A single view of performance for your entire system

Interrogates all your system's hardware counters to reveal where all the time has been spent in one easy to interpret visualization. Alongside hardware counters, Streamline analyzes other key performance metrics including Open CL applications and power usage from the Arm Energy Probe or National Instruments DAC.

Try for free   Buy Development Studio

 ×

CPU-based sampler

Samples the program counter to show you where in the system your CPU is spending its time, from system level right down to line by line application source code

 ×

Focuses your analysis

Filters out information that isn't relevant to your analysis, to focus on a certain bottleneck, or inspect the performance of a particular application or thread.

 ×

Fully automatable

Captures data from the command line, to integrate into your existing build system and enable problem fixes as soon as they present in development phases.

 ×

Investigate OpenCL

Use Streamline in Open CL Timeline Mode to investigate Open CL applications. Visualise which kernel is running on the GPU, for how long and any dependencies which will cause kernels to stall.


Use Cases

Customize to add your own IP

Provides the source code to the collection agent, enabling the addition of hardware performance counters for your own system IP and visualizations for non-Arm systems.

Plot power usage

Streamline plots power information from the Arm Energy Probe or National Instruments DAC against other hardware counters and PC sampling data, to determine the cause of power surges.

Advanced investigation

Using contextual annotations against your source code in Streamline enriches your understanding of cause and effect and can relate specific parts of your code with key points of graphical analysis.


Bare-metal system support

Streamline supports bare-metal systems and can be used with various RTOSes for performance analysis across everything from the smallest Cortex-M to the biggest Cortex-A based systems. Streamline Bare-metal works by compiling generated agent code that is unique to the system. This code collects relevant performance data from the system for transporting and importing into Streamline. Once complete, users can generate analysis and visualizations in the same way as they would with Linux.

Fix performance bottlenecks in your bare-metal system

Streamline enables the analysis of your entire system for identifying bottlenecks, using hardware counters and a range of other performance metrics. Streamline for Bare-metal systems offers the same level of performance analysis but specifically for Cortex-R and Cortex-M based devices where a Linux based operating system is not present.

Instruction trace

Streamline for Bare-metal systems imports Instruction Trace directly from your system, showing you exactly where your application has been, and populating all the standard Streamline views. This feature shows you accurate call paths and exactly where the CPU has spent its time. Streamline Bare-metal works by compiling generated agent code that is unique to the system. This code collects performance data for transporting from the system and into Streamline.

For multiple trace mechanisms

We understand that there's a wealth of different systems, each with different trace capabilities, so Streamline for Bare-metal systems enables data extraction from the production device via STM ITM and ETM. Where there is no trace mechanism, the tool places the data in memory, allowing users to decide how best to transport it to the PC.

Resources

Video

Profiling Intel SoC FPGAs with Arm Streamline

Learn how to use Streamline to perform performance analysis of applications running on Intel SoC FPGA devices.

Watch on YouTube


Streamline Community Blogs

Get in touch

If you would like to get in contact with us about Development Studio, please do so through our online enquiry form.