2.2. Linux 09.01.00 Performance Guide

Read This First

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

Name

Description

AM437x-gpevm

AM437x-gpevm Evaluation Module rev 1.5A with ARM running at 1000MHz, DDR3-400 (400MHz/800 MT/S), TMDSEVM437X

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.

If You Need Assistance

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

2.2.1. System Benchmarks

2.2.1.1. 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 http://lmbench.sourceforge.net/whatis_lmbench.html and http://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

Benchmarks

am43xx-gpevm: perf

af_unix_sock_stream_latency (microsec)

31.75

af_unix_socket_stream_bandwidth (MBs)

196.34

bw_file_rd-io-1mb (MB/s)

199.78

bw_file_rd-o2c-1mb (MB/s)

164.61

bw_mem-bcopy-16mb (MB/s)

193.25

bw_mem-bcopy-1mb (MB/s)

198.33

bw_mem-bcopy-2mb (MB/s)

198.83

bw_mem-bcopy-4mb (MB/s)

197.07

bw_mem-bcopy-8mb (MB/s)

198.09

bw_mem-bzero-16mb (MB/s)

672.58

bw_mem-bzero-1mb (MB/s)

434.40 (min 198.33, max 670.47)

bw_mem-bzero-2mb (MB/s)

436.23 (min 198.83, max 673.63)

bw_mem-bzero-4mb (MB/s)

433.71 (min 197.07, max 670.35)

bw_mem-bzero-8mb (MB/s)

435.78 (min 198.09, max 673.46)

bw_mem-cp-16mb (MB/s)

194.28

bw_mem-cp-1mb (MB/s)

437.96 (min 201.49, max 674.42)

bw_mem-cp-2mb (MB/s)

436.93 (min 200.68, max 673.17)

bw_mem-cp-4mb (MB/s)

437.07 (min 200.73, max 673.40)

bw_mem-cp-8mb (MB/s)

437.09 (min 201.01, max 673.17)

bw_mem-fcp-16mb (MB/s)

243.87

bw_mem-fcp-1mb (MB/s)

458.69 (min 246.91, max 670.47)

bw_mem-fcp-2mb (MB/s)

457.97 (min 242.31, max 673.63)

bw_mem-fcp-4mb (MB/s)

455.64 (min 240.92, max 670.35)

bw_mem-fcp-8mb (MB/s)

459.15 (min 244.84, max 673.46)

bw_mem-frd-16mb (MB/s)

202.94

bw_mem-frd-1mb (MB/s)

227.82 (min 208.72, max 246.91)

bw_mem-frd-2mb (MB/s)

224.55 (min 206.78, max 242.31)

bw_mem-frd-4mb (MB/s)

220.84 (min 200.75, max 240.92)

bw_mem-frd-8mb (MB/s)

223.39 (min 201.93, max 244.84)

bw_mem-fwr-16mb (MB/s)

672.13

bw_mem-fwr-1mb (MB/s)

441.57 (min 208.72, max 674.42)

bw_mem-fwr-2mb (MB/s)

439.98 (min 206.78, max 673.17)

bw_mem-fwr-4mb (MB/s)

437.08 (min 200.75, max 673.40)

bw_mem-fwr-8mb (MB/s)

437.55 (min 201.93, max 673.17)

bw_mem-rd-16mb (MB/s)

314.02

bw_mem-rd-1mb (MB/s)

294.78 (min 273.19, max 316.36)

bw_mem-rd-2mb (MB/s)

291.84 (min 270.05, max 313.63)

bw_mem-rd-4mb (MB/s)

292.39 (min 269.96, max 314.81)

bw_mem-rd-8mb (MB/s)

292.56 (min 270.78, max 314.33)

bw_mem-rdwr-16mb (MB/s)

245.76

bw_mem-rdwr-1mb (MB/s)

225.89 (min 201.49, max 250.28)

bw_mem-rdwr-2mb (MB/s)

223.19 (min 200.68, max 245.70)

bw_mem-rdwr-4mb (MB/s)

223.16 (min 200.73, max 245.59)

bw_mem-rdwr-8mb (MB/s)

223.75 (min 201.01, max 246.48)

bw_mem-wr-16mb (MB/s)

270.81

bw_mem-wr-1mb (MB/s)

261.74 (min 250.28, max 273.19)

bw_mem-wr-2mb (MB/s)

257.88 (min 245.70, max 270.05)

bw_mem-wr-4mb (MB/s)

257.78 (min 245.59, max 269.96)

bw_mem-wr-8mb (MB/s)

258.63 (min 246.48, max 270.78)

bw_mmap_rd-mo-1mb (MB/s)

210.39

bw_mmap_rd-o2c-1mb (MB/s)

161.06

bw_pipe (MB/s)

391.78

bw_unix (MB/s)

196.34

lat_connect (us)

76.04

lat_ctx-2-128k (us)

0.00

lat_ctx-2-256k (us)

0.00

lat_ctx-4-128k (us)

0.00

lat_ctx-4-256k (us)

0.00

lat_fs-0k (num_files)

233.00

lat_fs-10k (num_files)

75.00

lat_fs-1k (num_files)

120.00

lat_fs-4k (num_files)

123.00

lat_mem_rd-stride128-sz1000k (ns)

98.27

lat_mem_rd-stride128-sz125k (ns)

16.01

lat_mem_rd-stride128-sz250k (ns)

42.07

lat_mem_rd-stride128-sz31k (ns)

4.01

lat_mem_rd-stride128-sz50 (ns)

4.01

lat_mem_rd-stride128-sz500k (ns)

88.77

lat_mem_rd-stride128-sz62k (ns)

16.91

lat_mmap-1m (us)

76.00

lat_ops-double-add (ns)

5.02

lat_ops-double-div (ns)

25.09

lat_ops-double-mul (ns)

7.02

lat_ops-float-add (ns)

5.04

lat_ops-float-div (ns)

15.05

lat_ops-float-mul (ns)

6.02

lat_ops-int-add (ns)

1.03

lat_ops-int-bit (ns)

0.76

lat_ops-int-div (ns)

96.06

lat_ops-int-mod (ns)

23.91

lat_ops-int-mul (ns)

4.14

lat_ops-int64-add (ns)

1.12

lat_ops-int64-bit (ns)

0.67

lat_ops-int64-div (ns)

200.99

lat_ops-int64-mod (ns)

42.62

lat_ops-int64-mul (ns)

4.16

lat_pagefault (us)

1.42

lat_pipe (us)

29.91

lat_proc-exec (us)

1353.25

lat_proc-fork (us)

988.17

lat_proc-proccall (us)

0.01

lat_select (us)

33.01

lat_sem (us)

4.47

lat_sig-catch (us)

5.81

lat_sig-install (us)

0.88

lat_sig-prot (us)

0.57

lat_syscall-fstat (us)

3.43

lat_syscall-null (us)

0.39

lat_syscall-open (us)

263.14

lat_syscall-read (us)

0.67

lat_syscall-stat (us)

5.04

lat_syscall-write (us)

0.55

lat_tcp (us)

0.87

lat_unix (us)

31.75

latency_for_0.50_mb_block_size (nanosec)

88.77

latency_for_1.00_mb_block_size (nanosec)

49.13 (min 0.00, max 98.27)

pipe_bandwidth (MBs)

391.78

pipe_latency (microsec)

29.91

procedure_call (microsec)

0.01

select_on_200_tcp_fds (microsec)

33.01

semaphore_latency (microsec)

4.47

signal_handler_latency (microsec)

0.88

signal_handler_overhead (microsec)

5.81

tcp_ip_connection_cost_to_localhost (microsec)

76.04

tcp_latency_using_localhost (microsec)

0.87

Table: LM Bench Metrics

2.2.1.2. Dhrystone

Dhrystone is a core only benchmark that runs from warm L1 caches in all modern processors. It scales linearly with clock speed. For standard ARM cores the DMIPS/MHz score will be identical with the same compiler and flags.

Execute the benchmark with the following:

runDhrystone

Benchmarks

am43xx-gpevm: perf

cpu_clock (MHz)

1000.00

dhrystone_per_mhz (DMIPS/MHz)

2.50

dhrystone_per_second (DhrystoneP)

4444444.50

Table: Dhrystone Benchmark

2.2.1.3. Whetstone

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

Execute the benchmark with the following:

runWhetstone

Benchmarks

am43xx-gpevm: perf

whetstone (MIPS)

3333.30

Table: Whetstone Benchmark

2.2.1.4. Linpack

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

Benchmarks

am43xx-gpevm: perf

linpack (Kflops)

192209.00

Table: Linpack Benchmark

2.2.1.5. 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

Benchmarks

am43xx-gpevm: perf

assignment (Iterations)

7.20

fourier (Iterations)

13053.00

fp_emulation (Iterations)

75.00

huffman (Iterations)

855.02

idea (Iterations)

1830.40

lu_decomposition (Iterations)

338.30

neural_net (Iterations)

9.63

numeric_sort (Iterations)

503.24

string_sort (Iterations)

64.02

Table: NBench Benchmarks

2.2.1.6. 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

Benchmarks

am43xx-gpevm: perf

add (MB/s)

327.10

copy (MB/s)

435.10

scale (MB/s)

654.20

triad (MB/s)

381.50

Table: Stream

2.2.2. Boot-time Measurement

2.2.2.1. Boot media: MMCSD

Boot Configuration

am43xx-gpevm: boot time (sec)

Kernel boot time test when bootloader, kernel and sdk-rootfs are in mmc-sd

46.96 (min 46.73, max 47.46)

Kernel boot time test when init is /bin/sh and bootloader, kernel and sdk-rootfs are in mmc-sd

8.90 (min 8.79, max 9.26)

Table: Boot time MMC/SD

2.2.3. ALSA SoC Audio Driver

  1. Access type - RW_INTERLEAVED

  2. Channels - 2

  3. Format - S16_LE

  4. Period size - 64

Sampling Rate (Hz)

am43xx-gpevm: Throughput (bits/sec)

am43xx-gpevm: CPU Load (%)

8000

255997.00

0.54

11025

352795.00

0.63

16000

511993.00

0.75

22050

705591.00

1.09

24000

705588.00

0.87

32000

1023983.00

0.36

44100

1411177.00

1.43

48000

1535975.00

1.66

88200

2822352.00

2.86

96000

3071948.00

0.65

Table: Audio Capture


Sampling Rate (Hz)

am43xx-gpevm: Throughput (bits/sec)

am43xx-gpevm: CPU Load (%)

8000

256102.00

0.70

11025

352941.00

0.52

16000

512204.00

1.67

22050

705881.00

0.96

24000

705880.00

0.78

32000

1024406.00

0.16

44100

1411758.00

1.27

48000

1536609.00

2.42

88200

2823517.00

2.55

96000

3073215.00

0.57

Table: Audio Playback


2.2.4. NAND Driver

2.2.4.1. AM43XX-GPEVM

Buffer size (bytes)

am43xx-gpevm: Write UBIFS Throughput (Mbytes/sec)

am43xx-gpevm: Write UBIFS CPU Load (%)

am43xx-gpevm: Read UBIFS Throughput (Mbytes/sec)

am43xx-gpevm: Read UBIFS CPU Load (%)

102400

5.17 (min 5.10, max 5.37)

77.66 (min 76.81, max 77.96)

10.67

39.02

262144

5.11 (min 5.08, max 5.16)

77.82 (min 77.70, max 78.00)

10.66

39.98

524288

5.12 (min 5.08, max 5.15)

77.62 (min 77.22, max 77.94)

10.66

40.94

1048576

5.11 (min 5.09, max 5.14)

77.67 (min 77.47, max 78.14)

10.67

41.04

5242880

5.12 (min 5.07, max 5.16)

77.81 (min 77.47, max 78.25)

10.66

40.25

2.2.5. MMC/SD Driver

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.


2.2.5.1. AM43XX-GPEVM


Buffer size (bytes)

am43xx-gpevm: Write EXT4 Throughput (Mbytes/sec)

am43xx-gpevm: Write EXT4 CPU Load (%)

am43xx-gpevm: Read EXT4 Throughput (Mbytes/sec)

am43xx-gpevm: Read EXT4 CPU Load (%)

1m

5.90

3.40

26.60

8.92

4m

9.41

2.41

26.60

4.40

4k

2.56

12.38

8.59

31.77

256k

7.97

3.26

26.60

9.68


The performance numbers were captured using the following:

  • SanDisk 8GB MicroSDHC Class 10 Memory Card

  • Partition was mounted with async option


2.2.6. UBoot MMC/SD Driver


2.2.6.1. AM43XX-GPEVM

File size (bytes in hex)

am43xx-gpevm: Write Throughput (Kbytes/sec)

am43xx-gpevm: Read Throughput (Kbytes/sec)

400000

10556.70

18618.18

800000

9649.00

19095.57

1000000

7713.75

19389.35

The performance numbers were captured using the following:

  • SanDisk 8GB MicroSDHC Class 10 Memory Card


2.2.7. USB Driver

2.2.7.1. USB Device Controller

Number of Blocks

am43xx-gpevm: Throughput (MB/sec)

150

18.30

Table: USBDEVICE HIGHSPEED SLAVE READ THROUGHPUT



Number of Blocks

am43xx-gpevm: Throughput (MB/sec)

150

21.90

Table: USBDEVICE HIGHSPEED SLAVE WRITE THROUGHPUT