This doesn't seem to work anymore ... Athena now uses native (ie en..) NIC naming.

Ubuntu 16.04LTS introduced a rather arcane naming convention for ethernet ports. To be able to continue using the traditional eth0/eth1 names I added the following file. Note that you will need to substitute the appropriate NIC addresses in the 'ATTR{address}==' fields:

/etc/udev/rules.d/70-persistent-net.rules:

# This file was automatically generated by the /lib/udev/write_net_rules
      # program, run by the persistent-net-generator.rules rules file.
      #
      # You can modify it, as long as you keep each rule on a single
      # line, and change only the value of the NAME= key.

      # PCI device  (eno1)
      SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="9c:5c:8e:50:01:nn", ATTR{dev_id}=="0x0", ATTR{type}=="1", NAME="eth0"

      # PCI device  (enp18s0)
      SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="9c:5c:8e:50:01:nn", ATTR{dev_id}=="0x0", ATTR{type}=="1", NAME="eth1"

The following technique for reserving CPUs is currently unused.

Since we want realtime response from our dfcSDR programs, we can reserve one or more CPUs exclusively for the program. Edit /etc/default/grub by adding isolcpus= to the line: GRUB_CMDLINE_LINUX_DEFAULT=... In this example I know that I have 8 cores (actually 4 dual threaded cores). So I reserve the last 2, ie. #6 & #7, by setting the line:

/etc/default/grub:

 GRUB_CMDLINE_LINUX_DEFAULT="quiet splash isolcpus=6,7"

After modifying the grub file you must update grub's database and reboot: 

$ sudo update-grub
$ sudo reboot

You can enable a dfcSDR program to use the reserved CPU(s) with this code: 

#include <sched.h>

...
    /*
     * Schedule this process to run on a specific CPU.
     */
    size_t size = CPU_ALLOC_SIZE(2);		// unsigned long int granularity,
    cpu_set_t* setPtr = CPU_ALLOC(size);	//  a 32 bit machine will = 32 CPUs minimim
    CPU_ZERO_S(size, setPtr);			//  a 64 bit machine will = 64 CPUs minimum

    /* Get list of available CPUs. */
    if (sched_getaffinity(0, size, setPtr) < 0) {
        perror("can't getaffinity");
        exit(-1);
    }
    
    /* Assume the highest numbered CPU is reserved. */
    int xx = CPU_COUNT_S(size, setPtr);		// actual number of available CPUs
    int useCpu = 0;
    for (int x = 0; x < xx; ++x) {
        if (CPU_ISSET(x, setPtr) && (x > useCpu))
            useCpu = x;
    }

    /* Set affinity. */
    CPU_ZERO_S(size, setPtr);
    CPU_SET(useCpu, setPtr);
    if (sched_setaffinity(0, size, setPtr) < 0) {
        perror("can't setaffinity");
        exit(-1);
    }

    /* Free resources. */
    CPU_FREE(setPtr);
    
...
    
    /* Verify which CPU we are using. */
    printf(stderr, "cudaLoop() running on CPU: %d", sched_getcpu());