pqos, pqos-msr, pqos-os - Intel(R) Resource Director Technology monitoring and
Intel(R) Resource Director Technology is designed to monitor and manage cpu
resources and improve performance of applications and virtual machines.
Intel(R) Resource Director Technology includes monitoring and control
technologies. Monitoring technologies include CMT (Cache Monitoring
Technology), which monitors occupancy of last level cache, and MBM (Memory
Bandwidth Monitoring). Control technologies include CAT (Cache Allocation
Technology) and CDP (Code Data Prioritization).
pqos supports CMT and MBM on a per core or hardware thread basis. MBM supports
two types of events reporting local and remote memory bandwidth.
pqos-msr and pqos-os are simple pqos wrapper scripts that automatically select
the MSR or OS/Kernel library interface to program the technologies.
Please see the -I option below for more information.
For hardware information please refer to the README located on:
pqos options are as follow:
- -h, --help
- show help
- -v, --verbose
- verbose mode
- -V, --super-verbose
- super-verbose mode
- -l FILE, --log-file=FILE
- log messages into selected log FILE
- -s, --show
- show the current allocation and monitoring
- -f FILE, --config-file=FILE
- load commands from selected configuration FILE
- -e CLASSDEF, --alloc-class=CLASSDEF
- define the allocation classes on all CPU sockets. CLASSDEF
format is "TYPE:ID=DEFINITION;...".
define classes for selected CPU resources. CLASSDEF format is
For CAT, TYPE is "llc" for the last level cache (aka l3) and
"l2" for level 2 cache, ID is a CLOS number and DEFINITION is a
For MBA, TYPE is "mba", ID is a CLOS number and DEFINITION is a
value between 1 and 100 representing the percentage of available memory
RESOURCE_ID is a unique number that can represent a socket or l2/l3 cache
identifier. The RESOURCE_ID for each logical CPU can be found using
Note: When CDP is on, ID can be postfixed with 'D' for data or 'C' for
code. CDP is not available for l2.
"-e l2:2=0x3f" means that COS2 for all L2 cache clusters is changed to
"-e l2@2:1=0xf" means that COS1 for L2 cache cluster 2 is changed to
"-e mba:1=30" means that COS1, on all sockets, can utilize up to 30%
of available memory bandwidth.
- -a CLASS2ID, --alloc-assoc=CLASS2ID
- associate allocation classes with cores. CLASS2ID format is
"TYPE:ID=CORE_LIST;..." or "TYPE:ID=TASK_LIST;...".
For CAT, TYPE is "llc", "core" or "pid" and ID
is a class number. CORE_LIST is comma or dash separated list of cores.
TASK_LIST is comma or dash separated list of process/task ID's.
"-a llc:0=0,2,4,6-10;llc:1=1;" associates cores 0, 2, 4, 6, 7, 8, 9,
10 with CAT class 0 and core 1 with class 1.
"-a core:0=0,2,4,6-10;core:1=1;" associates cores 0, 2, 4, 6, 7, 8, 9,
10 with CAT class 0 and core 1 with class 1.
"-I -a pid:0=3543,7643,4556;pid:1=7644;" associates process ID 3543,
7643, 4556 with CAT class 0 and process ID 7644 with class 1.
The -I option must be used for PID association.
- -R [CONFIG], --alloc-reset[=CONFIG]
- reset allocation setting (L3 CAT, L2 CAT) and reconfigure
L3 CDP. CONFIG is one of the following options:
l3cdp-on sets L3 CDP on
l3cdp-off sets L3 CDP off
l3cdp-any keep current L3 CDP setting (default)
- -m EVTCORES, --mon-core=EVTCORES
- select the cores and events for monitoring, EVTCORES format
is "EVENT:CORE_LIST". Valid EVENT settings are:
- "llc" for CMT (LLC occupancy)
- "mbr" for MBR (remote memory bandwidth)
- "mbl" for MBL (local memory bandwidth)
- "all" or "" for all detected event types
CORE_LIST is comma or dash separated list of cores.
Example "-m all:0,2,4-10;llc:1,3;mbr:11-12".
Core statistics can be grouped by enclosing the core list in square
Example "-m llc:[0-3];all:[4,5,6];mbr:[0-3],7,8".
- -p EVTPIDS, --mon-pid=EVTPIDS
- select the process ids and events to monitor, EVTPIDS
format is "EVENT:PID_LIST". See -m option for valid EVENT
settings. PID_LIST is comma separated list of process ids.
Examples "-p llc:22,25673" or "-p all:892,4588-4592".
Note: it is not possible to track both processes and cores at the same
Note: Process tracking depends on the library to be built with PID
- -T, --mon-top
- enable top like monitoring output sorted by highest LLC
- -o FILE, --mon-file FILE
- select output FILE to store monitored data in, the default
- -u TYPE, --mon-file-type=TYPE
- select the output format TYPE for monitored data. Supported
TYPE settings are: "text" (default), "xml" and
- -i INTERVAL, --mon-interval=INTERVAL
- define monitoring sampling INTERVAL in 100ms units,
1=100ms, default 10=10x100ms=1s
- -t SECONDS, --mon-time=SECONDS
- define monitoring time in seconds, use 'inf' or 'infinite'
for infinite monitoring. Use CTRL+C to stop monitoring at any time.
- -r, --mon-reset
- reset monitoring and use all RMID's and cores in the
- -H, --profile-list
- list supported allocation profiles
- -c PROFILE, --profile-set=PROFILE
- select a PROFILE from predefined allocation classes, use -H
to list available profiles
- -I, --iface-os
- set the library interface to use the kernel implementation.
If not set the default implementation is to program the MSR's
CMT, MBM and CAT are configured using Model Specific Registers (MSRs). The pqos
software executes in user space, and access to the MSRs is obtained through a
standard Linux* interface. The msr file interface is protected and requires
root privileges. The msr driver might not be auto-loaded and on some modular
kernels the driver may need to be loaded manually:
sudo modprobe msr
sudo kldload cpuctl
If you require system wide interface enforcement you can do so by setting the
"RDT_IFACE" environment variable.
pqos was written by Tomasz Kantecki <firstname.lastname@example.org>, Marcel
Cornu <email@example.com>, Aaron Hetherington
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR