Processor Assignment

On a multiprocessor machine that is configured for RTX dedicated mode, you configure how many of the processors are dedicated to Windows and how many are dedicated to the Real-Time Subsystem (RTSS). Depending upon the edition of RTX Runtime, up to 31 processors can be dedicated to RTSS. If your system has more than eight processors, or if clustering is required by the hardware in a system with eight or fewer processors, the system will be set to dedicated cluster mode. In a dedicated cluster mode system, only four processors can be dedicated to Windows.

You use the RTX Properties control panel to configure the number of processors dedicated to Windows and the number dedicated to RTSS.

When configuring processors, it is helpful to remember the following rules:

Windows processors are assigned first, starting with Processor 0, and then RTSS processors are assigned.
The total number of RTSS processors allowed depends upon the RTX runtime edition installed on your system. See your Installation Guide for more information on RTX editions.
A maximum of four processors can be dedicated to Windows on a dedicated cluster mode system.

Processors are identified by their system-wide processor number when using an affinity mask within an RTAPI call. For example, in the following illustration three processors are dedicated to Windows and five are dedicated to RTSS. Physically, this configuration translates into dedicating Processors 0, 1, and 2 to Windows and Processors 3 through 7 to RTSS.

The sum of processors dedicated to Windows and to RTSS should equal the total number of processors in your system. If you do not configure your system in this way, the unassigned processors will not be used by either Windows or RTSS and you will lose valuable processors.

On systems with multi-core processors, Windows cores are assigned first, beginning with the first core in each multi-core processor. The following configuration illustrates two quad-core processors for which five cores are dedicated to RTSS.

Each execution core in a multi-core processor can access the processor's L2 cache. This means that if one core has minimal cache requirements, other cores can increase the percentage of L2 cache available to them.