1.. SPDX-License-Identifier: BSD-3-Clause 2 Copyright(c) 2010-2014 Intel Corporation. 3 4L3 Forwarding with Power Management Sample Application 5====================================================== 6 7Introduction 8------------ 9 10The L3 Forwarding with Power Management application is an example of power-aware packet processing using the DPDK. 11The application is based on existing L3 Forwarding sample application, 12with the power management algorithms to control the P-states and 13C-states of the Intel processor via a power management library. 14 15Overview 16-------- 17 18The application demonstrates the use of the Power libraries in the DPDK to implement packet forwarding. 19The initialization and run-time paths are very similar to those of the :doc:`l3_forward`. 20The main difference from the L3 Forwarding sample application is that this application introduces power-aware optimization algorithms 21by leveraging the Power library to control P-state and C-state of processor based on packet load. 22 23The DPDK includes poll-mode drivers to configure Intel NIC devices and their receive (Rx) and transmit (Tx) queues. 24The design principle of this PMD is to access the Rx and Tx descriptors directly without any interrupts to quickly receive, 25process and deliver packets in the user space. 26 27In general, the DPDK executes an endless packet processing loop on dedicated IA cores that include the following steps: 28 29* Retrieve input packets through the PMD to poll Rx queue 30 31* Process each received packet or provide received packets to other processing cores through software queues 32 33* Send pending output packets to Tx queue through the PMD 34 35In this way, the PMD achieves better performance than a traditional interrupt-mode driver, 36at the cost of keeping cores active and running at the highest frequency, 37hence consuming the maximum power all the time. 38However, during the period of processing light network traffic, 39which happens regularly in communication infrastructure systems due to well-known "tidal effect", 40the PMD is still busy waiting for network packets, which wastes a lot of power. 41 42Processor performance states (P-states) are the capability of an Intel processor 43to switch between different supported operating frequencies and voltages. 44If configured correctly, according to system workload, this feature provides power savings. 45CPUFreq is the infrastructure provided by the Linux* kernel to control the processor performance state capability. 46CPUFreq supports a user space governor that enables setting frequency via manipulating the virtual file device from a user space application. 47The Power library in the DPDK provides a set of APIs for manipulating a virtual file device to allow user space application 48to set the CPUFreq governor and set the frequency of specific cores. 49 50This application includes a P-state power management algorithm to generate a frequency hint to be sent to CPUFreq. 51The algorithm uses the number of received and available Rx packets on recent polls to make a heuristic decision to scale frequency up/down. 52Specifically, some thresholds are checked to see whether a specific core running a DPDK polling thread needs to increase frequency 53a step up based on the near to full trend of polled Rx queues. 54Also, it decreases frequency a step if packet processed per loop is far less than the expected threshold 55or the thread's sleeping time exceeds a threshold. 56 57C-States are also known as sleep states. 58They allow software to put an Intel core into a low power idle state from which it is possible to exit via an event, such as an interrupt. 59However, there is a tradeoff between the power consumed in the idle state and the time required to wake up from the idle state (exit latency). 60Therefore, as you go into deeper C-states, the power consumed is lower but the exit latency is increased. Each C-state has a target residency. 61It is essential that when entering into a C-state, the core remains in this C-state for at least as long as the target residency in order 62to fully realize the benefits of entering the C-state. 63CPUIdle is the infrastructure provide by the Linux kernel to control the processor C-state capability. 64Unlike CPUFreq, CPUIdle does not provide a mechanism that allows the application to change C-state. 65It actually has its own heuristic algorithms in kernel space to select target C-state to enter by executing privileged instructions like HLT and MWAIT, 66based on the speculative sleep duration of the core. 67In this application, we introduce a heuristic algorithm that allows packet processing cores to sleep for a short period 68if there is no Rx packet received on recent polls. 69In this way, CPUIdle automatically forces the corresponding cores to enter deeper C-states 70instead of always running to the C0 state waiting for packets. 71 72.. note:: 73 74 To fully demonstrate the power saving capability of using C-states, 75 it is recommended to enable deeper C3 and C6 states in the BIOS during system boot up. 76 77Compiling the Application 78------------------------- 79 80To compile the sample application see :doc:`compiling`. 81 82The application is located in the ``l3fwd-power`` sub-directory. 83 84Running the Application 85----------------------- 86 87The application has a number of command line options: 88 89.. code-block:: console 90 91 ./<build_dir>/examples/dpdk-l3fwd_power [EAL options] -- -p PORTMASK [-P] --config(port,queue,lcore)[,(port,queue,lcore)] [--max-pkt-len PKTLEN] [--no-numa] 92 93where, 94 95* -p PORTMASK: Hexadecimal bitmask of ports to configure 96 97* -P: Sets all ports to promiscuous mode so that packets are accepted regardless of the packet's Ethernet MAC destination address. 98 Without this option, only packets with the Ethernet MAC destination address set to the Ethernet address of the port are accepted. 99 100* --config (port,queue,lcore)[,(port,queue,lcore)]: determines which queues from which ports are mapped to which cores. 101 102* --max-pkt-len: optional, maximum packet length in decimal (64-9600) 103 104* --no-numa: optional, disables numa awareness 105 106* --empty-poll: Traffic Aware power management. See below for details 107 108* --telemetry: Telemetry mode. 109 110* --pmd-mgmt: PMD power management mode. 111 112See :doc:`l3_forward` for details. 113The L3fwd-power example reuses the L3fwd command line options. 114 115Explanation 116----------- 117 118The following sections provide some explanation of the sample application code. 119As mentioned in the overview section, 120the initialization and run-time paths are identical to those of the L3 forwarding application. 121The following sections describe aspects that are specific to the L3 Forwarding with Power Management sample application. 122 123Power Library Initialization 124~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 125 126The Power library is initialized in the main routine. 127It changes the P-state governor to userspace for specific cores that are under control. 128The Timer library is also initialized and several timers are created later on, 129responsible for checking if it needs to scale down frequency at run time by checking CPU utilization statistics. 130 131.. note:: 132 133 Only the power management related initialization is shown. 134 135.. literalinclude:: ../../../examples/l3fwd-power/main.c 136 :language: c 137 :start-after: Power library initialized in the main routine. 8< 138 :end-before: >8 End of power library initialization. 139 140Monitoring Loads of Rx Queues 141~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 142 143In general, the polling nature of the DPDK prevents the OS power management subsystem from knowing 144if the network load is actually heavy or light. 145In this sample, sampling network load work is done by monitoring received and 146available descriptors on NIC Rx queues in recent polls. 147Based on the number of returned and available Rx descriptors, 148this example implements algorithms to generate frequency scaling hints and speculative sleep duration, 149and use them to control P-state and C-state of processors via the power management library. 150Frequency (P-state) control and sleep state (C-state) control work individually for each logical core, 151and the combination of them contributes to a power efficient packet processing solution when serving light network loads. 152 153The rte_eth_rx_burst() function and the newly-added rte_eth_rx_queue_count() function are used in the endless packet processing loop 154to return the number of received and available Rx descriptors. 155And those numbers of specific queue are passed to P-state and C-state heuristic algorithms 156to generate hints based on recent network load trends. 157 158.. note:: 159 160 Only power control related code is shown. 161 162.. literalinclude:: ../../../examples/l3fwd-power/main.c 163 :language: c 164 :start-after: Main processing loop. 8< 165 :end-before: >8 End of main processing loop. 166 167P-State Heuristic Algorithm 168~~~~~~~~~~~~~~~~~~~~~~~~~~~ 169 170The power_freq_scaleup_heuristic() function is responsible for generating a frequency hint for the specified logical core 171according to available descriptor number returned from rte_eth_rx_queue_count(). 172On every poll for new packets, the length of available descriptor on an Rx queue is evaluated, 173and the algorithm used for frequency hinting is as follows: 174 175* If the size of available descriptors exceeds 96, the maximum frequency is hinted. 176 177* If the size of available descriptors exceeds 64, a trend counter is incremented by 100. 178 179* If the length of the ring exceeds 32, the trend counter is incremented by 1. 180 181* When the trend counter reached 10000 the frequency hint is changed to the next higher frequency. 182 183.. note:: 184 185 The assumption is that the Rx queue size is 128 and the thresholds specified above 186 must be adjusted accordingly based on actual hardware Rx queue size, 187 which are configured via the rte_eth_rx_queue_setup() function. 188 189In general, a thread needs to poll packets from multiple Rx queues. 190Most likely, different queue have different load, so they would return different frequency hints. 191The algorithm evaluates all the hints and then scales up frequency in an aggressive manner 192by scaling up to highest frequency as long as one Rx queue requires. 193In this way, we can minimize any negative performance impact. 194 195On the other hand, frequency scaling down is controlled in the timer callback function. 196Specifically, if the sleep times of a logical core indicate that it is sleeping more than 25% of the sampling period, 197or if the average packet per iteration is less than expectation, the frequency is decreased by one step. 198 199C-State Heuristic Algorithm 200~~~~~~~~~~~~~~~~~~~~~~~~~~~ 201 202Whenever recent rte_eth_rx_burst() polls return 5 consecutive zero packets, 203an idle counter begins incrementing for each successive zero poll. 204At the same time, the function power_idle_heuristic() is called to generate speculative sleep duration 205in order to force logical to enter deeper sleeping C-state. 206There is no way to control C- state directly, and the CPUIdle subsystem in OS is intelligent enough 207to select C-state to enter based on actual sleep period time of giving logical core. 208The algorithm has the following sleeping behavior depending on the idle counter: 209 210* If idle count less than 100, the counter value is used as a microsecond sleep value through rte_delay_us() 211 which execute pause instructions to avoid costly context switch but saving power at the same time. 212 213* If idle count is between 100 and 999, a fixed sleep interval of 100 μs is used. 214 A 100 μs sleep interval allows the core to enter the C1 state while keeping a fast response time in case new traffic arrives. 215 216* If idle count is greater than 1000, a fixed sleep value of 1 ms is used until the next timer expiration is used. 217 This allows the core to enter the C3/C6 states. 218 219.. note:: 220 221 The thresholds specified above need to be adjusted for different Intel processors and traffic profiles. 222 223If a thread polls multiple Rx queues and different queue returns different sleep duration values, 224the algorithm controls the sleep time in a conservative manner by sleeping for the least possible time 225in order to avoid a potential performance impact. 226 227Empty Poll Mode 228------------------------- 229Additionally, there is a traffic aware mode of operation called "Empty 230Poll" where the number of empty polls can be monitored to keep track 231of how busy the application is. Empty poll mode can be enabled by the 232command line option --empty-poll. 233 234See :doc:`Power Management<../prog_guide/power_man>` chapter in the DPDK Programmer's Guide for empty poll mode details. 235 236.. code-block:: console 237 238 ./<build_dir>/examples/dpdk-l3fwd-power -l xxx -n 4 -a 0000:xx:00.0 -a 0000:xx:00.1 \ 239 -- -p 0x3 -P --config="(0,0,xx),(1,0,xx)" --empty-poll="0,0,0" -l 14 -m 9 -h 1 240 241Where, 242 243--empty-poll: Enable the empty poll mode instead of original algorithm 244 245--empty-poll="training_flag, med_threshold, high_threshold" 246 247* ``training_flag`` : optional, enable/disable training mode. Default value is 0. If the training_flag is set as 1(true), then the application will start in training mode and print out the trained threshold values. If the training_flag is set as 0(false), the application will start in normal mode, and will use either the default thresholds or those supplied on the command line. The trained threshold values are specific to the user’s system, may give a better power profile when compared to the default threshold values. 248 249* ``med_threshold`` : optional, sets the empty poll threshold of a modestly busy system state. If this is not supplied, the application will apply the default value of 350000. 250 251* ``high_threshold`` : optional, sets the empty poll threshold of a busy system state. If this is not supplied, the application will apply the default value of 580000. 252 253* -l : optional, set up the LOW power state frequency index 254 255* -m : optional, set up the MED power state frequency index 256 257* -h : optional, set up the HIGH power state frequency index 258 259Empty Poll Mode Example Usage 260~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 261To initially obtain the ideal thresholds for the system, the training 262mode should be run first. This is achieved by running the l3fwd-power 263app with the training flag set to “1”, and the other parameters set to 2640. 265 266.. code-block:: console 267 268 ./<build_dir>/examples/dpdk-l3fwd-power -l 1-3 -- -p 0x0f --config="(0,0,2),(0,1,3)" --empty-poll "1,0,0" –P 269 270This will run the training algorithm for x seconds on each core (cores 2 271and 3), and then print out the recommended threshold values for those 272cores. The thresholds should be very similar for each core. 273 274.. code-block:: console 275 276 POWER: Bring up the Timer 277 POWER: set the power freq to MED 278 POWER: Low threshold is 230277 279 POWER: MED threshold is 335071 280 POWER: HIGH threshold is 523769 281 POWER: Training is Complete for 2 282 POWER: set the power freq to MED 283 POWER: Low threshold is 236814 284 POWER: MED threshold is 344567 285 POWER: HIGH threshold is 538580 286 POWER: Training is Complete for 3 287 288Once the values have been measured for a particular system, the app can 289then be started without the training mode so traffic can start immediately. 290 291.. code-block:: console 292 293 ./<build_dir>/examples/dpdk-l3fwd-power -l 1-3 -- -p 0x0f --config="(0,0,2),(0,1,3)" --empty-poll "0,340000,540000" –P 294 295Telemetry Mode 296-------------- 297 298The telemetry mode support for ``l3fwd-power`` is a standalone mode, in this mode 299``l3fwd-power`` does simple l3fwding along with calculating empty polls, full polls, 300and busy percentage for each forwarding core. The aggregation of these 301values of all cores is reported as application level telemetry to metric 302library for every 500ms from the main core. 303 304The busy percentage is calculated by recording the poll_count 305and when the count reaches a defined value the total 306cycles it took is measured and compared with minimum and maximum 307reference cycles and accordingly busy rate is set to either 0% or 30850% or 100%. 309 310.. code-block:: console 311 312 ./<build_dir>/examples/dpdk-l3fwd-power --telemetry -l 1-3 -- -p 0x0f --config="(0,0,2),(0,1,3)" --telemetry 313 314The new stats ``empty_poll`` , ``full_poll`` and ``busy_percent`` can be viewed by running the script 315``/usertools/dpdk-telemetry-client.py`` and selecting the menu option ``Send for global Metrics``. 316 317PMD power management Mode 318------------------------- 319 320The PMD power management mode support for ``l3fwd-power`` is a standalone mode. 321In this mode, ``l3fwd-power`` does simple l3fwding 322along with enabling the power saving scheme on specific port/queue/lcore. 323Main purpose for this mode is to demonstrate 324how to use the PMD power management API. 325 326.. code-block:: console 327 328 ./build/examples/dpdk-l3fwd-power -l 1-3 -- --pmd-mgmt -p 0x0f --config="(0,0,2),(0,1,3)" 329 330PMD Power Management Mode 331------------------------- 332 333There is also a traffic-aware operating mode that, 334instead of using explicit power management, 335will use automatic PMD power management. 336This mode is limited to one queue per core, 337and has three available power management schemes: 338 339``monitor`` 340 This will use ``rte_power_monitor()`` function to enter 341 a power-optimized state (subject to platform support). 342 343``pause`` 344 This will use ``rte_power_pause()`` or ``rte_pause()`` 345 to avoid busy looping when there is no traffic. 346 347``scale`` 348 This will use frequency scaling routines 349 available in the ``librte_power`` library. 350 351See :doc:`Power Management<../prog_guide/power_man>` chapter 352in the DPDK Programmer's Guide for more details on PMD power management. 353 354.. code-block:: console 355 356 ./<build_dir>/examples/dpdk-l3fwd-power -l 1-3 -- -p 0x0f --config="(0,0,2),(0,1,3)" --pmd-mgmt=scale 357