DS-5 Streamline Performance Analyzer Overview The growing need for high-performance and energy-efficient consumer products has changed the shape of embedded systems and consequently the requirements of software developers. The ARM(R) Development Studio 5 (DS-5TM) toolchain, through its ARM StreamlineTM Performance Analyzer component, delivers a simple and intuitive way to analyze and optimize complex Linux and AndroidTM based platforms. Key Features and Benefits * Locate ARM application processor hotspots at process, thread and source code level * Analyze the efficiency of parallel code on multicore platforms * Find and optimize bottlenecks across Cortex processors and ARM MaliTM GPUs * Zero-in on the top functions causing performance-penalty events (e.g. cache misses, and branch mispredictions) * Improve energy efficiency with actual power measurement data * Long term data capture over TCP/IP interface * Simultaneous analysis of multiple applications. Timeline Analysis Collected data is displayed on a timeline that enables viewing of software and system-level activity over time. This helps to identify performance issues that arise from the interaction of the software with the hardware blocks on the SoC. Key features: * Multicore-aware performance charts, process and thread activity, and instrumented annotations synchronized over time * Over 40 core-dependant PMU counters, over 300 Mali GPU counters, plus several OS performance metrics available for monitoring * Time-based filtering on software profile analysis * Thread activity mapping per core on SMP systems. ARM Mali GPU Support The Streamline Performance Analyzer gives you visibility of performance data across the Cortex processor and Mali GPU, so that bottlenecks on either side can be easily spotted. It enables developers to easily balance the overall computing load to make efficient usage of system resources. Flexible Architecture Easily customize the Streamline Performance Analyzer to collect and visualize SoC specific data. This flexibility can be used to monitor statistics from peripherals, fabric, and other processors interconnected with the Cortex processor. CPU, GPU, and OS-level performance graphs are time-correlated with process/thread activity per core and software profiling. The Architecture of the Digital World(R) Annotations Reconcile debug and performance analysis with a simple and powerful solution; code annotations. From simply tracking machine state changes on a timeline or cross-relating the screen content with performance issues, this simple code instrumentation links your software to the Streamline Performance Analyzer. The Call Paths view displays CPU time or PMU event count (e.g., cache misses), per process, thread, and function call tree. Software Profiling Profiling reports offer process-to-source-code drill-down analysis of the hotspots on your CPU. The three available views can be based on either processor time or PMU counters such as instructions, cycles, cache misses. * Call Paths - powerful hierarchical software profile view that can be used to see statistics per process, thread, library and functional call chain * Functions - flat function level list of hotspots * Code - the ultimate resource to pin-point hotspots within functions at both source code and disassembly levels. Applications and the kernel can write text messages or graphics frames into the Streamline driver for visualization. Energy Analysis The ARM Energy Probe is an easily deployable USB accessory that can sample voltage, current and power from up to three probe points in the system. It then synchronizes the data with system performance metrics and the software execution trace. Key Benefits: * Low Total Cost of Ownership - no additional debug hardware is required. All you need is the Streamline Performance Analyzer and the Energy Probe. * Everything in one place - make informed decisions based on an integrated view of the ARM application processor and Mali GPU performance counters, software profiling, code annotations and power consumption data. Power, current and voltage data can be displayed alongside thread activity and other performance metrics to enable energy optimization www.arm.com/optimize All brand names or product names are the property of their respective holders. Neither the whole nor any part of the information contained in, or the product described in, this document may be adapted or reproduced in any material form except with the prior written permission of the copyright holder. The product described in this document is subject to continuous developments and improvements. All particulars of the product and its use contained in this document are given in good faith. All warranties implied or expressed, including but not limited to implied warranties of satisfactory quality or fitness for purpose are excluded. This document is intended only to provide information to the reader about the product. To the extent permitted by local laws ARM shall not be liable for any loss or damage arising from the use of any information in this document or any error or omission in such information. Copyright (c) 2012 ARM Ltd. ARM Ltd. www.arm.com UK T: +44 1223 400400 FRANCE T: +33 1 39 30 47 89 JAPAN T: +81 45 477 5260 TAIWAN T: +886 2 2627 1681 CHINA T: +86 21 62351296 US T: +1 408 576 1500 GERMANY T: +49 89 456040-20 KOREA T: +82 31 712 8234 ISRAEL T: +972 9 7632000 INDIA T: +91 80 5138 4000 DS-5 Mali | 02.12