Skip to content

Latest commit

 

History

History
920 lines (735 loc) · 42.6 KB

File metadata and controls

920 lines (735 loc) · 42.6 KB

Linux 12.00.00 Performance Guide

Read This First

All performance numbers provided in this document are gathered using following Evaluation Modules unless otherwise specified.

Name Description
AM62Lx EVM AM62Lx Evaluation Module rev E1-1 with ARM running at 1.4GHz, DDR data rate 1600 MT/S

Table: Evaluation Modules

About This Manual

This document provides performance data for each of the device drivers which are part of the Processor SDK Linux package. This document should be used in conjunction with release notes and user guides provided with the Processor SDK Linux package for information on specific issues present with drivers included in a particular release.

For further information or to report any problems, contact https://e2e.ti.com/ or https://support.ti.com/


System Benchmarks


LMBench

LMBench is a collection of microbenchmarks of which the memory bandwidth and latency related ones are typically used to estimate processor memory system performance. More information about lmbench at https://lmbench.sourceforge.net/whatis_lmbench.html and https://lmbench.sourceforge.net/man/lmbench.8.html

Latency: lat_mem_rd-stride128-szN, where N is equal to or smaller than the cache size at given level measures the cache miss penalty. N that is at least double the size of last level cache is the latency to external memory.

Bandwidth: bw_mem_bcopy-N, where N is equal to or smaller than the cache size at a given level measures the achievable memory bandwidth from software doing a memcpy() type operation. Typical use is for external memory bandwidth calculation. The bandwidth is calculated as byte read and written counts as 1 which should be roughly half of STREAM copy result.

Execute the LMBench with the following:

cd /opt/ltp
./runltp -P j721e-idk-gw -f ddt/lmbench -s LMBENCH_L_PERF_0001
LMBench Benchmarks
Benchmarks am62lxx_evm-fs: perf
af_unix_sock_stream_latency (microsec) 28.47 (min 25.57, max 29.86)
af_unix_socket_stream_bandwidth (mbs) 473.68 (min 441.13, max 551.95)
bw_file_rd-io-1mb (mb/s) 952.82 (min 909.92, max 1066.29)
bw_file_rd-o2c-1mb (mb/s) 391.04 (min 55.18, max 547.95)
bw_mem-bcopy-16mb (mb/s) 749.66 (min 705.87, max 858.60)
bw_mem-bcopy-1mb (mb/s) 769.84 (min 694.44, max 909.09)
bw_mem-bcopy-2mb (mb/s) 752.70 (min 690.49, max 913.94)
bw_mem-bcopy-4mb (mb/s) 775.04 (min 704.72, max 954.65)
bw_mem-bcopy-8mb (mb/s) 809.57 (min 727.14, max 971.23)
bw_mem-bzero-16mb (mb/s) 2279.05 (min 2106.37, max 2725.72)
bw_mem-bzero-1mb (mb/s) 1526.26 (min 694.44, max 2725.29)
bw_mem-bzero-2mb (mb/s) 1515.47 (min 690.49, max 2727.45)
bw_mem-bzero-4mb (mb/s) 1526.69 (min 704.72, max 2724.33)
bw_mem-bzero-8mb (mb/s) 1546.56 (min 727.14, max 2725.72)
bw_mem-cp-16mb (mb/s) 426.86 (min 395.35, max 500.74)
bw_mem-cp-1mb (mb/s) 1439.36 (min 395.41, max 2878.53)
bw_mem-cp-2mb (mb/s) 1400.78 (min 386.25, max 2819.38)
bw_mem-cp-4mb (mb/s) 1398.29 (min 434.36, max 2764.34)
bw_mem-cp-8mb (mb/s) 1393.47 (min 432.85, max 2750.09)
bw_mem-fcp-16mb (mb/s) 751.74 (min 710.38, max 853.20)
bw_mem-fcp-1mb (mb/s) 1557.43 (min 776.04, max 2725.29)
bw_mem-fcp-2mb (mb/s) 1551.17 (min 764.33, max 2727.45)
bw_mem-fcp-4mb (mb/s) 1563.30 (min 790.28, max 2724.33)
bw_mem-fcp-8mb (mb/s) 1574.04 (min 798.08, max 2725.72)
bw_mem-frd-16mb (mb/s) 1329.96 (min 1253.62, max 1537.28)
bw_mem-frd-1mb (mb/s) 1049.23 (min 776.04, max 1494.21)
bw_mem-frd-2mb (mb/s) 1043.79 (min 764.33, max 1423.99)
bw_mem-frd-4mb (mb/s) 1062.05 (min 790.28, max 1526.14)
bw_mem-frd-8mb (mb/s) 1092.66 (min 798.08, max 1551.89)
bw_mem-fwr-16mb (mb/s) 2295.38 (min 2122.02, max 2738.79)
bw_mem-fwr-1mb (mb/s) 1857.24 (min 1178.25, max 2878.53)
bw_mem-fwr-2mb (mb/s) 1819.10 (min 1196.53, max 2819.38)
bw_mem-fwr-4mb (mb/s) 1803.17 (min 1166.35, max 2764.34)
bw_mem-fwr-8mb (mb/s) 1816.38 (min 1237.43, max 2750.09)
bw_mem-rd-16mb (mb/s) 1348.11 (min 1269.04, max 1574.80)
bw_mem-rd-1mb (mb/s) 973.75 (min 532.67, max 1533.45)
bw_mem-rd-2mb (mb/s) 954.80 (min 508.78, max 1548.95)
bw_mem-rd-4mb (mb/s) 965.06 (min 582.16, max 1563.11)
bw_mem-rd-8mb (mb/s) 1028.38 (min 671.03, max 1574.49)
bw_mem-rdwr-16mb (mb/s) 742.83 (min 671.11, max 871.98)
bw_mem-rdwr-1mb (mb/s) 507.25 (min 395.41, max 758.58)
bw_mem-rdwr-2mb (mb/s) 508.60 (min 386.25, max 742.12)
bw_mem-rdwr-4mb (mb/s) 554.93 (min 434.36, max 758.58)
bw_mem-rdwr-8mb (mb/s) 594.72 (min 432.85, max 831.51)
bw_mem-wr-16mb (mb/s) 761.79 (min 696.29, max 875.47)
bw_mem-wr-1mb (mb/s) 597.49 (min 503.36, max 758.58)
bw_mem-wr-2mb (mb/s) 580.37 (min 508.78, max 742.12)
bw_mem-wr-4mb (mb/s) 649.66 (min 582.16, max 809.88)
bw_mem-wr-8mb (mb/s) 712.22 (min 660.23, max 831.51)
bw_mmap_rd-mo-1mb (mb/s) 1186.31 (min 1095.29, max 1369.06)
bw_mmap_rd-o2c-1mb (mb/s) 504.63 (min 466.13, max 534.00)
bw_pipe (mb/s) 397.94 (min 332.79, max 502.07)
bw_unix (mb/s) 473.68 (min 441.13, max 551.95)
lat_connect (us) 110.85 (min 108.39, max 113.45)
lat_ctx-2-128k (us) 40.79 (min 10.02, max 53.07)
lat_ctx-2-256k (us) 122.00 (min 19.85, max 160.05)
lat_ctx-4-128k (us) 79.14 (min 11.48, max 104.99)
lat_ctx-4-256k (us) 95.97 (min 8.50, max 179.58)
lat_fs-0k (num_files) 203.25 (min 195.00, max 212.00)
lat_fs-10k (num_files) 89.00 (min 81.00, max 96.00)
lat_fs-1k (num_files) 134.25 (min 118.00, max 146.00)
lat_fs-4k (num_files) 128.25 (min 120.00, max 136.00)
lat_mem_rd-stride128-sz1000k (ns) 49.87 (min 42.84, max 52.35)
lat_mem_rd-stride128-sz125k (ns) 6.18 (min 6.17, max 6.18)
lat_mem_rd-stride128-sz250k (ns) 11.22 (min 7.65, max 14.89)
lat_mem_rd-stride128-sz31k (ns) 3.16 (min 2.41, max 3.90)
lat_mem_rd-stride128-sz50 (ns) 2.41
lat_mem_rd-stride128-sz500k (ns) 45.38 (min 39.37, max 47.67)
lat_mem_rd-stride128-sz62k (ns) 5.74 (min 5.72, max 5.77)
lat_mmap-1m (us) 65.00 (min 57.00, max 68.00)
lat_ops-double-add (ns) 3.21
lat_ops-double-div (ns) 17.66 (min 17.65, max 17.66)
lat_ops-double-mul (ns) 3.21
lat_ops-float-add (ns) 3.21
lat_ops-float-div (ns) 10.44 (min 10.43, max 10.45)
lat_ops-float-mul (ns) 3.21 (min 3.21, max 3.22)
lat_ops-int-add (ns) 0.80
lat_ops-int-bit (ns) 0.53
lat_ops-int-div (ns) 4.82 (min 4.81, max 4.82)
lat_ops-int-mod (ns) 5.08 (min 5.08, max 5.09)
lat_ops-int-mul (ns) 3.49
lat_ops-int64-add (ns) 0.80
lat_ops-int64-bit (ns) 0.54 (min 0.53, max 0.54)
lat_ops-int64-div (ns) 7.63 (min 7.62, max 7.63)
lat_ops-int64-mod (ns) 5.89 (min 5.88, max 5.92)
lat_ops-int64-mul (ns) 3.98 (min 3.97, max 3.99)
lat_pagefault (us) 0.87 (min 0.77, max 0.93)
lat_pipe (us) 22.18 (min 20.56, max 23.06)
lat_proc-exec (us) 1469.13 (min 1298.00, max 1543.25)
lat_proc-fork (us) 1296.60 (min 1160.40, max 1372.00)
lat_proc-proccall (us) 0.01
lat_select (us) 35.38 (min 35.29, max 35.56)
lat_sem (us) 3.55 (min 3.33, max 3.84)
lat_sig-catch (us) 6.24 (min 6.16, max 6.34)
lat_sig-install (us) 0.66 (min 0.65, max 0.70)
lat_sig-prot (us) 0.95 (min 0.88, max 1.01)
lat_syscall-fstat (us) 2.02 (min 1.97, max 2.12)
lat_syscall-null (us) 0.42 (min 0.41, max 0.42)
lat_syscall-open (us) 485.80 (min 413.85, max 568.44)
lat_syscall-read (us) 0.83 (min 0.82, max 0.83)
lat_syscall-stat (us) 4.74 (min 4.69, max 4.79)
lat_syscall-write (us) 0.75
lat_tcp (us) 0.86
lat_unix (us) 28.47 (min 25.57, max 29.86)
latency_for_0.50_mb_block_size (nanosec) 45.38 (min 39.37, max 47.67)
latency_for_1.00_mb_block_size (nanosec) 22.16 (min 0.00, max 52.35)
pipe_bandwidth (mbs) 397.94 (min 332.79, max 502.07)
pipe_latency (microsec) 22.18 (min 20.56, max 23.06)
procedure_call (microsec) 0.01
select_on_200_tcp_fds (microsec) 35.38 (min 35.29, max 35.56)
semaphore_latency (microsec) 3.55 (min 3.33, max 3.84)
signal_handler_latency (microsec) 0.66 (min 0.65, max 0.70)
signal_handler_overhead (microsec) 6.24 (min 6.16, max 6.34)
tcp_ip_connection_cost_to_localhost (microsec) 110.85 (min 108.39, max 113.45)
tcp_latency_using_localhost (microsec) 0.86

Dhrystone

Dhrystone is a core only benchmark that runs from warm L1 caches in all modern processors. It scales linearly with clock speed.

Please take note, different run may produce different slightly results. This is advised to run this test multiple times in order to get maximum performance numbers.

Execute the benchmark with the following:

runDhrystone
Dhrystone Benchmarks
Benchmarks am62lxx_evm-fs: perf
cpu_clock (mhz) 1250.00
dhrystone_per_mhz (dmips/mhz) 2.80
dhrystone_per_second (dhrystonep) 6250000.00

Whetstone

Whetstone is a benchmark primarily measuring floating-point arithmetic performance.

Execute the benchmark with the following:

runWhetstone
Whetstone Benchmarks
Benchmarks am62lxx_evm-fs: perf
whetstone (mips) 5000.00

Linpack

Linpack measures peak double precision (64 bit) floating point performance in solving a dense linear system.

Linpack Benchmarks
Benchmarks am62lxx_evm-fs: perf
linpack (kflops) 516120.50 (min 514627.00, max 517614.00)

NBench

NBench which stands for Native Benchmark is used to measure macro benchmarks for commonly used operations such as sorting and analysis algorithms. More information about NBench at https://en.wikipedia.org/wiki/NBench and https://nbench.io/articles/index.html

NBench Benchmarks
Benchmarks am62lxx_evm-fs: perf
assignment (iterations) 12.64 (min 12.62, max 12.67)
fourier (iterations) 18560.75 (min 18556.00, max 18568.00)
fp_emulation (iterations) 182.43 (min 182.35, max 182.50)
huffman (iterations) 1063.13 (min 1063.00, max 1063.20)
idea (iterations) 2738.40 (min 2738.10, max 2738.60)
lu_decomposition (iterations) 473.09 (min 467.17, max 475.75)
neural_net (iterations) 8.63 (min 8.59, max 8.64)
numeric_sort (iterations) 483.39 (min 482.54, max 484.16)
string_sort (iterations) 150.19 (min 150.17, max 150.21)

Stream

STREAM is a microbenchmark for measuring data memory system performance without any data reuse. It is designed to miss on caches and exercise data prefetcher and speculative accesses. It uses double precision floating point (64bit) but in most modern processors the memory access will be the bottleneck. The four individual scores are copy, scale as in multiply by constant, add two numbers, and triad for multiply accumulate. For bandwidth, a byte read counts as one and a byte written counts as one, resulting in a score that is double the bandwidth LMBench will show.

Execute the benchmark with the following:

stream_c
Stream Benchmarks
Benchmarks am62lxx_evm-fs: perf
add (mb/s) 1523.87 (min 1498.10, max 1538.30)
copy (mb/s) 1629.27 (min 1610.40, max 1639.00)
scale (mb/s) 1672.47 (min 1632.10, max 1713.80)
triad (mb/s) 1512.27 (min 1488.30, max 1526.00)

CoreMarkPro

CoreMark®-Pro is a comprehensive, advanced processor benchmark that works with and enhances the market-proven industry-standard EEMBC CoreMark® benchmark. While CoreMark stresses the CPU pipeline, CoreMark-Pro tests the entire processor, adding comprehensive support for multicore technology, a combination of integer and floating-point workloads, and data sets for utilizing larger memory subsystems.

CoreMarkPro Benchmarks
Benchmarks am62lxx_evm-fs: perf
cjpeg-rose7-preset (workloads/) 37.22 (min 37.04, max 37.31)
core (workloads/) 0.27
coremark-pro () 719.19 (min 697.16, max 733.93)
linear_alg-mid-100x100-sp (workloads/) 13.08
loops-all-mid-10k-sp (workloads/) 0.59 (min 0.56, max 0.60)
nnet_test (workloads/) 0.96
parser-125k (workloads/) 7.62 (min 7.04, max 8.06)
radix2-big-64k (workloads/) 19.26 (min 16.30, max 21.85)
sha-test (workloads/) 71.11 (min 69.44, max 72.46)
zip-test (workloads/) 19.04 (min 17.86, max 20.41)
CoreMarkProTwoCore Benchmarks
Benchmarks am62lxx_evm-fs: perf
cjpeg-rose7-preset (workloads/) 72.20 (min 71.94, max 72.46)
core (workloads/) 0.53 (min 0.53, max 0.54)
coremark-pro () 1209.22 (min 1186.40, max 1232.04)
linear_alg-mid-100x100-sp (workloads/) 26.04 (min 26.03, max 26.06)
loops-all-mid-10k-sp (workloads/) 0.99 (min 0.98, max 1.00)
nnet_test (workloads/) 1.91 (min 1.91, max 1.92)
parser-125k (workloads/) 7.61 (min 5.97, max 9.26)
radix2-big-64k (workloads/) 24.15 (min 23.08, max 25.21)
sha-test (workloads/) 137.01 (min 135.14, max 138.89)
zip-test (workloads/) 33.37 (min 32.26, max 34.48)

MultiBench

MultiBench™ is a suite of benchmarks that allows processor and system designers to analyze, test, and improve multicore processors. It uses three forms of concurrency: Data decomposition: multiple threads cooperating on achieving a unified goal and demonstrating a processor’s support for fine grain parallelism. Processing multiple data streams: uses common code running over multiple threads and demonstrating how well a processor scales over scalable data inputs. Multiple workload processing: shows the scalability of general-purpose processing, demonstrating concurrency over both code and data. MultiBench combines a wide variety of application-specific workloads with the EEMBC Multi-Instance-Test Harness (MITH), compatible and portable with most any multicore processors and operating systems. MITH uses a thread-based API (POSIX-compliant) to establish a common programming model that communicates with the benchmark through an abstraction layer and provides a flexible interface to allow a wide variety of thread-enabled workloads to be tested.

Multibench Benchmarks
Benchmarks am62lxx_evm-fs: perf
4m-check (workloads/) 287.19 (min 265.48, max 320.97)
4m-check-reassembly (workloads/) 66.74 (min 51.05, max 83.26)
4m-check-reassembly-tcp (workloads/) 41.39 (min 33.69, max 47.80)
4m-check-reassembly-tcp-cmykw2-rotatew2 (workloads/) 16.75 (min 15.29, max 18.62)
4m-check-reassembly-tcp-x264w2 (workloads/) 0.87 (min 0.78, max 0.95)
4m-cmykw2 (workloads/) 107.63 (min 106.10, max 110.01)
4m-cmykw2-rotatew2 (workloads/) 22.14 (min 18.79, max 26.33)
4m-reassembly (workloads/) 74.80 (min 47.60, max 98.62)
4m-rotatew2 (workloads/) 24.65 (min 19.73, max 28.48)
4m-tcp-mixed (workloads/) 102.76 (min 98.16, max 108.84)
4m-x264w2 (workloads/) 0.93 (min 0.92, max 0.96)
empty-wld (workloads/) 1.00
idct-4m (workloads/) 17.23 (min 16.47, max 17.62)
idct-4mw1 (workloads/) 17.23 (min 16.48, max 17.62)
ippktcheck-4m (workloads/) 288.80 (min 267.09, max 323.63)
ippktcheck-4mw1 (workloads/) 288.26 (min 265.73, max 323.71)
ipres-4m (workloads/) 67.60 (min 60.10, max 78.25)
ipres-4mw1 (workloads/) 67.64 (min 59.78, max 78.21)
md5-4m (workloads/) 19.49 (min 18.37, max 20.95)
md5-4mw1 (workloads/) 19.53 (min 18.49, max 21.01)
rgbcmyk-4m (workloads/) 57.07 (min 56.69, max 57.47)
rgbcmyk-4mw1 (workloads/) 57.04 (min 56.64, max 57.29)
rotate-4ms1 (workloads/) 20.24 (min 17.49, max 23.21)
rotate-4ms1w1 (workloads/) 20.27 (min 17.48, max 23.20)
rotate-4ms64 (workloads/) 20.37 (min 17.64, max 23.22)
rotate-4ms64w1 (workloads/) 20.36 (min 17.62, max 23.29)
x264-4mq (workloads/) 0.50 (min 0.50, max 0.51)
x264-4mqw1 (workloads/) 0.50 (min 0.49, max 0.51)

Boot-time Measurement

Boot media: MMCSD

Linux boot time MMCSD
Boot Configuration am62lxx_evm-fs: Boot time in seconds: avg(min,max)
Linux boot time from SD with default rootfs (20 boot cycles) 21.87 (min 19.45, max 24.14)

Boot time numbers [avg, min, max] are measured from "Starting kernel" to Linux prompt across 20 boot cycles.


ALSA SoC Audio Driver

  1. Access type - RW_INTERLEAVED
  2. Channels - 2
  3. Format - S16_LE
  4. Period size - 64
Audio Capture
Sampling Rate (Hz) am62lxx_evm-fs: Throughput (bits/sec) am62lxx_evm-fs: CPU Load (%)
11025 352795.00 (min 352794.00, max 352796.00) 0.36 (min 0.29, max 0.42)
16000 511994.00 0.37 (min 0.31, max 0.42)
22050 705588.00 (min 705587.00, max 705589.00) 0.36 (min 0.35, max 0.36)
24000 705590.00 (min 705589.00, max 705591.00) 0.38 (min 0.36, max 0.40)
32000 1023985.50 (min 1023984.00, max 1023987.00) 0.22 (min 0.21, max 0.23)
44100 1411180.00 (min 1411179.00, max 1411181.00) 0.54 (min 0.53, max 0.54)
48000 1535978.50 (min 1535977.00, max 1535980.00) 0.59 (min 0.19, max 0.99)
88200 2822351.50 (min 2822349.00, max 2822354.00) 0.83 (min 0.81, max 0.84)
96000 3071928.50 (min 3071926.00, max 3071931.00) 1.18 (min 0.39, max 1.96)
Audio Playback
Sampling Rate (Hz) am62lxx_evm-fs: Throughput (bits/sec) am62lxx_evm-fs: CPU Load (%)
11025 352943.67 (min 352943.00, max 352944.00) 0.45 (min 0.31, max 0.63)
16000 512209.00 (min 512208.00, max 512210.00) 0.33 (min 0.22, max 0.46)
22050 705886.67 (min 705885.00, max 705890.00) 0.36 (min 0.30, max 0.48)
24000 705888.00 (min 705887.00, max 705890.00) 0.42 (min 0.36, max 0.53)
32000 1024417.67 (min 1024416.00, max 1024420.00) 0.54 (min 0.24, max 1.00)
44100 1411774.67 (min 1411772.00, max 1411778.00) 0.52 (min 0.42, max 0.67)
48000 1536624.67 (min 1536622.00, max 1536628.00) 0.48 (min 0.33, max 0.77)
88200 2823538.00 (min 2823532.00, max 2823546.00) 0.74 (min 0.62, max 0.91)
96000 3073210.00 (min 3073190.00, max 3073224.00) 0.48 (min 0.37, max 0.55)

Ethernet

Ethernet performance benchmarks were measured using :command:`netperf` 2.7.1 https://hewlettpackard.github.io/netperf/doc/netperf.html Test procedures were modeled after those defined in RFC-2544: https://tools.ietf.org/html/rfc2544, where the DUT is the TI device and the "tester" used was a Linux PC. To produce consistent results, it is recommended to carry out performance tests in a private network and to avoid running NFS on the same interface used in the test. In these results, CPU utilization was captured as the total percentage used across all cores on the device, while running the performance test over one external interface.

UDP Throughput (0% loss) was measured by the procedure defined in RFC-2544 section 26.1: Throughput. In this scenario, :command:`netperf` options burst_size (-b) and wait_time (-w) are used to limit bandwidth during different trials of the test, with the goal of finding the highest rate at which no loss is seen. For example, to limit bandwidth to 500Mbits/sec with 1472B datagram:

burst_size = <bandwidth (bits/sec)> / 8 (bits -> bytes) / <UDP datagram size> / 100 (seconds -> 10 ms)
burst_size = 500000000 / 8 / 1472 / 100 = 425

wait_time = 10 milliseconds (minimum supported by Linux PC used for testing)

UDP Throughput (possible loss) was measured by capturing throughput and packet loss statistics when running the :command:`netperf` test with no bandwidth limit (remove -b/-w options).

In order to start a :command:`netperf` client on one device, the other device must have :command:`netserver` running. To start :command:`netserver`:

netserver [-p <port_number>] [-4 (IPv4 addressing)] [-6 (IPv6 addressing)]

Running the following shell script from the DUT will trigger :command:`netperf` clients to measure bidirectional TCP performance for 60 seconds and report CPU utilization. Parameter -k is used in client commands to summarize selected statistics on their own line and -j is used to gain additional timing measurements during the test.

#!/bin/bash
for i in 1
do
   netperf -H <tester ip> -j -c -l 60 -t TCP_STREAM --
      -k DIRECTION,THROUGHPUT,MEAN_LATENCY,LOCAL_CPU_UTIL,REMOTE_CPU_UTIL,LOCAL_BYTES_SENT,REMOTE_BYTES_RECVD,LOCAL_SEND_SIZE &

   netperf -H <tester ip> -j -c -l 60 -t TCP_MAERTS --
      -k DIRECTION,THROUGHPUT,MEAN_LATENCY,LOCAL_CPU_UTIL,REMOTE_CPU_UTIL,LOCAL_BYTES_SENT,REMOTE_BYTES_RECVD,LOCAL_SEND_SIZE &
done

Running the following commands will trigger :command:`netperf` clients to measure UDP burst performance for 60 seconds at various burst/datagram sizes and report CPU utilization.

netperf -H <tester ip> -j -c -l 60 -t UDP_STREAM -b <burst_size> -w <wait_time> -- -m <UDP datagram size>
   -k DIRECTION,THROUGHPUT,MEAN_LATENCY,LOCAL_CPU_UTIL,REMOTE_CPU_UTIL,LOCAL_BYTES_SENT,REMOTE_BYTES_RECVD,LOCAL_SEND_SIZE
netperf -H <DUT ip> -j -C -l 60 -t UDP_STREAM -b <burst_size> -w <wait_time> -- -m <UDP datagram size>
   -k DIRECTION,THROUGHPUT,MEAN_LATENCY,LOCAL_CPU_UTIL,REMOTE_CPU_UTIL,LOCAL_BYTES_SENT,REMOTE_BYTES_RECVD,LOCAL_SEND_SIZE

CPSW/CPSW2g/CPSW3g Ethernet

TCP Bidirectional Throughput
CPSW2g TCP Bidirectional Throughput
Command Used am62lxx_evm-fs: THROUGHPUT (Mbits/sec) am62lxx_evm-fs: CPU Load % (LOCAL_CPU_UTIL)
netperf -H 192.168.0.1 -j -c -C -l 60 -t TCP_STREAM; netperf -H 192.168.0.1 -j -c -C -l 60 -t TCP_MAERTS 1048.02 (min 1035.65, max 1060.38) 99.10 (min 98.47, max 99.72)
TCP Bidirectional Throughput Interrupt Pacing
CPSW2g TCP Bidirectional Throughput Interrupt Pacing
Command Used am62lxx_evm-fs: THROUGHPUT (Mbits/sec) am62lxx_evm-fs: CPU Load % (LOCAL_CPU_UTIL)
netperf -H 192.168.0.1 -j -c -C -l 60 -t TCP_STREAM; netperf -H 192.168.0.1 -j -c -C -l 60 -t TCP_MAERTS 1182.95 (min 1179.68, max 1186.21) 95.50 (min 95.39, max 95.60)
UDP Throughput
CPSW2g UDP Egress Throughput 0 loss
Frame Size(bytes) am62lxx_evm-fs: UDP Datagram Size(bytes) (LOCAL_SEND_SIZE) am62lxx_evm-fs: THROUGHPUT (Mbits/sec) am62lxx_evm-fs: Packets Per Second (kPPS) am62lxx_evm-fs: CPU Load % (LOCAL_CPU_UTIL)
64   36.89 (min 36.82, max 36.96) 72.00 75.48 (min 75.29, max 75.67)
128   80.20 78.00 77.85
256   145.24 71.00 75.58
1024   355.15 (min 137.62, max 572.68) 43.50 (min 17.00, max 70.00) 49.89 (min 24.19, max 75.58)
1518   623.50 (min 619.61, max 627.38) 51.50 (min 51.00, max 52.00) 72.61 (min 71.63, max 73.58)
CPSW2g UDP Ingress Throughput 0 loss
Frame Size(bytes) am62lxx_evm-fs: UDP Datagram Size(bytes) (LOCAL_SEND_SIZE) am62lxx_evm-fs: THROUGHPUT (Mbits/sec) am62lxx_evm-fs: Packets Per Second (kPPS) am62lxx_evm-fs: CPU Load % (LOCAL_CPU_UTIL)
64   1.54 (min 1.48, max 1.59) 3.00 1.97 (min 1.84, max 2.10)
128   4.46 (min 4.40, max 4.51) 4.00 3.70 (min 2.67, max 4.73)
256   10.14 (min 10.04, max 10.24) 5.00 3.06 (min 3.00, max 3.11)
1024   42.19 (min 41.78, max 42.60) 5.00 3.49 (min 3.40, max 3.57)
1518   34.74 (min 8.24, max 61.23) 3.00 (min 1.00, max 5.00) 3.65 (min 1.02, max 6.27)
CPSW2g UDP Ingress Throughput possible loss
Frame Size(bytes) am62lxx_evm-fs: UDP Datagram Size(bytes) (LOCAL_SEND_SIZE) am62lxx_evm-fs: THROUGHPUT (Mbits/sec) am62lxx_evm-fs: Packets Per Second (kPPS) am62lxx_evm-fs: CPU Load % (LOCAL_CPU_UTIL) am62lxx_evm-fs: Packet Loss %
64   47.26 (min 46.91, max 47.61) 92.50 (min 92.00, max 93.00) 65.65 (min 64.95, max 66.34) 79.50 (min 79.03, max 79.97)
128   89.46 (min 88.85, max 90.07) 87.50 (min 87.00, max 88.00) 67.22 (min 67.01, max 67.42) 83.49 (min 80.73, max 86.25)
256   187.89 (min 178.98, max 196.80) 91.50 (min 87.00, max 96.00) 67.38 (min 65.87, max 68.89) 55.23 (min 51.03, max 59.42)
1024   590.71 (min 544.95, max 636.47) 72.50 (min 67.00, max 78.00) 74.71 (min 70.00, max 79.41) 21.13 (min 16.49, max 25.77)
1518   784.17 (min 775.84, max 792.50) 66.50 (min 66.00, max 67.00) 76.55 (min 75.73, max 77.37) 14.76 (min 12.32, max 17.20)

EMMC

Warning

IMPORTANT: The performance numbers can be severely affected if the media is mounted in sync mode. Hot plug scripts in the filesystem mount removable media in sync mode to ensure data integrity. For performance sensitive applications, umount the auto-mounted filesystem and re-mount in async mode.

EMMC EXT4 FIO 1G

EMMC EXT4 FIO 1G
Buffer size (bytes) am62lxx_evm-fs: Write EXT4 Throughput (Mbytes/sec) am62lxx_evm-fs: Write EXT4 CPU Load (%) am62lxx_evm-fs: Read EXT4 Throughput (Mbytes/sec) am62lxx_evm-fs: Read EXT4 CPU Load (%)
1m 128.33 (min 128.00, max 129.00) 8.92 (min 8.60, max 9.20) 179.33 (min 179.00, max 180.00) 7.83 (min 7.36, max 8.35)
4m 128.67 (min 128.00, max 129.00) 5.52 (min 5.34, max 5.80) 179.33 (min 179.00, max 180.00) 5.08 (min 4.98, max 5.22)
4k 87.67 (min 86.30, max 88.60) 57.12 (min 56.47, max 57.84) 93.83 (min 90.80, max 97.50) 55.19 (min 54.00, max 57.17)
256k 123.00 12.30 (min 11.83, max 13.08) 177.67 (min 175.00, max 179.00) 10.40 (min 9.43, max 11.55)

EMMC EXT4

EMMC EXT4
Buffer size (bytes) am62lxx_evm-fs: Write EXT4 Throughput (Mbytes/sec) am62lxx_evm-fs: Write EXT4 CPU Load (%) am62lxx_evm-fs: Read EXT4 Throughput (Mbytes/sec) am62lxx_evm-fs: Read EXT4 CPU Load (%)
102400 104.84 (min 95.48, max 108.41) 15.02 (min 13.33, max 20.00) 175.11 (min 158.41, max 183.74) 29.09 (min 25.19, max 33.64)
262144 106.77 (min 100.69, max 110.53) 12.13 (min 10.05, max 16.93) 185.91 (min 185.64, max 186.39) 27.61 (min 25.00, max 31.48)
524288 106.64 (min 99.47, max 110.24) 12.17 (min 9.74, max 15.57) 186.99 (min 186.68, max 187.16) 23.49 (min 20.91, max 25.00)
1048576 106.92 (min 102.07, max 109.98) 11.92 (min 10.26, max 15.71) 187.10 (min 187.05, max 187.17) 23.32 (min 22.22, max 24.11)
5242880 107.71 (min 100.42, max 111.86) 12.14 (min 10.47, max 15.05) 187.16 (min 187.07, max 187.23) 23.25 (min 22.73, max 24.07)

EMMC VFAT

EMMC VFAT
Buffer size (bytes) am62lxx_evm-fs: Write VFAT Throughput (Mbytes/sec) am62lxx_evm-fs: Write VFAT CPU Load (%) am62lxx_evm-fs: Read VFAT Throughput (Mbytes/sec) am62lxx_evm-fs: Read VFAT CPU Load (%)
102400 32.78 (min 16.04, max 37.19) 15.45 (min 12.96, max 20.27) 174.45 (min 173.98, max 175.01) 39.26 (min 36.52, max 43.97)
262144 47.10 (min 19.77, max 59.67) 18.04 (min 14.56, max 21.99) 175.94 (min 175.78, max 176.17) 36.81 (min 34.51, max 38.94)
524288 55.87 (min 21.16, max 65.96) 18.69 (min 15.92, max 25.61) 177.09 (min 176.63, max 177.56) 32.66 (min 31.03, max 34.19)
1048576 61.77 (min 22.07, max 76.62) 19.90 (min 18.12, max 22.93) 175.57 (min 174.58, max 177.21) 35.72 (min 33.04, max 40.52)
5242880 71.75 (min 23.10, max 85.16) 21.11 (min 19.44, max 22.40) 176.47 (min 176.20, max 176.65) 33.14 (min 32.48, max 34.45)

UBoot EMMC

UBOOT EMMC RAW
File size (bytes in hex) am62lxx_evm-fs: Write Throughput (Kbytes/sec) am62lxx_evm-fs: Read Throughput (Kbytes/sec)
2000000 111077.97 175229.95
4000000 124830.48 178086.96

MMCSD

Warning

IMPORTANT: The performance numbers can be severely affected if the media is mounted in sync mode. Hot plug scripts in the filesystem mount removable media in sync mode to ensure data integrity. For performance sensitive applications, umount the auto-mounted filesystem and re-mount in async mode.

MMC EXT4 FIO 1G

MMC EXT4 FIO 1G
Buffer size (bytes) am62lxx_evm-fs: Write EXT4 Throughput (Mbytes/sec) am62lxx_evm-fs: Write EXT4 CPU Load (%) am62lxx_evm-fs: Read EXT4 Throughput (Mbytes/sec) am62lxx_evm-fs: Read EXT4 CPU Load (%)
1m 42.60 (min 42.20, max 43.00) 4.13 (min 3.97, max 4.29) 86.40 (min 85.50, max 87.30) 5.47 (min 5.44, max 5.50)
4m 41.50 (min 40.30, max 42.70) 2.71 (min 2.63, max 2.78) 86.10 (min 84.70, max 87.50) 3.13 (min 3.12, max 3.14)
4k 2.86 (min 2.84, max 2.89) 7.03 (min 6.58, max 7.47) 12.95 (min 12.90, max 13.00) 12.01 (min 11.69, max 12.32)
256k 39.40 (min 38.90, max 39.90) 5.83 (min 5.38, max 6.28) 83.85 (min 83.70, max 84.00) 6.30 (min 6.26, max 6.34)

MMC EXT4

MMC EXT4
Buffer size (bytes) am62lxx_evm-fs: Write Raw Throughput (Mbytes/sec) am62lxx_evm-fs: Write Raw CPU Load (%) am62lxx_evm-fs: Read Raw Throughput (Mbytes/sec) am62lxx_evm-fs: Read Raw CPU Load (%)
102400 10.74 (min 10.57, max 10.92) 1.53 (min 1.20, max 2.08) 10.92 (min 10.64, max 11.08) 2.10 (min 1.85, max 2.37)
262144 10.70 (min 10.25, max 11.01) 1.52 (min 1.11, max 2.42) 11.12 (min 10.99, max 11.19) 1.63 (min 1.34, max 1.83)
524288 10.74 (min 10.42, max 10.96) 1.33 (min 1.04, max 1.80) 11.36 (min 11.06, max 11.52) 1.34 (min 1.32, max 1.38)
1048576 10.83 (min 10.22, max 11.21) 1.42 (min 0.94, max 2.00) 11.50 (min 11.02, max 12.02) 1.36 (min 1.16, max 1.55)
5242880 10.93 (min 10.69, max 11.19) 1.29 (min 0.92, max 2.16) 12.01 (min 11.99, max 12.02) 1.49 (min 1.32, max 1.61)

The performance numbers were captured using the following:

  • SanDisk Max Endurance SD card (SDSQQVR-032G-GN6IA)
  • Partition was mounted with async option

UBoot MMCSD

UBOOT MMCSD FAT
UBOOT MMCSD FAT
File size (bytes in hex) am62lxx_evm-fs: Write Throughput (Kbytes/sec) am62lxx_evm-fs: Read Throughput (Kbytes/sec)
400000 28068.99 (min 10422.39, max 39009.52) 63923.25 (min 11636.36, max 83591.84)
800000 35679.64 (min 10908.12, max 49053.89) 67640.33 (min 11719.60, max 88086.02)
1000000 36338.73 (min 11033.00, max 47489.86) 69037.04 (min 11744.80, max 90021.98)

The performance numbers were captured using the following:

  • SanDisk Max Endurance SD card (SDSQQVR-032G-GN6IA)

USB

USB Device Controller

USBDEVICE HIGHSPEED SLAVE_READ_THROUGHPUT
Number of Blocks am62lxx_evm-fs: Throughput (MB/sec)
150 37.55 (min 36.00, max 40.60)
USBDEVICE HIGHSPEED SLAVE_WRITE_THROUGHPUT
Number of Blocks am62lxx_evm-fs: Throughput (MB/sec)
150 31.63 (min 31.20, max 32.20)

CRYPTO

OpenSSL Performance

OpenSSL Performance
Algorithm Buffer Size (in bytes) am62lxx_evm-fs: throughput (KBytes/Sec)
aes-128-cbc 1024 820153.17 (min 819911.34, max 820533.93)
aes-128-cbc 16 88945.15 (min 88836.08, max 89002.73)
aes-128-cbc 16384 939063.98 (min 937519.79, max 940032.00)
aes-128-cbc 256 581491.09 (min 581083.82, max 581764.44)
aes-128-cbc 64 277695.28 (min 277243.09, max 278112.00)
aes-128-cbc 8192 930250.75 (min 929565.35, max 930660.35)
aes-128-ecb 1024 899457.71 (min 896871.42, max 901688.66)
aes-128-ecb 16 94404.48 (min 92735.88, max 95448.03)
aes-128-ecb 16384 1068426.58 (min 1067881.81, max 1068979.54)
aes-128-ecb 256 594218.41 (min 589874.01, max 597775.02)
aes-128-ecb 64 259665.47 (min 256530.28, max 261640.53)
aes-128-ecb 8192 1056195.93 (min 1055719.42, max 1057030.14)
aes-192-cbc 1024 635496.19 (min 634212.01, max 636192.09)
aes-192-cbc 16 84958.61 (min 83445.07, max 85605.02)
aes-192-cbc 16384 704898.39 (min 702851.75, max 706418.01)
aes-192-cbc 256 480254.10 (min 477063.59, max 481556.39)
aes-192-cbc 64 250222.52 (min 246972.93, max 251633.17)
aes-192-cbc 8192 701138.94 (min 700844.71, max 701322.58)
aes-192-ecb 1024 796659.71 (min 794786.82, max 797450.58)
aes-192-ecb 16 90855.18 (min 89027.61, max 91505.64)
aes-192-ecb 16384 929252.69 (min 928235.52, max 929824.77)
aes-192-ecb 256 544550.23 (min 540975.10, max 546148.27)
aes-192-ecb 64 245857.87 (min 242240.04, max 247096.23)
aes-192-ecb 8192 919371.09 (min 918964.91, max 919639.38)
aes-256-cbc 1024 533614.59 (min 532546.56, max 534222.85)
aes-256-cbc 16 82838.72 (min 81336.68, max 83377.88)
aes-256-cbc 16384 582748.84 (min 582489.43, max 582909.95)
aes-256-cbc 256 419748.05 (min 417439.23, max 420707.93)
aes-256-cbc 64 232853.02 (min 229933.29, max 233899.99)
aes-256-cbc 8192 579334.14 (min 579215.36, max 579499.35)
aes-256-ecb 1024 709590.02 (min 708103.51, max 710752.26)
aes-256-ecb 16 87171.66 (min 86051.96, max 88241.35)
aes-256-ecb 16384 813151.57 (min 812346.03, max 814159.19)
aes-256-ecb 256 488434.03 (min 485834.50, max 490525.18)
aes-256-ecb 64 232332.60 (min 230068.37, max 234072.00)
aes-256-ecb 8192 801607.00 (min 801270.44, max 802234.37)
des3 1024 4550.06 (min 4524.03, max 4562.26)
des3 16 4316.02 (min 4306.80, max 4319.40)
des3 16384 4564.31 (min 4560.21, max 4565.67)
des3 256 4548.33 (min 4545.19, max 4549.89)
des3 64 4497.16 (min 4490.35, max 4499.69)
des3 8192 4557.48 (min 4546.56, max 4565.67)
md5 1024 144407.38 (min 144327.00, max 144474.79)
md5 16 10784.95 (min 10761.82, max 10811.65)
md5 16384 179810.30 (min 179754.33, max 179869.01)
md5 256 88550.12 (min 88464.73, max 88638.55)
md5 64 34590.38 (min 34213.35, max 34737.00)
md5 8192 175732.74 (min 171794.43, max 177081.00)
sha1 1024 332384.51 (min 329384.28, max 333542.74)
sha1 16 12155.34 (min 12111.18, max 12188.26)
sha1 16384 557408.26 (min 555111.77, max 558410.41)
sha1 256 146047.32 (min 145553.07, max 146227.97)
sha1 64 44891.43 (min 44686.68, max 44967.42)
sha1 8192 534735.53 (min 533981.87, max 535374.51)
sha224 1024 319523.93 (min 317158.06, max 320510.98)
sha224 16 11875.43 (min 11692.84, max 11952.49)
sha224 16384 528467.29 (min 527777.79, max 529017.51)
sha224 256 141599.27 (min 139762.60, max 142371.58)
sha224 64 43810.79 (min 43110.55, max 44092.07)
sha224 8192 506508.63 (min 506161.83, max 507008.34)
sha256 1024 316993.96 (min 315484.84, max 318002.86)
sha256 16 11887.42 (min 11801.78, max 11934.95)
sha256 16384 528214.70 (min 527750.49, max 528438.61)
sha256 256 139981.29 (min 139145.39, max 140377.69)
sha256 64 43094.56 (min 42625.62, max 43298.41)
sha256 8192 505567.23 (min 504321.37, max 506585.09)
sha512 1024 99841.62 (min 99679.91, max 99963.90)
sha512 16 6977.87 (min 6945.17, max 7007.74)
sha512 16384 126693.38 (min 126621.01, max 126795.78)
sha512 256 59556.44 (min 59256.32, max 59719.51)
sha512 64 27937.12 (min 27885.93, max 28021.01)
sha512 8192 124555.95 (min 124523.86, max 124630.36)
OpenSSL CPU Load
Algorithm am62lxx_evm-fs: CPU Load
aes-128-cbc 98.75 (min 98.00, max 99.00)
aes-128-ecb 99.00
aes-192-cbc 99.00
aes-192-ecb 99.00
aes-256-cbc 99.00
aes-256-ecb 99.00
des3 99.00
md5 99.00
sha1 99.00
sha224 99.00
sha256 99.00
sha386 81.00 (min 78.00, max 85.00)
sha512 96.75 (min 90.00, max 99.00)

Listed for each algorithm are the code snippets used to run each benchmark test.

time -v openssl speed -elapsed -evp aes-128-cbc

Crypto Performance Comparison

The following table shows different AES/SHA algorithms throughput measured using openssl speed across DTHEv2 accelerator, ARM Cryptographic Extension (CE), and baseline ARM CPU.

Crypto Accelerator Performance
Algorithm Size (bytes) Accelerator (MB/s) ARM CE (MB/s) ARM (MB/s)
aes-128-cbc 16 0.28 86.86 28.02
aes-128-cbc 64 1.16 271.43 35.57
aes-128-cbc 256 4.63 568.25 38.33
aes-128-cbc 1024 17.15 800.83 39.08
aes-128-cbc 8192 82.06 908.72 39.31
aes-128-cbc 16384 112.48 917.23 39.02
aes-128-ecb 16 bytes 0.25 93.16 29.65
aes-128-ecb 64 bytes 1.04 255.43 36.71
aes-128-ecb 256 bytes 4.27 584.00 39.18
aes-128-ecb 1024 bytes 15.85 880.24 39.87
aes-128-ecb 8192 bytes 78.94 1016.25 40.06
aes-128-ecb 16384 bytes 109.80 1041.75 40.05
aes-192-cbc 16 bytes 0.19 83.46 25.01
aes-192-cbc 64 bytes 1.05 245.60 30.80
aes-192-cbc 256 bytes 4.22 470.37 32.16
aes-192-cbc 1024 bytes 15.82 621.54 33.41
aes-192-cbc 8192 bytes 56.12 685.22 33.57
aes-192-cbc 16384 bytes 94.11 688.92 33.57
aes-192-ecb 16 bytes 0.19 88.85 26.24
aes-192-ecb 64 bytes 0.95 241.29 31.64
aes-192-ecb 256 bytes 4.28 533.39 33.46
aes-192-ecb 1024 bytes 16.08 778.58 33.21
aes-192-ecb 8192 bytes 74.64 897.90 34.09
aes-192-ecb 16384 bytes 100.83 906.14 34.08
aes-256-cbc 16 bytes 0.19 81.34 22.37
aes-256-cbc 64 bytes 1.05 228.13 27.06
aes-256-cbc 256 bytes 4.35 410.47 28.68
aes-256-cbc 1024 bytes 15.88 521.18 29.06
aes-256-cbc 8192 bytes 70.04 565.39 29.25
aes-256-cbc 16384 bytes 90.28 567.93 29.25
sha2-256 16 bytes 0.20 11.66 7.69
sha2-256 64 bytes 0.60 42.28 21.87
sha2-256 256 bytes 1.43 137.06 48.05
sha2-256 1024 bytes 6.81 310.52 68.74
sha2-256 8192 bytes 57.47 494.56 78.58
sha2-256 16384 bytes 99.41 514.70 79.38
sha2-512 16 bytes 0.13 6.80 6.89
sha2-512 64 bytes 0.67 27.30 27.19
sha2-512 256 bytes 1.87 57.70 58.17
sha2-512 1024 bytes 6.26 92.97 97.56
sha2-512 8192 bytes 53.19 121.49 121.70
sha2-512 16384 bytes 103.87 123.73 123.76
CPU Usage %
Algorithm Accelerator (%) ARM CE (%) ARM (%)
aes-128-cbc 18% 99% 99%
aes-128-ecb 15% 99% 99%
aes-192-cbc 16% 99% 99%
aes-192-ecb 15% 99% 99%
aes-256-cbc 14% 99% 99%
sha2-256 16% 99% 99%
sha2-512 20% 98% 99%

Low Power Performance

Power Performance

These are power measurements taken while the device is in various low power modes like Deep Sleep, RTC Only + DDR, and RTC Only.

Deep Sleep Power Performance
Rail name Rail voltage(V) Power (mW)
vdd_core 0.75 7.17
soc_dvdd_1v8 1.80 1.47
soc_dvdd_3v3 3.29 3.38
vdda_1v8 1.80 1.02
vdd_lpddr4_pmic2 1.10 0.99
vdd_rtc 0.74 0.03
vdd_rtc_1v8 1.80 0.02
Total   14.07
RTC Only + DDR Power Performance
Rail name Rail voltage(V) Power (mW)
vdd_core 0.00 0.00
soc_dvdd_1v8 1.80 1.46
soc_dvdd_3v3 3.29 0.99
vdda_1v8 0.00 0.00
vdd_lpddr4_pmic2 1.10 0.95
vdd_rtc 0.75 0.02
vdd_rtc_1v8 1.80 0.01
Total   3.44
RTC Only Power Performance
Rail name Rail voltage(V) Power (mW)
vdd_core 0.00 0.00
soc_dvdd_1v8 0.00 0.00
soc_dvdd_3v3 0.00 0.00
vdda_1v8 0.00 0.00
vdd_lpddr4_pmic2 0.00 0.00
vdd_rtc 0.75 0.03
vdd_rtc_1v8 1.80 0.01
Total   0.05
Resume Latency Performance
LPM Resume Latency Performance
Low Power Mode Total Resume Latency (ms)
RTC Only Full normal boot time (~seconds)
RTC Only + DDR 385
Deep Sleep 171

The performance numbers are measured without the Linux printk logs. To remove the Linux printk logs, run the following commands in the terminal:

# Detach kernel serial console
consoles=$(find /sys/bus/platform/devices/*.serial/ -name console)
for console in ${consoles}; do
     echo -n N > ${console}
done

Note

The measurements shown are from using the default SDK with no extra optimizations.

Further optimizations are possible for these low power modes. Please refer to the AM62x Power Consumption App Note (https://www.ti.com/lit/pdf/spradg1)