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

AM335x EVM

AM335x Evaluation Module rev 1.5B with ARM running at 1000MHz, DDR3-400 (400MHz/800 MT/S), TMDXEVM3358

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

am335x-evm: perf

af_unix_sock_stream_latency (microsec)

40.96

af_unix_socket_stream_bandwidth (MBs)

186.32

bw_file_rd-io-1mb (MB/s)

157.62

bw_file_rd-o2c-1mb (MB/s)

147.34

bw_mem-bcopy-16mb (MB/s)

203.35

bw_mem-bcopy-1mb (MB/s)

205.80

bw_mem-bcopy-2mb (MB/s)

206.70

bw_mem-bcopy-4mb (MB/s)

200.71

bw_mem-bcopy-8mb (MB/s)

193.36

bw_mem-bzero-16mb (MB/s)

996.33

bw_mem-bzero-1mb (MB/s)

595.43 (min 205.80, max 985.06)

bw_mem-bzero-2mb (MB/s)

595.40 (min 206.70, max 984.09)

bw_mem-bzero-4mb (MB/s)

592.30 (min 200.71, max 983.89)

bw_mem-bzero-8mb (MB/s)

588.81 (min 193.36, max 984.25)

bw_mem-cp-16mb (MB/s)

193.71

bw_mem-cp-1mb (MB/s)

582.73 (min 182.98, max 982.48)

bw_mem-cp-2mb (MB/s)

589.61 (min 195.12, max 984.09)

bw_mem-cp-4mb (MB/s)

598.18 (min 192.08, max 1004.27)

bw_mem-cp-8mb (MB/s)

587.30 (min 185.47, max 989.12)

bw_mem-fcp-16mb (MB/s)

184.96

bw_mem-fcp-1mb (MB/s)

588.96 (min 192.86, max 985.06)

bw_mem-fcp-2mb (MB/s)

586.05 (min 188.01, max 984.09)

bw_mem-fcp-4mb (MB/s)

584.21 (min 184.53, max 983.89)

bw_mem-fcp-8mb (MB/s)

581.35 (min 178.44, max 984.25)

bw_mem-frd-16mb (MB/s)

250.22

bw_mem-frd-1mb (MB/s)

230.66 (min 192.86, max 268.46)

bw_mem-frd-2mb (MB/s)

219.60 (min 188.01, max 251.19)

bw_mem-frd-4mb (MB/s)

217.29 (min 184.53, max 250.05)

bw_mem-frd-8mb (MB/s)

214.27 (min 178.44, max 250.09)

bw_mem-fwr-16mb (MB/s)

996.02

bw_mem-fwr-1mb (MB/s)

625.47 (min 268.46, max 982.48)

bw_mem-fwr-2mb (MB/s)

617.64 (min 251.19, max 984.09)

bw_mem-fwr-4mb (MB/s)

627.16 (min 250.05, max 1004.27)

bw_mem-fwr-8mb (MB/s)

619.61 (min 250.09, max 989.12)

bw_mem-rd-16mb (MB/s)

252.78

bw_mem-rd-1mb (MB/s)

627.70 (min 270.82, max 984.57)

bw_mem-rd-2mb (MB/s)

620.30 (min 255.04, max 985.55)

bw_mem-rd-4mb (MB/s)

628.24 (min 252.84, max 1003.64)

bw_mem-rd-8mb (MB/s)

628.84 (min 252.53, max 1005.15)

bw_mem-rdwr-16mb (MB/s)

204.39

bw_mem-rdwr-1mb (MB/s)

195.46 (min 182.98, max 207.94)

bw_mem-rdwr-2mb (MB/s)

199.62 (min 195.12, max 204.12)

bw_mem-rdwr-4mb (MB/s)

198.43 (min 192.08, max 204.77)

bw_mem-rdwr-8mb (MB/s)

194.90 (min 185.47, max 204.32)

bw_mem-wr-16mb (MB/s)

996.02

bw_mem-wr-1mb (MB/s)

596.26 (min 207.94, max 984.57)

bw_mem-wr-2mb (MB/s)

594.84 (min 204.12, max 985.55)

bw_mem-wr-4mb (MB/s)

604.21 (min 204.77, max 1003.64)

bw_mem-wr-8mb (MB/s)

604.74 (min 204.32, max 1005.15)

bw_mmap_rd-mo-1mb (MB/s)

253.55

bw_mmap_rd-o2c-1mb (MB/s)

167.36

bw_pipe (MB/s)

322.54

bw_unix (MB/s)

186.32

lat_connect (us)

85.15

lat_ctx-2-128k (us)

28.67

lat_ctx-2-256k (us)

5.36

lat_ctx-4-128k (us)

34.51

lat_ctx-4-256k (us)

0.00

lat_fs-0k (num_files)

198.00

lat_fs-10k (num_files)

74.00

lat_fs-1k (num_files)

111.00

lat_fs-4k (num_files)

101.00

lat_mem_rd-stride128-sz1000k (ns)

220.85

lat_mem_rd-stride128-sz125k (ns)

11.74

lat_mem_rd-stride128-sz250k (ns)

62.97

lat_mem_rd-stride128-sz31k (ns)

3.03

lat_mem_rd-stride128-sz50 (ns)

3.00

lat_mem_rd-stride128-sz500k (ns)

183.34

lat_mem_rd-stride128-sz62k (ns)

8.87

lat_mmap-1m (us)

117.00

lat_ops-double-add (ns)

8.97

lat_ops-double-div (ns)

57.32

lat_ops-double-mul (ns)

11.08

lat_ops-float-add (ns)

8.97

lat_ops-float-div (ns)

33.20

lat_ops-float-mul (ns)

10.06

lat_ops-int-add (ns)

1.01

lat_ops-int-bit (ns)

0.67

lat_ops-int-div (ns)

72.83

lat_ops-int-mod (ns)

25.49

lat_ops-int-mul (ns)

6.64

lat_ops-int64-add (ns)

1.06

lat_ops-int64-bit (ns)

0.68

lat_ops-int64-div (ns)

188.60

lat_ops-int64-mod (ns)

52.12

lat_ops-int64-mul (ns)

6.07

lat_pagefault (us)

1.62

lat_pipe (us)

41.00

lat_proc-exec (us)

1521.00

lat_proc-fork (us)

1191.40

lat_proc-proccall (us)

0.02

lat_select (us)

41.44

lat_sem (us)

6.11

lat_sig-catch (us)

8.24

lat_sig-install (us)

1.33

lat_sig-prot (us)

0.74

lat_syscall-fstat (us)

4.55

lat_syscall-null (us)

0.49

lat_syscall-open (us)

316.11

lat_syscall-read (us)

0.86

lat_syscall-stat (us)

6.38

lat_syscall-write (us)

0.67

lat_tcp (us)

1.05

lat_unix (us)

40.96

latency_for_0.50_mb_block_size (nanosec)

183.34

latency_for_1.00_mb_block_size (nanosec)

110.42 (min 0.00, max 220.85)

pipe_bandwidth (MBs)

322.54

pipe_latency (microsec)

41.00

procedure_call (microsec)

0.02

select_on_200_tcp_fds (microsec)

41.44

semaphore_latency (microsec)

6.11

signal_handler_latency (microsec)

1.33

signal_handler_overhead (microsec)

8.24

tcp_ip_connection_cost_to_localhost (microsec)

85.15

tcp_latency_using_localhost (microsec)

1.05

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

am335x-evm: perf

cpu_clock (MHz)

1000.00

dhrystone_per_mhz (DMIPS/MHz)

2.10

dhrystone_per_second (DhrystoneP)

3773585.00

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

am335x-evm: perf

whetstone (MIPS)

833.30

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

am335x-evm: perf

linpack (Kflops)

50419.00

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

am335x-evm: perf

assignment (Iterations)

7.77

fourier (Iterations)

2891.70

fp_emulation (Iterations)

70.38

huffman (Iterations)

727.35

idea (Iterations)

1706.30

lu_decomposition (Iterations)

76.53

neural_net (Iterations)

2.21

numeric_sort (Iterations)

418.37

string_sort (Iterations)

67.17

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

am335x-evm: perf

add (MB/s)

528.70

copy (MB/s)

486.60

scale (MB/s)

614.30

triad (MB/s)

445.60

Table: Stream

2.2.2. ALSA SoC Audio Driver

  1. Access type - RW_INTERLEAVED

  2. Channels - 2

  3. Format - S16_LE

  4. Period size - 64

Sampling Rate (Hz)

am335x-evm: Throughput (bits/sec)

am335x-evm: CPU Load (%)

8000

255998.00

2.27

11025

352796.00

2.61

16000

511995.00

3.75

22050

705594.00

4.14

24000

705592.00

3.44

32000

1023988.00

7.47

44100

1411183.00

5.03

48000

1535983.00

2.48

88200

2822364.00

7.95

96000

3071961.00

12.37

Table: Audio Capture


Sampling Rate (Hz)

am335x-evm: Throughput (bits/sec)

am335x-evm: CPU Load (%)

8000

256101.00

0.85

11025

352940.00

0.78

16000

512203.00

0.17

22050

705880.00

1.55

24000

705878.00

1.29

32000

1024404.00

0.29

44100

1411756.00

2.13

48000

1536606.00

4.54

88200

2823510.00

4.36

96000

3073208.00

0.77

Table: Audio Playback


2.2.3. NAND Driver

2.2.3.1. AM335X-EVM

Buffer size (bytes)

am335x-evm: Write UBIFS Throughput (Mbytes/sec)

am335x-evm: Write UBIFS CPU Load (%)

am335x-evm: Read UBIFS Throughput (Mbytes/sec)

am335x-evm: Read UBIFS CPU Load (%)

102400

3.45 (min 3.41, max 3.56)

68.18 (min 67.31, max 68.65)

6.32

32.26

262144

3.51 (min 3.47, max 3.58)

65.30 (min 64.51, max 65.89)

6.30

29.16

524288

3.42 (min 3.39, max 3.44)

68.32 (min 67.94, max 69.29)

6.33

29.78

1048576

3.41 (min 3.38, max 3.42)

68.15 (min 67.20, max 68.90)

6.34

29.78

5242880

3.42 (min 3.39, max 3.44)

68.06 (min 66.88, max 69.18)

6.34

28.61

2.2.4. USB Driver

2.2.4.1. USB Device Controller

Number of Blocks

am335x-evm: Throughput (MB/sec)

150

18.00

Table: USBDEVICE HIGHSPEED SLAVE READ THROUGHPUT



Number of Blocks

am335x-evm: Throughput (MB/sec)

150

7.60

Table: USBDEVICE HIGHSPEED SLAVE WRITE THROUGHPUT