xmesh - Reports utilization percentages of EV7 based
AlphaServer systems mesh components.
/usr/bin/X11/xmesh
The xmesh application accepts all of the standard X
Toolkit command line options, which are documented in the
OPTIONS section in the X(1X) reference page.
You use the xmesh application to view a mesh, which is the
interconnection of CPUs on EV7 based AlphaServer systems.
The xmesh application reports the percent of utilization
for the hardware components associated with each CPU. A
color spectrum indicates levels of utilization for each
hardware component. There are 10 colors in the spectrum,
ranging from purple, which indicates low utilization, to
red, which indicates high utilization.
The xmesh application reports utilization for the following
EV7 based AlphaServer systems mesh components: the
CPU, the Zbox (the memory controller), the RBox I/O (Rbox
I/O port utilization), IO7 (ports 0, 1, 2, and 3), the Up
and Down Hoses, and the North, South, East and West interprocessor
ports on the EV7 chip.
You can use the information obtained from the xmesh
application to monitor data flow between CPUs, determine
resource bottlenecks and watch for non-optimal performance,
and identify applications that are memory or CPU
intensive.
Although software can treat a CC-NUMA (Cache Coherent NonUniform
Memory Access) system (such as an EV7 based
AlphaServer system) as a traditional SMP system and still
be programmatically correct, obtaining optimal performance
from a CC-NUMA system depends on appropriate use of its
capabilities. Starting with Tru64 UNIX Version 5.1, the
operating system includes kernel algorithms, utilities,
and programming APIs that are NUMA aware. These algorithms
and user interfaces are used to maximize the ratio of
local to remote memory accesses and thereby help ensure
optimal performance on CC-NUMA hardware.
If the xmesh application reveals that utilization factors
are reaching maximum capacity and there are performance
bottlenecks, you have several alternatives. For instance,
if there is too much I/O targetted to one CPU's disks,
there is contention for access to memory on a particular
CPU, or perhaps a CPU is reaching its maximum of memory
utilization, consider modifying your applications to run
on a particular CPU, allocate memory to a particular CPU,
or bind processes to a particular RAD (Resource Affinity
Domain).
Users do not require special privileges to run the xmesh
application, however, only one instance of xmesh can run
at a time.
Another xmesh is running on this system. Only one xmesh
can be run against a system.
The owner of that process is: <username>
The process id is: <pid>
Explanation:
Only one process can open the /dev/marvel_pfm file,
therefore only one user can run xmesh at a time.
User Action:
Determine who is running xmesh and ask them to exit
the application. Xmesh is not supported on this
platform. Please refer to the "xmesh" man page for
additional information.
Explanation:
The xmesh application is supported only on EV7
based AlphaServer platforms because it is of little
benefit on any other platform.
The driver that collects low-level statistics for xmesh.
Contains the xmesh help volume. Contains the default
application's X resources
Commands: runon(1)
Functions: numa_intro(3)
Files: numa_scheduling_groups(4), numa_types(4)
NUMA Overview
The NUMA Overview is a web-only document that includes a
complete NUMA programming example. Starting with Tru64
UNIX Version 5.1, this web-only document can be accessed
through the version-specific web pages for Tru64 UNIX documentation.
Links to documentation sets for different
product versions are available at the following URL: '
http://h30097.www3.hp.com/docs/'>
xmesh(1)
[ Back ] |
