Category:Lmbench for OMAPL1x LCDK

=OMAPL-138 LCDK=

About Lmbench
Lmbench is a suite of simple, portable, ANSI/C microbenchmarks for UNIX/POSIX. In general, it measures two key features: latency and bandwidth. Lmbench is intended to give system developers an insight into basic costs of key operations. Supports-
 * Bandwidth benchmarks
 * Cached file read
 * Memory copy (bcopy)
 * Memory read
 * Memory write
 * Pipe
 * TCP


 * Latency benchmarks
 * Context switching.
 * Networking: connection establishment, pipe, TCP, UDP, and RPC hot potato
 * File system creates and deletes.
 * Process creation.
 * Signal handling
 * System call overhead
 * Memory read latency


 * Miscellaneous
 * Processor clock rate calculation

Visit lmbench web page for more information.

Source Download Location

 * Visit Download lmbench

Cross compiling

 * Cross compilation command - make CC=$(TOOL_CHAIN_PREFIX)-gcc
 * Example: $make CC=arm-arago-linux-gnueabi-gcc
 * The TOOL_CHAIN_PREFIX corresponds to the tool chain in use. Set this based on your tool chain. Also make path to toolchain is exported as part of $PATH.

Test setup for OMAPL138 LCDK board

 * EVM booted up with SD boot configuration.

Execution with logs
root@omapl138-lcdk:/# ./lmbench_script.sh

=
===================================================== L M B E N C H  T E S T   R E S U L T S   O N   L C D K   B O A R D

=
=====================================================

=
========= BANDWIDTH MEASUREMENTS

=
=========

File read bandwidth with openclose: "bw_file_rd 1M open2close test.pdf" --- 1.00 39.52

File read bandwidth with io only: "bw_file_rd 1M io_only test.pdf" - 1.00 39.88

Memory read: "bw_mem 1M rd" -- 1.00 137.97

Mem write: "bw_mem 1M wr" --- 1.00 931.97

Mem Read/write: "bw_mem 1M rdwr" --- 1.00 81.92

Memory copy: "bw_mem 1M cp" -- 1.00 103.87

Mem- file write: "bw_mem 1M fwr" -- 1.00 253.58

Mem- File read: "bw_mem 1M frd"

1.00 110.88

Mem- File cp: "bw_mem 1M fcp" --- 1.00 79.67

Mem- bzero: "bw_mem 1M bzero" --- 1.00 253.68

Mem- bcopy: "bw_mem 1M bcopy" --- 1.00 82.98

Mmap with openclose: "bw_mmap_rd 1M open2close test.pdf" -- 1.00 70.82

Mmap only: "bw_mmap_rd 1M mmap_only test.pdf" - 1.00 111.33

tcp bandwidth: "bw_tcp localhost" --- 0.065536 26.36 MB/sec

Socket: "bw_unix"

AF_UNIX sock stream bandwidth: 32.28 MB/sec

=
========= LATENCY MEASUREMENTS

=
=========

latency for ls command: "lat_cmd ls" --- lat_cmd: 2768.3333 microseconds

latency for connect: "lat_connect localhost" -- TCP/IP connection cost to localhost: 2435.0000 microseconds

Latency- context switch: "lat_ctx -s 128K processes 2"

"size=128k ovr=4083.80

latency DRAM: "lat_dram_page -M 1M"

769.416264

Latency - fcntl: "lat_fcntl" - Fcntl lock latency: 362.8739 microseconds

Latency - fs: "lat_fs" --- 0k     15      1510    2482 1k     19      1228    2621 4k     17      857     3002 10k    12      753     2460

Latency - Mem rd: "lat_mem_rd 1M" -- "stride=128 0.00049 21.208 0.00098 21.332 0.00195 28.494 0.00293 28.257 0.00391 28.325 0.00586 28.837 0.00781 29.097 0.01172 28.749 0.01562 28.069 0.02344 371.544 0.03125 522.790 0.04688 591.058 0.06250 598.073 0.09375 608.639 0.12500 587.671 0.18750 603.131 0.25000 600.909 0.37500 622.018 0.50000 616.423 0.75000 632.235 1.00000 614.799

Latency -Mmap: "lat_mmap 1M test.pdf" - 1.000000 1052

Latency - Operations: "lat_ops" --- integer bit: 11.26 nanoseconds integer add: 8.21 nanoseconds integer mul: 4.25 nanoseconds integer div: 1233.25 nanoseconds integer mod: 429.01 nanoseconds int64 bit: 22.31 nanoseconds uint64 add: 22.31 nanoseconds int64 mul: 8.56 nanoseconds int64 div: 5768.79 nanoseconds int64 mod: 3445.91 nanoseconds float add: 350.16 nanoseconds float mul: 295.26 nanoseconds float div: 1509.59 nanoseconds double add: 643.04 nanoseconds double mul: 413.11 nanoseconds double div: 6207.06 nanoseconds float bogomflops: 3719.50 nanoseconds double bogomflops: 9978.50 nanoseconds

Latency -Page fault: "lat_pagefault test.pdf" -- Pagefaults on test.pdf: 38.1243 microseconds

Latency-Pipe: "lat_pipe" - Pipe latency: 805.7552 microseconds

latency-semaphore: "lat_sem" - Semaphore latency: 117.3243 microseconds

Latency -TCP select: "lat_select tcp"

Select on 200 tcp fd's: 682.0370 microseconds

Latency-Signals: "lat_sig install" -- Signal handler installation: 12.3278 microseconds Signal handler overhead: 55.6735 microseconds Protection fault: 17.4318 microseconds

Latency - system calls: "lat_syscall fstat test.pdf" -- Simple fstat: 12.9910 microseconds Simple stat: 39.1774 microseconds Simple open/close: 90.7987 microseconds Simple write: 10.6913 microseconds Simple read: 11.8267 microseconds Simple syscall: 7.9688 microseconds

Latency tcp: " lat_tcp -s; lat_tcp localhost" -- TCP latency using localhost: 2130.9959 microseconds Latency udp: "lat_udp -s; lat_udp localhost" -- UDP latency using localhost: 1611.1758 microseconds

latency-sockets: "lat_unix"

AF_UNIX sock stream latency: 1606.3333 microseconds

Latency sleep: "lat_usleep -u usleep 100"

usleep 100 microseconds: 335.1875 microseconds nanosleep 100 microseconds: 331.1765 microseconds select 100 microseconds: 538.8271 microseconds itimer 100 microseconds: 1286.5714 microseconds

Cache line size: "line" - ./lmbench_script.sh: line 213: line: command not found

lmdd testing: "lmdd if=internal of=/tmp/file count=1000 fsync=1" - 8.1920 MB in 0.2715 secs, 30.1703 MB/sec

Memory size: "memsize" ---

1

CPU frequency: "mhz" - -1 System too busy

Test sleep - 3 secs: "msleep 3000" -

Parallel memory ops: "par_mem -M 1M" -- 0.004096 17.39 0.008192 1.56 0.016384 18.30 0.032768 13.82 0.065536 9.35 0.131072 7.65 0.262144 19.90 0.524288 25.61

Parallel ops: "par_ops"

integer bit parallelism: 1.02 integer add parallelism: 1.30 integer mul parallelism: 1.07 integer div parallelism: 1.00 integer mod parallelism: 1.00 int64 bit parallelism: 1.05 int64 add parallelism: 1.08 int64 mul parallelism: 1.04 int64 div parallelism: 1.00 int64 mod parallelism: 1.01 float add parallelism: 1.00 float mul parallelism: 1.00 float div parallelism: 1.00 double add parallelism: 1.00 double mul parallelism: 1.00 double div parallelism: 1.00

STREAM ops: "stream -M 128K" - STREAM copy latency: 96.81 nanoseconds STREAM copy bandwidth: 165.28 MB/sec STREAM scale latency: 267.67 nanoseconds STREAM scale bandwidth: 59.77 MB/sec STREAM add latency: 479.34 nanoseconds STREAM add bandwidth: 50.07 MB/sec STREAM triad latency: 1284.27 nanoseconds STREAM triad bandwidth: 18.69 MB/sec

Translation lookaside buffer: "tlb -M 1M" -- tlb: 8 pages

=
=============== LMBENCH BENCHMARK COMPLETED

=
=============== root@omapl138-lcdk:/#

Test setup for OMAPL138 SDI (Spectrum Digital Inc) EVM board

 * EVM booted up with NFS boot configuration.

Execution with logs
root@omapl138-evm:/# ./lmbench_script.sh

=
===================================================== L M B E N C H	T E S T	 R E S U L T S   O N   L C D K   B O A R D

=
=====================================================

=
========= BANDWIDTH MEASUREMENTS

=
=========

File read bandwidth with openclose: "bw_file_rd 1M open2close test.pdf" --- 1.00 329.49

File read bandwidth with io only: "bw_file_rd 1M io_only test.pdf" - 1.00 645.37

Memory read: "bw_mem 1M rd" -- 1.00 121.70

Mem write: "bw_mem 1M wr" --- 1.00 864.30

Mem Read/write: "bw_mem 1M rdwr" --- 1.00 75.58

Memory copy: "bw_mem 1M cp" -- 1.00 93.62

Mem- file write: "bw_mem 1M fwr" -- 1.00 236.69

Mem- File read: "bw_mem 1M frd"

1.00 101.35

Mem- File cp: "bw_mem 1M fcp" --- 1.00 74.42

Mem- bzero: "bw_mem 1M bzero" --- 1.00 235.88

Mem- bcopy: "bw_mem 1M bcopy" --- 1.00 77.18

Mmap with openclose: "bw_mmap_rd 1M open2close test.pdf" -- 1.00 inf

Mmap only: "bw_mmap_rd 1M mmap_only test.pdf" - 1.00 inf

tcp bandwidth: "bw_tcp localhost" --- 0.065536 18.30 MB/sec

Socket: "bw_unix"

AF_UNIX sock stream bandwidth: 25.83 MB/sec

=
========= LATENCY MEASUREMENTS

=
=========

latency for ls command: "lat_cmd ls" --- lat_cmd: 3480.5000 microseconds

latency for connect: "lat_connect localhost" -- TCP/IP connection cost to localhost: 4835.0000 microseconds

Latency- context switch: "lat_ctx -s 128K processes 2"

"size=128k ovr=4887.60

latency DRAM: "lat_dram_page -M 1M"

1010.947621

Latency - fcntl: "lat_fcntl" - Fcntl lock latency: 391.7500 microseconds

Latency - fs: "lat_fs" --- 0k	15	608	695 1k	10	459	600 4k	9	408	766 10k	7	1140	708

Latency - Mem rd: "lat_mem_rd 1M" -- "stride=128 0.00049 18.340 0.00098 20.305 0.00195 29.064 0.00293 27.796 0.00391 29.321 0.00586 28.011 0.00781 28.988 0.01172 28.046 0.01562 29.267 0.02344 372.199 0.03125 633.957 0.04688 659.171 0.06250 661.608 0.09375 634.755 0.12500 685.739 0.18750 676.165 0.25000 711.864 0.37500 717.554 0.50000 648.273 0.75000 656.263 1.00000 617.067

Latency -Mmap: "lat_mmap 1M test.pdf" -

Latency - Operations: "lat_ops" --- integer bit: 13.65 nanoseconds integer add: 9.30 nanoseconds integer mul: 5.23 nanoseconds integer div: 1776.14 nanoseconds integer mod: 431.00 nanoseconds int64 bit: 21.28 nanoseconds uint64 add: 22.55 nanoseconds int64 mul: 10.68 nanoseconds int64 div: 5610.00 nanoseconds int64 mod: 3829.77 nanoseconds float add: 415.04 nanoseconds float mul: 313.12 nanoseconds float div: 1573.33 nanoseconds double add: 730.89 nanoseconds double mul: 419.82 nanoseconds double div: 7657.59 nanoseconds float bogomflops: 4128.33 nanoseconds double bogomflops: 9809.00 nanoseconds

Latency -Page fault: "lat_pagefault test.pdf" --

Latency-Pipe: "lat_pipe" - Pipe latency: 1024.9239 microseconds

latency-semaphore: "lat_sem" - Semaphore latency: 139.6828 microseconds

Latency -TCP select: "lat_select tcp"

Select on 200 tcp fd's: 704.1481 microseconds

Latency-Signals: "lat_sig install" -- Signal handler installation: 25.5830 microseconds Signal handler overhead: 68.3860 microseconds Protection fault: 8.1411 microseconds

Latency - system calls: "lat_syscall fstat test.pdf" -- Simple fstat: 69.5961 microseconds Simple stat: 290.1600 microseconds Simple open/close: 5054.0000 microseconds Simple write: 23.3622 microseconds Simple read: 11.6025 microseconds Simple syscall: 4.8681 microseconds

Latency tcp: " lat_tcp -s; lat_tcp localhost" -- TCP latency using localhost: 2672.8475 microseconds Latency udp: "lat_udp -s; lat_udp localhost" -- UDP latency using localhost: 1966.9336 microseconds

latency-sockets: "lat_unix"

AF_UNIX sock stream latency: 2102.7143 microseconds

Latency sleep: "lat_usleep -u usleep 100"

usleep 100 microseconds: 738.5833 microseconds nanosleep 100 microseconds: 377.0714 microseconds select 100 microseconds: 1040.2143 microseconds itimer 100 microseconds: 1105.0556 microseconds

lmdd testing: "lmdd if=internal of=/tmp/file count=1000 fsync=1" - 8.1920 MB in 0.5973 secs, 13.7151 MB/sec

Memory size: "memsize" ---

1

CPU frequency: "mhz" - 275 MHz, 3.6364 nanosec clock

Test sleep - 3 secs: "msleep 3000" -

Parallel memory ops: "par_mem -M 1M" -- 0.004096 4.32 0.008192 35.19 0.016384 73.18 0.032768 3.70 0.065536 18.46 0.131072 20.33 0.262144 1.00 0.524288 1.19

Parallel ops: "par_ops"

integer bit parallelism: 1.02 integer add parallelism: 3.25 integer mul parallelism: 3.34 integer div parallelism: 1.00 integer mod parallelism: 1.00 int64 bit parallelism: 1.00 int64 add parallelism: 2.91 int64 mul parallelism: 2.81 int64 div parallelism: 1.01 int64 mod parallelism: 1.01 float add parallelism: 1.00 float mul parallelism: 1.00 float div parallelism: 1.00 double add parallelism: 1.69 double mul parallelism: 1.00 double div parallelism: 1.00

STREAM ops: "stream -M 128K" - STREAM copy latency: 106.02 nanoseconds STREAM copy bandwidth: 150.92 MB/sec STREAM scale latency: 656.33 nanoseconds STREAM scale bandwidth: 24.38 MB/sec STREAM add latency: 1081.51 nanoseconds STREAM add bandwidth: 22.19 MB/sec STREAM triad latency: 2026.25 nanoseconds STREAM triad bandwidth: 11.84 MB/sec

Translation lookaside buffer: "tlb -M 1M" -- tlb: 8 pages

=
=============== LMBENCH BENCHMARK COMPLETED

=
=============== root@omapl138-evm:/#

Test setup for OMAPL137 SDI (Spectrum Digital Inc) EVM board

 * EVM booted up with NFS boot configuration.

Execution with logs
root@omapl137-evm:/# ./lmbench_script.sh

=
===================================================== L M B E N C H	T E S T	 R E S U L T S   O N   L C D K   B O A R D

=
=====================================================

=
========= BANDWIDTH MEASUREMENTS

=
=========

File read bandwidth with openclose: "bw_file_rd 1M open2close test.pdf" --- 1.00 358.94

File read bandwidth with io only: "bw_file_rd 1M io_only test.pdf" - 1.00 691.09

Memory read: "bw_mem 1M rd" -- 1.00 138.56

Mem write: "bw_mem 1M wr" --- 1.00 938.09

Mem Read/write: "bw_mem 1M rdwr" --- 1.00 94.75

Memory copy: "bw_mem 1M cp" -- 1.00 86.72

Mem- file write: "bw_mem 1M fwr" -- 1.00 254.74

Mem- File read: "bw_mem 1M frd"

1.00 113.60

Mem- File cp: "bw_mem 1M fcp" --- 1.00 70.30

Mem- bzero: "bw_mem 1M bzero" --- 1.00 254.36

Mem- bcopy: "bw_mem 1M bcopy" --- 1.00 88.70

Mmap with openclose: "bw_mmap_rd 1M open2close test.pdf" -- 1.00 inf

Mmap only: "bw_mmap_rd 1M mmap_only test.pdf" - 1.00 inf

tcp bandwidth: "bw_tcp localhost" --- 0.065536 26.86 MB/sec

Socket: "bw_unix"

AF_UNIX sock stream bandwidth: 36.64 MB/sec

=
========= LATENCY MEASUREMENTS

=
=========

latency for ls command: "lat_cmd ls" --- lat_cmd: 3170.0000 microseconds

latency for connect: "lat_connect localhost" -- TCP/IP connection cost to localhost: 3509.4286 microseconds

Latency- context switch: "lat_ctx -s 128K processes 2"

"size=128k ovr=5163.40

latency DRAM: "lat_dram_page -M 1M"

828.768493

Latency - fcntl: "lat_fcntl" - Fcntl lock latency: 352.4714 microseconds

Latency - fs: "lat_fs" --- 0k	14	550	829 1k	10	517	951 4k	10	392	917 10k	2	393	763

Latency - Mem rd: "lat_mem_rd 1M" -- "stride=128 0.00049 20.963 0.00098 20.180 0.00195 24.186 0.00293 27.005 0.00391 25.792 0.00586 27.119 0.00781 29.907 0.01172 26.380 0.01562 28.630 0.02344 332.807 0.03125 529.307 0.04688 577.818 0.06250 519.881 0.09375 520.526 0.12500 526.923 0.18750 549.832 0.25000 521.169 0.37500 543.594 0.50000 574.296 0.75000 609.014 1.00000 592.644

Latency -Mmap: "lat_mmap 1M test.pdf" -

Latency - Operations: "lat_ops" --- integer bit: 12.50 nanoseconds integer add: 8.45 nanoseconds integer mul: 4.52 nanoseconds integer div: 1530.31 nanoseconds integer mod: 439.00 nanoseconds int64 bit: 18.48 nanoseconds uint64 add: 26.26 nanoseconds int64 mul: 15.12 nanoseconds int64 div: 5238.18 nanoseconds int64 mod: 4089.78 nanoseconds float add: 390.14 nanoseconds float mul: 274.37 nanoseconds float div: 1808.37 nanoseconds double add: 667.85 nanoseconds double mul: 467.74 nanoseconds double div: 7996.15 nanoseconds float bogomflops: 3651.00 nanoseconds double bogomflops: 10352.00 nanoseconds

Latency -Page fault: "lat_pagefault test.pdf" --

Latency-Pipe: "lat_pipe" - Pipe latency: 947.8743 microseconds

latency-semaphore: "lat_sem" - Semaphore latency: 145.9000 microseconds

Latency -TCP select: "lat_select tcp"

Select on 200 tcp fd's: 842.5000 microseconds

Latency-Signals: "lat_sig install" -- Signal handler installation: 31.2089 microseconds Signal handler overhead: 92.6519 microseconds Protection fault: 41.6277 microseconds

Latency - system calls: "lat_syscall fstat test.pdf" -- Simple fstat: 62.1209 microseconds Simple stat: 199.9535 microseconds Simple open/close: 4579.5000 microseconds Simple write: 25.4886 microseconds Simple read: 11.8981 microseconds Simple syscall: 4.7163 microseconds

Latency tcp: " lat_tcp -s; lat_tcp localhost" -- TCP latency using localhost: 2517.3024 microseconds Latency udp: "lat_udp -s; lat_udp localhost" -- UDP latency using localhost: 1855.7055 microseconds

latency-sockets: "lat_unix"

AF_UNIX sock stream latency: 1948.3333 microseconds

Latency sleep: "lat_usleep -u usleep 100"

usleep 100 microseconds: 361.2500 microseconds nanosleep 100 microseconds: 351.9375 microseconds select 100 microseconds: 779.8125 microseconds itimer 100 microseconds: 900.5714 microseconds

lmdd testing: "lmdd if=internal of=/tmp/file count=1000 fsync=1" - 8.1920 MB in 0.4803 secs, 17.0548 MB/sec

Memory size: "memsize" ---

1

CPU frequency: "mhz" - -1 System too busy

Test sleep - 3 secs: "msleep 3000" -

Parallel memory ops: "par_mem -M 1M" -- 0.004096 11.84 0.008192 42.47 0.016384 17.25 0.032768 15.05 0.065536 10.54 0.131072 14.06 0.262144 12.92 0.524288 31.31

Parallel ops: "par_ops"

integer bit parallelism: 1.02 integer add parallelism: 2.97 integer mul parallelism: 3.37 integer div parallelism: 1.00 integer mod parallelism: 1.01 int64 bit parallelism: 1.00 int64 add parallelism: 3.10 int64 mul parallelism: 2.61 int64 div parallelism: 1.00 int64 mod parallelism: 1.01 float add parallelism: 1.00 float mul parallelism: 1.00 float div parallelism: 1.00 double add parallelism: 1.00 double mul parallelism: 1.00 double div parallelism: 1.00

STREAM ops: "stream -M 128K" - STREAM copy latency: 80.21 nanoseconds STREAM copy bandwidth: 199.47 MB/sec STREAM scale latency: 260.27 nanoseconds STREAM scale bandwidth: 61.48 MB/sec STREAM add latency: 462.87 nanoseconds STREAM add bandwidth: 51.85 MB/sec STREAM triad latency: 1525.66 nanoseconds STREAM triad bandwidth: 15.73 MB/sec

Translation lookaside buffer: "tlb -M 1M" -- tlb: 8 pages

=
=============== LMBENCH BENCHMARK COMPLETED

=
=============== root@omapl137-evm:/#

Download lmbench script