2. Release Specific¶

2.1.1. Overview¶

The Processor Software Development Kit (Processor-SDK) for Linux provides a fundamental software platform for development, deployment and execution of Linux based applications and includes the following:

• Bootloaders & Filesystems
• SDK Installer
• Setup Scripts
• Makefiles
• Matrix Application Launcher
• Example Applications
• WLAN support (Wilink 8)
• Code Composer Studio v7

2.1.2. Licensing¶

Please refer to the software manifest, which outlines the licensing status for all packages included in this release. The manifest can be found on the SDK download page. The manifest can be found on the SDK download page or in the installed directory as indicated below. In addition, see GPLv3 Disclaimer

2.1.3. Documentation¶

• Processor SDK Linux Software Developer’s Guide: Provides information on features, functions, delivery package and, compile tools for the Processor SDK Linux release. This also provides detailed information regarding software elements and software infrastructure to allow developers to start creating applications.
• Getting Started Guide: Provides information on getting the software and running examples/demonstrations bundled in the SDK.
• Software Manifest: Provides license information on software included in the SDK release. This document is in the release at [INSTALL-DIR]/docs.
• EVM Quick Start Guide: Provides information on hardware setup and running the demonstration application that is loaded on flash. This document is provided as part of the EVM kit.

2.1.4. Release 05.00.00 Highlights¶

Released Jul 2018

What’s New

• 2018 LTS Update: First release in 2018 LTS (Kernel 4.14, U-boot: 2018.01, gcc: 7.2-2017.11, Yocto 2.4 (rocko). For more details, see the migration guide
• AM57x: Enable Deep Learning at the edge using TI-DL that leverages TI’s proprietary, highly optimized CNN/DNN implementation on the EVE and C66x DSP compute engines. Please see here for more details
• AM57x: PTP: Support Ordinary Clock (OC) Master/Slave with HSR redundancy on PRU-ICSS. Please refer to HSR OC TC for more details
• AM57x: PTP: Support Hybrid Clock (HC), i.e. Transparent Clock (TC) + Ordinary Clock (OC) in HSR mode. Please refer to HSR OC TC for more details
• AM57x: HSR/PRP: Add Broadcast Storm Prevention
• AM57x: HSR/PRP: Improve throughput at smaller packet sizes, details here
• AM437x, AM57x: Implement Simple Open Real-Time Ethernet (SORTE) with Linux. Please see here for more details

SDK Components & Versions

2.1.5. Supported Platforms¶

See Supported Platforms and Versions for a list of supported platforms and links to more information.

2.1.6. Build Information¶

2.1.7. Issues Tracker¶

2.2.1. Kernel Performance Guide¶

Read This First

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

Table: Evaluation Modules

About This Manual

This document provides performance data for each of the device drivers which are part of the Process SDK Linux package. This document should be used in conjunction with release notes and user guides provided with the Process 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://community.ti.com/ or http://support.ti.com/

Linux Kernel

System Benchmarks

LMBench

Table: LM Bench Metrics

Dhrystone

Table: Dhrystone Benchmark

Whestone

Table: Whestone Benchmark

Linpack

Table: Linpack Benchmark

NBench

Table: NBench Benchmarks

Stream

Table: Stream

Interrupt Latency Measurement

The following table gives interrupt latency measurements taken across 1001 measurement samples.

Table: Interrupt Latency Measurement

Boot-time Measurement

Boot media: MMCSD

Table: ‘

ALSA SoC Audio Driver

1. Access type - RW_INTERLEAVED
2. Channels - 2
3. Format - S16_LE
4. Period size - 64

Table: Audio Capture

Table: Audio Playback

Sensor Capture

Capture video frames (MMAP buffers) with v4l2c-ctl and record the reported fps

Table: Sensor Capture

Display Driver

Table: Display performance (LCD)

Table: Display performance (HDMI)

Graphics SGX Driver

Run GLBenchmark and capture performance reported Display rate (Fps), Fill rate, Vertex Throughput, etc. All display outputs (HDMI and LCD) are connected when running these tests

Performance (Fps)

Table: GLBenchmark 2.5 Performance

Vertex Throughput

Table: GLBenchmark 2.5 Vertex Throughput

Pixel Throughput

Table: GLBenchmark 2.5 Pixel Throughput

Multimedia (Decode)

Run gstreamer pipeline “gst-launch-1.0 playbin uri=file://<Path to stream> video-sink=”kmssink sync=false connector=<connector id>” audio-sink=fakesink” and calculate performance based on the execution time reported. All display display outputs (HDMI and LCD) were connected when running these tests, but playout was forced to LCD via the connector=<connector id> option.

H264

Table: Gstreamer H264 in AVI Container Decode Performance

MPEG4

Table: GStreamer MPEG4 in 3GP Container Decode Performance

MPEG2

Table: GStreamer MPEG2 in MP4 Container Decode Performance

Ethernet Driver

PCIe Driver

PCIe-ETH

Table: ‘

NAND Driver

AM335X-EVM

Table: ‘

AM43XX-GPEVM

Table: ‘

OMAPL138-LCDK

Table: ‘

QSPI Flash Driver

K2G-EVM

Table: ‘

SPI Flash Driver

K2G-EVM

Table: ‘

EMMC Driver

AM43XX-GPEVM

Table: ‘

Table: ‘

Table: ‘

AM57XX-EVM

Table: ‘

Table: ‘

Table: ‘

K2G-EVM

Table: ‘

Table: ‘

Table: ‘

SATA Driver

AM57XX-EVM

SATA

Table: ‘

Table: ‘

Table: ‘

• Filesize used is : 1G
• SATA II Harddisk used is: Seagate ST3500514NS 500G

mSATA

Table: ‘

Table: ‘

Table: ‘

• Filesize used is : 1G
• MSATA Harddisk used is: SMS200S3/30G Kingston mSATA SSD drive

OMAPL138-LCDK

Table: ‘

Table: ‘

Table: ‘

MMC/SD Driver

AM335X-EVM

Table: ‘

Table: ‘

Table: ‘

The performance numbers were captured using the following:

• SanDisk 8GB MicroSDHC Class 10 Memory Card
• Partition was mounted with async option

AM43XX-GPEVM

Table: ‘

Table: ‘

The performance numbers were captured using the following:

• Sandisk Ultra 8GB Micro SDHC Flash Card Class10
• Partition was mounted with async option

AM57XX-EVM

Table: ‘

Table: ‘

Table: ‘

The performance numbers were captured using the following:

• Sandisk Ultra 8GB Micro SDHC Flash Card Class10
• Partition was mounted with async option

K2G-EVM

Table: ‘

Table: ‘

Table: ‘

The performance numbers were captured using the following:

• Sandisk Ultra 8GB Micro SDHC Flash Card Class10
• Partition was mounted with async option

OMAPL138-LCDK

Table: ‘

Table: ‘

Table: ‘

• Sandisk Ultra 8GB Micro SDHC Flash Card Class10
• Partition was mounted with async option

UART Driver

Performance and Benchmarks not available in this release.

I2C Driver

Performance and Benchmarks not available in this release.

EDMA Driver

Performance and Benchmarks not available in this release.

Touchscreen Driver

Performance and Benchmarks not available in this release.

USB Driver

MUSB/XHCI Host controller

Setup : Western Digital 1TB (My Book Essential) USB hard disk is connected to usb0 port. File read/write performance data on usb0 port is captured.

Table: ‘

Table: ‘

USB Mass Storage Class Slave Driver

# dd if=/dev/zero of=/dev/shm/50M bs=1M count=50"

Insert the USB mass storage gadget with /dev/shm/50M as the backing storage:

# insmod g_mass_storage.ko file=/dev/shm/50M

For test setup details refer to USB MSC device setup

AM335X USB SLAVE DEVICE PERFORMANCE

Table: AM335X USB Slave Read Performance values

Table: AM335X USB Slave Write Performance values

AM43XX GPEVM USB SLAVE DEVICE PERFORMANCE

Table: AM43XX USB Slave Read Performance values

Table: AM43XX USB Slave WritePerformance values

AM57XX GPEVM USB SLAVE DEVICE PERFORMANCE

Table: AM57XX USB Slave Read Performance values

Table: AM57XX USB Slave WritePerformance values

OMAPL138-LCDK USB SLAVE DEVICE PERFORMANCE

Table: OMAPL138-LCDK USB Slave Read Performance values

Table: OMAPL138-LCDK USB Slave WritePerformance values

USB CDC/RNDIS Slave Driver

Performance benchmarks were collected using the Iperf tool and default options were used to collect the throughput numbers.

AM335X USB CDC Performance

Table: AM335X USB CDC Performance values - Client

AM43XX GPEVM USB CDC Performance

Table: AM43XX GPEVM USB CDC Performance values - Client

AM57XX USB CDC Performance

Table: AM57XX USB CDC Performance values - Client

K2G-EVM USB CDC Performance

Table: K2G-EVM USB CDC Performance values - Client

OMAPL138-LCDK USB CDC Performance

Table: OMAPL138-LCDK USB CDC Performance values - Client

AM335X USB RNDIS Performance

Table: AM335XX USB RNDIS Performance values - Client

AM43XX GPEVM USB RNDIS Performance

Table: AM43XX GPEVM USB RNDIS Performance values - Client

AM57XX USB RNDIS Performance

Table: AM57XX USB RNDIS Performance values - Client

OMAPL138-LCDK GPEVM USB RNDIS Performance

Table: OMAPL138-LCDK USB RNDIS Performance values - Client

CRYPTO Driver

OpenSSL Performance

Table: ‘

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

IPSec Performance

Note: queue_len is set to 300 and software fallback threshold set to 9 to enable software support for optimal performance

Table: ‘

DCAN Driver

Performance and Benchmarks not available in this release.

2.2.2. RT Kernel Performance Guide¶

Read This First

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

Table: Evaluation Modules

About This Manual

This document provides performance data for each of the device drivers which are part of the LCPD Core SDK package. This document should be used in conjunction with release notes and user guides provided with the LCPD Core SDK 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://community.ti.com/ or http://support.ti.com/

Linux Kernel

System Benchmarks

Dhrystone

Table: Dhrystone Benchmark

Whestone

Table: Whestone Benchmark

Linpack

Table: Linpack Benchmark

NBench

Table: NBench Benchmarks

Stream

Table: Stream Benchmarks

Maximum Latency under different use cases

Shield (dedicated core) Case

shield_shell()
{
create_cgroup nonrt 0
create_cgroup rt 1
for pid in $(cat /sys/fs/cgroup/tasks); do /bin/echo $pid > /sys/fs/cgroup/nonrt/tasks; done
/bin/echo $$ > /sys/fs/cgroup/rt/tasks
}

Table: Max Latency for shielded use cases

OSADL Test

Latency test used by https://www.osadl.org/ Command: “cyclictest -l100000000 -m -n -a0 -t1 -p99 -i200 -h400 -q”

am574x-idk

# Total: 100000000
# Min Latencies: 00007
# Avg Latencies: 00011
# Max Latencies: 00038
# Histogram Overflows: 00000
# Histogram Overflow at cycle number:

# Histogram
000000 000000
000001 000000
000002 000000
000003 000000
000004 000000
000005 000000
000006 000000
000007 000006
000008 006778
000009 5852853
000010 2587813
000011 53313381
000012 33663230
000013 2055609
000014 515347
000015 229216
000016 1376865
000017 170278
000018 042501
000019 022922
000020 007325
000021 055718
000022 065202
000023 015769
000024 011486
000025 000835
000026 000640
000027 000700
000028 000490
000029 001261
000030 001190
000031 000909
000032 000830
000033 000525
000034 000214
000035 000074
000036 000019
000037 000010
000038 000004
000039 000000
000040 000000

Boot-time Measurement

Sensor Capture

Capture video frames (MMAP buffers) with v4l2c-ctl and record the reported fps

Table: Sensor Capture

Display Driver

Set different display modes using modetest’s sync plus flip option (modetest -v -s ...)

Table: Display performance (LCD)