| /dpdk/lib/latencystats/ |
| H A D | rte_latencystats.c | 160 float latency[nb_pkts]; in calc_latency() local 172 latency[cnt++] = now - *timestamp_dynfield(pkts[i]); in calc_latency() 188 glob_stats->jitter += (fabsf(prev_latency - latency[i]) in calc_latency() 191 glob_stats->min_latency = latency[i]; in calc_latency() 192 else if (latency[i] < glob_stats->min_latency) in calc_latency() 193 glob_stats->min_latency = latency[i]; in calc_latency() 194 else if (latency[i] > glob_stats->max_latency) in calc_latency() 195 glob_stats->max_latency = latency[i]; in calc_latency() 202 alpha * (latency[i] - glob_stats->avg_latency); in calc_latency() 203 prev_latency = latency[i]; in calc_latency()
|
| /dpdk/app/test-eventdev/ |
| H A D | test_perf_common.h | 35 uint64_t latency; member 136 uint64_t latency; in perf_process_last_stage_latency() local 150 latency = rte_get_timer_cycles() - m->timestamp; in perf_process_last_stage_latency() 153 latency = rte_get_timer_cycles() - m->timestamp; in perf_process_last_stage_latency() 156 w->latency += latency; in perf_process_last_stage_latency()
|
| H A D | test_perf_common.c | 444 total += t->worker[i].latency; in total_latency() 506 const uint64_t latency = total_latency(t); in perf_launch_lcores() local 519 (float)(latency/pkts)/freq_mhz); in perf_launch_lcores() 750 w->latency = 0; in perf_event_dev_port_setup()
|
| /dpdk/doc/guides/prog_guide/ |
| H A D | metrics_lib.rst | 239 The latency statistics library calculates the latency of packet 245 - ``min_latency_ns``: Minimum processing latency (nano-seconds) 246 - ``avg_latency_ns``: Average processing latency (nano-seconds) 247 - ``mac_latency_ns``: Maximum processing latency (nano-seconds) 248 - ``jitter_ns``: Variance in processing latency (nano-seconds) 265 rte_exit(EXIT_FAILURE, "Could not allocate latency data.\n"); 272 periodically so that latency statistics can be updated. 283 de-initialise the latency library. 289 Timestamp and latency calculation 296 timestamp and are used for the latency calculation.
|
| H A D | writing_efficient_code.rst | 136 Traditionally, there is a trade-off between throughput and latency. 138 but the end-to-end latency of an average packet will typically increase as a result. 140 a low end-to-end latency, at the cost of lower throughput. 156 However, this is not very desirable when tuning for low latency because 162 To consistently achieve low latency, even under heavy system load, 165 This will allow a single packet to be processed at a time, providing lower latency,
|
| H A D | traffic_management.rst | 135 based on latency measured and desired and whether the queuing latency is currently 136 trending up or down. Queuing latency can be obtained using direct measurement or
|
| H A D | profile_app.rst | 29 * Long latency instructions and exceptions
|
| /dpdk/doc/guides/faq/ |
| H A D | faq.rst | 75 How can I tune my network application to achieve lower latency? 78 Traditionally, there is a trade-off between throughput and latency. An application can be tuned to … 79 but the end-to-end latency of an average packet typically increases as a result. 80 Similarly, the application can be tuned to have, on average, a low end-to-end latency at the cost o… 90 However, this is not very desirable when tuning for low latency, because the first packet that was … 94 To consistently achieve low latency even under heavy system load, the application developer should … 96 This allows a single packet to be processed at a time, providing lower latency, but with the added … 119 …th memory domains, thus incurring QPI bandwidth accessing the other memory and much higher latency.
|
| /dpdk/drivers/bus/vmbus/ |
| H A D | vmbus_channel.c | 55 uint32_t latency) in rte_vmbus_set_latency() argument 60 if (latency >= UINT16_MAX * 100) { in rte_vmbus_set_latency() 61 VMBUS_LOG(ERR, "invalid latency value %u", latency); in rte_vmbus_set_latency() 72 dev->monitor_page->lat[trig_idx][trig_offs] = latency / 100; in rte_vmbus_set_latency()
|
| H A D | rte_bus_vmbus.h | 379 uint32_t latency);
|
| /dpdk/doc/guides/tools/ |
| H A D | cryptoperf.rst | 9 crypto tree. There are available two measurement types: throughput and latency. 86 latency 412 Call application for performance latency test of two Aesni MB PMD executed 419 --ptest latency --total-ops 10 474 tests, and histogram and boxplot graphs for latency tests. 532 both throughput and latency ptests are supported. 569 To run crypto-perf-qat latency test suite only: 573 ./dpdk-graph-crypto-perf configs/crypto-perf-qat -t latency 575 To run both crypto-perf-aesni-mb throughput and latency test suites 579 ./dpdk-graph-crypto-perf configs/crypto-perf-aesni-mb -t throughput latency [all …]
|
| H A D | testbbdev.rst | 9 Tests available for execution are: latency, throughput, validation, 55 ``./test-bbdev.py -c latency throughput`` 56 Runs latency and throughput test suites 122 * Latency measurement [-c latency] 125 - This measurement represents the full latency of a bbdev operation
|
| H A D | testeventdev.rst | 105 Perform forward latency measurement. 355 #. Measure the latency to forward an event. 405 value to compute the average latency to a forward packet. 482 #. Measure the latency to forward an event.
|
| /dpdk/doc/guides/sample_app_ug/ |
| H A D | rxtx_callbacks.rst | 9 packets. The application performs a simple latency check, using callbacks, to 17 also display the average latency since the packet was timestamped in hardware, 18 on top of the latency since the packet was received and processed by the RX
|
| H A D | intro.rst | 76 packets. The application calculates the latency of a packet between RX
|
| /dpdk/doc/guides/nics/ |
| H A D | avp.rst | 16 throughput and decreased latency to meet the demands of their performance 28 significantly improved throughput and latency over other device types.
|
| H A D | sfc_efx.rst | 451 - ``perf_profile`` [auto|throughput|low-latency] (default **throughput**) 454 low-latency. 467 - ``fw_variant`` [dont-care|full-feature|ultra-low-latency| 479 **ultra-low-latency** chooses firmware with fewer features but lower latency.
|
| H A D | netvsc.rst | 110 #. ``latency``: 115 is used by the host. Smaller values improve transmit latency, and larger
|
| H A D | ice.rst | 237 - ``Low Rx latency`` (default ``0``) 239 vRAN workloads require low latency DPDK interface for the front haul 242 to host memory and the Rx interrupt latency can be reduced to 2us::
|
| H A D | ena.rst | 30 Some of the ENA devices support a working mode called Low-latency 164 reduces the latency of the packets by pushing the header directly through
|
| /dpdk/doc/guides/eventdevs/ |
| H A D | opdl.rst | 9 processing workloads that have high throughput and low latency requirements.\ 87 that the implementation can achieve such high throughput and low latency
|
| /dpdk/drivers/net/netvsc/ |
| H A D | hn_ethdev.c | 167 hv->latency = v * 1000; in hn_set_parameter() 168 PMD_DRV_LOG(DEBUG, "set latency %u usec", hv->latency); in hn_set_parameter() 535 rte_vmbus_set_latency(hv->vmbus, new_sc, hv->latency); in hn_subchan_configure() 1175 hv->latency = HN_CHAN_LATENCY_NS; in eth_hn_dev_init() 1204 rte_vmbus_set_latency(hv->vmbus, hv->channels[0], hv->latency); in eth_hn_dev_init()
|
| H A D | hn_var.h | 166 uint32_t latency; member
|
| /dpdk/doc/guides/rel_notes/ |
| H A D | release_17_05.rst | 248 * **Added latency stats library.** 250 Added a library that measures packet latency. The collected statistics are 251 jitter and latency. For latency the minimum, average, and maximum is
|
| /dpdk/doc/guides/rawdevs/ |
| H A D | cnxk_gpio.rst | 29 for installing interrupt handlers for low latency signal processing.
|