPU Accelerator can deliver up to 18.7 teraflops
(TFLOPS) of half-precision and is capable of 720 GB/s of memory bandwidth. The accelerator is ideal for
high-performance-computing (HPC) and hyperscale workloads.
72
Power Systems: Installing the IBM Power System S822LC for Big Data (8001-22C)