AMD CodeXL 1.3 GA Release Notes

Similar documents
AMD CodeXL Quick Start Guide

image1d_t, image1d_array_t, image2d_array_t Limited support. User cannot view contents of image types while kernel debugging.

OPENCL TM APPLICATION ANALYSIS AND OPTIMIZATION MADE EASY WITH AMD APP PROFILER AND KERNELANALYZER

AMD gdebugger 6.2 for Linux

CodeXL Quick Start Guide

Intel Parallel Studio XE 2017 Composer Edition BETA C++ - Debug Solutions Release Notes

GORDON SELLEY TONY HOSIER AMD GPU DEVELOPER TOOLS TEAM OPTIMIZING LINUX GAMES FOR AMD GRAPHICS USING GPU PERFSTUDIO2

Addressing the Increasing Challenges of Debugging on Accelerated HPC Systems. Ed Hinkel Senior Sales Engineer

OpenACC Course. Office Hour #2 Q&A

NSIGHT ECLIPSE EDITION

Ati Radeon 7000 Not Supported On This Hardware

NVIDIA Parallel Nsight. Jeff Kiel

More performance options

NSIGHT ECLIPSE EDITION

NSIGHT ECLIPSE EDITION

Qualcomm Snapdragon Profiler

Tesla GPU Computing A Revolution in High Performance Computing

CLU: Open Source API for OpenCL Prototyping

Debugging Your CUDA Applications With CUDA-GDB

April 4-7, 2016 Silicon Valley

Guest Operating System Installation Guide. February 25, 2008

ARM. Streamline. Performance Analyzer. Using ARM Streamline. Copyright 2010 ARM Limited. All rights reserved. ARM DUI 0482A (ID100210)

CodeWarrior Development Studio for StarCore 3850 DSP v SP1

Using Intel VTune Amplifier XE and Inspector XE in.net environment

Visual Studio 2008 Load Symbols Manually

Colligo Manager for Outlook Release Notes. Release Notes

Compute Shaders. Christian Hafner. Institute of Computer Graphics and Algorithms Vienna University of Technology

Release Notes Compute Abstraction Layer (CAL) Stream Computing SDK New Features. 2 Resolved Issues. 3 Known Issues. 3.

Getting Started with Intel SDK for OpenCL Applications

Introduction to Multicore Programming

Release Notes. May be insufficient for 1080p+ or 60 fps workflows 2. Additional hard disk space required for record to disk. 3

Parallel Programming and Debugging with CUDA C. Geoff Gerfin Sr. System Software Engineer

TotalView Release Notes

DS-5 ARM. Using Eclipse. Version Copyright ARM. All rights reserved. ARM DUI 0480L (ID100912)

Integrated Software Environment. Part 2

ControlUp v7.1 Release Notes

physics on screen Supported platforms and minimum system requirements

The Rise of Open Programming Frameworks. JC BARATAULT IWOCL May 2015

Enterprise Architect. User Guide Series. Profiling

Enterprise Architect. User Guide Series. Profiling. Author: Sparx Systems. Date: 10/05/2018. Version: 1.0 CREATED WITH

Android Sdk Tutorial For Windows 7 64 Bit Full Version

Guest Operating System Installation Guide. March 14, 2008

Graphics Performance Analyzer for Android

Visualization of OpenCL Application Execution on CPU-GPU Systems

Kaltura Video Package for Moodle 2.x Quick Start Guide. Version: 3.1 for Moodle

Contents. Platform Compatibility. New Features. Secure Remote Access SonicWALL SSL VPN 2.5 Early Field Trial (EFT) for SSL-VPN 200

Colligo Briefcase for Mac. Release Notes

NightStar LX Installation Guide. Version 4.1

Debugging in AVR32 Studio

CUDA Development Using NVIDIA Nsight, Eclipse Edition. David Goodwin

The Road to CCSv4. Status Update

QNX Software Development Platform 6.6. Quickstart Guide

NEW CEIBO DEBUGGER. Menus and Commands

Installation Guide and Release Notes

uintellisense brings all the power of IntelliSense code hints to your Unity code.

Ati opengl 1.4 windows 7 32bit. Ati opengl 1.4 windows 7 32bit.zip

ECE QNX Real-time Lab

1.00 Lecture 2. What s an IDE?

Mali Developer Resources. Kevin Ho ARM Taiwan FAE

Hands-on CUDA exercises

HETEROGENEOUS SYSTEM ARCHITECTURE: PLATFORM FOR THE FUTURE

Guest Operating System Installation Guide. May 28, 2008

April 4-7, 2016 Silicon Valley. CUDA DEBUGGING TOOLS IN CUDA8 Vyas Venkataraman, Kudbudeen Jalaludeen, April 6, 2016

Eclipse Debug Find Current Instruction Pointer

Microsoft Visual Studio 2008 Full Version For Windows 7 64 Bit

NightStar LX Installation Guide

Visual Profiler. User Guide

Heterogeneous Computing

HSA Foundation! Advanced Topics on Heterogeneous System Architectures. Politecnico di Milano! Seminar Room (Bld 20)! 15 December, 2017!

COBOL-IT Developer Studio Getting Started The Debugger Perspective Version 2.0

3D-Hub Player System Requirements V2.2

System Debug. This material exempt per Department of Commerce license exception TSU Xilinx, Inc. All Rights Reserved

Parallel Debugging. ª Objective. ª Contents. ª Learn the basics of debugging parallel programs

Programming and Simulating Fused Devices. Part 2 Multi2Sim

EW The Source Browser might fail to start data collection properly in large projects until the Source Browser window is opened manually.

3.1 Introduction. Computers perform operations concurrently

PROFILER USER'S GUIDE. DU _v7.5 September 2015

CSE 591: GPU Programming. Programmer Interface. Klaus Mueller. Computer Science Department Stony Brook University

GPU Computing Master Clss. Development Tools

Genesys Pulse. Known Issues and Recommendations

ExpertCAD 2012 Release Summary April 2012

Xcode Release Notes. Apple offers a number of resources where you can get Xcode development support:

Copyright

Debugging HPC Applications. David Lecomber CTO, Allinea Software

AMD ACCELERATING TECHNOLOGIES FOR EXASCALE COMPUTING FELLOW 3 OCTOBER 2016

Xcode Release Notes. Apple offers a number of resources where you can get Xcode development support:

NEW DEVELOPER TOOLS FEATURES IN CUDA 8.0. Sanjiv Satoor

AMD APP SDK v2.8.1 FAQ. 1 General Questions


Your code must have been compiled with the -g compiler option. Example:

AMD APU and Processor Comparisons. AMD Client Desktop Feb 2013 AMD

Laboratory Assignment #4 Debugging in Eclipse CDT 1

Introduction to Computation and Problem Solving

ECE 574 Cluster Computing Lecture 17

Windows 7 Read The Manually Installation Driver Automatically

Creating a robot project

CUDA. Schedule API. Language extensions. nvcc. Function type qualifiers (1) CUDA compiler to handle the standard C extensions.

Martin Dubois, ing. Contents

TDT 1.2 Release Notes and FAQ March 2002

Collecting Linux Trace without using CodeWarrior

Transcription:

AMD CodeXL 1.3 GA Release Notes Thank you for using CodeXL. We appreciate any feedback you have! Please use the CodeXL Forum to provide your feedback. You can also check out the Getting Started guide on the CodeXL Web Page and the latest CodeXL blog at AMD Developer Central - Blogs This version contains: For Windows for 32-bit and 64-bit Windows platforms o CodeXL Standalone application o CodeXL Microsoft Visual Studio 2010 extension o CodeXL Microsoft Visual Studio 2012 extension o CodeXL Remote Agent For 64-bit Linux platforms o CodeXL Standalone application o CodeXL Remote Agent Note about 32-bit Windows CodeXL 1.3 Upgrade Error On 32-bit Windows platforms, upgrading from CodeXL 1.0 using the CodeXL 1.3 installer will remove the previous version and then display an error message without installing CodeXL 1.3. The recommended method is to uninstall previous CodeXL versions before installing CodeXL 1.3. If you ran the 1.3 installer to upgrade a previous installation and encountered the error mentioned above, ignore the error and run the installer again to install CodeXL 1.3. Note about installing CodeAnalyst after installing CodeXL for Windows CodeXL can be safely installed on a Windows station where AMD CodeAnalyst is already installed. However, do not install CodeAnalyst on a Windows station already installed with CodeXL. Uninstall CodeXL first, and then install CodeAnalyst. System Requirements CodeXL contains a host of development features with varying system requirements: GPU Profiling and OpenCL Kernel Debugging o An AMD GPU (Radeon HD 5000 series or newer, desktop or mobile version) or APU is required.

o The AMD Catalyst Driver must be installed, release 13.11 Beta or later. Catalyst Beta 13.11 is the recommended version. See "Getting the latest Catalyst release" section below. For GPU API-Level Debugging, a working OpenCL/OpenGL configuration is required (AMD or other). CPU Profiling o Time-Based Profiling can be performed on any x86 or AMD64 (x86-64) CPU/APU. o The Event-Based Profiling (EBP) and Instruction-Based Sampling (IBS) session types require an AMD CPU or APU processor. Supported platforms: Windows platforms o Windows 7, 8 and 8.1, 32-bit and 64-bit o Visual Studio 2010/2012 must be installed on the station before the CodeXL Visual Studio 2010/2012 Package is installed, respectively. o If Visual Studio 2010 is not installed, the VC++ 2010 x86 Redistributable package must be installed for CPU Profiling Source View of.net applications to work. Linux platforms o Red Hat 6 u3 64-bit o Ubuntu 12.04 64-bit Getting the latest Catalyst release The way to get the latest beta driver is to use the links "Latest Windows Beta Driver" and "Latest Linux Beta Driver" on the Graphics Drivers support page: http://support.amd.com/us/gpudownload/pages/index.aspx New in this version The following items are new in this version: CodeXL includes a Remote Agent application for Remote GPU Profiling and Remote GPU Debugging. Developers can now run the CodeXL client application on their development station and launch their product application for GPU debugging or profiling on a separate target platform such as a server node. The AMD APP KernelAnalyzer application is now obsolete. Instead, CodeXL delivers a brand new user experience for Static OpenCL Kernel Analysis as part of the CodeXL application, encapsulated under a new work mode named Analyze Mode. This includes activation of the OpenCL offline compiler and detailed emulation of kernel performance resulting in analysis of multiple flow paths. The offline kernel performance analysis has been extended and supports target devices from the Southern Islands (Radeon HD 7000 series) and Sea Islands (Radeon HD 8000 series) families of GPUs. CPU Profiling of Cache Line Utilization. Developers can now use CodeXL to profile the utilization of the CPU's cache.

CPU Profiling support for Java inline functions. New design for the CPU Profiling Source Code /Disassembly View. GPU Debugging support for OpenGL 4.3. Support includes coverage of API functions and objects new in OpenGL releases up to and including 4.3. Support for the complete set of OpenCL 1.2 data types and API. Fixed Issues The following fixes were not part of the 1.2 release and are new to this version: Visual Studio crashes on launch after re-naming a GPU Profiling session and deleting it. (405809) While GPU Debugging, CodeXL points to incorrect source line on first step into kernel. (370960) - fixed in Catalyst driver 13.11 Beta CodeXL crashes during GPU Debugging if OpenCL-OpenGL interop VBO/CL buffer is clicked. (406102) - fixed in Catalyst driver 13.11 Beta CodeXL crashes during GPU Debugging when stepping into a kernel that contains a variable of structure type containing an unnamed union. (406284) Debugged application crashes when exiting. (363607) In GPU Debugging, stepping through some OpenCL kernels advances to non-sequential source lines. (396656, 404045, 387422) - fixed in Catalyst driver 13.8 Building an OpenCL kernel in Analyze mode (formerly KernelAnalyzer) - further builds are not allowed after the offline compiler reports "clbuildprogram had an unhandled exception". (399444) - fixed in Catalyst driver 13.11 Beta Building an OpenCL kernel analysis in Analyze mode (formerly KernelAnalyzer) won't show ISA tab/code for Devastator and Scrapper GPUs. (403765) - fixed in Catalyst driver 13.8. Building a kernel in Analyze mode (formerly KernelAnalyzer) doesn't generate ISA code for CPU target device. (398323) - fixed in Catalyst driver 13.11 Beta OpenCL kernel analysis in Analyze mode (formerly KernelAnalyzer) shows all zeros for Oland target devices. (399435) Building a kernel which uses double precision in Analyze mode (formerly KernelAnalyzer) causes "Frontend phase failed compilation" error. (401636) Building the APP SDK AtomicCounters kernel in Analyze mode (formerly KernelAnalyzer), fails to generate IL/ISA. (406669) Building the APP SDK BlackScholesDP kernel in Analyze mode (formerly KernelAnalyzer) fails. (406673) During GPU Debugging, Visual Studio may stop responding when kernel debugging while a window displaying an OpenGL object is open. (385320) During GPU Debugging, adding a large number of Breakpoints makes Visual Studio slow or sometimes unresponsive. (404178) CPU Profiling on Intel i7 Ivy Bridge may cause a system crash (BSoD). (405682) In CPU Profiling mode, importing a thread-based session generated from CodeAnalyst sometimes results in CodeXL crash. (356479) CodeXL crashes when performing CPU: Assess Performance profile of APP SDK sample LUDecomposition. (394149) For CPU Profiling sessions that include Call-Stack Sampling, CodeXL may hang during the postprocessing stage. (406095)

After performing CPU Profiling of OpenCL applications, post processing takes a very long time. (369505, 387530) On Linux, CPU Profiling Instruction-Based Sampling sessions of Java applications may attribute event samples to libjvm. (403250) When viewing a CPU Profiling session, choosing to display data per core / per node causes columns that were not selected for display to be shown. (403724) When viewing a CPU Profiling session, selecting a module and clicking "Display in Functions View" command, the functions view's Modules Filter should contain only the selected module. (404154) No CodeXL icon in Ubuntu dashboard panel & alt + tab list. (404193) Viewing CPU Profiling session data on Linux, Call Graph view shows Self/Deep samples but the functions and Overview tables are empty. (404458) In GPU Profiling, navigation to Source Code does not work for GL samples. (404851) CodeXL installer should install the Visual C++ redistributables if Visual Studio 2010 or later is not installed. (404870) GPU Profiler reports 0 for all performance counters of APP SDK QuasiRandomSequence on Cape Verde platforms (357699) When navigation to source code is enabled in GPU Profiling Application Trace profiles, navigation to source code would not work for later API calls in the session (404851) GPU Debugging on Linux, if gdb reports an error, CodeXL should realize the application has terminated. (405281) The GPU Debugger shows a type of unnamed struct for variables of the built-in type int4. (362152) On Linux with KDE, while GPU Debugging, the debugger displays static buffers, textures and VBOs in the explorer tree but won't show their content in the Object views. (406260) Building a kernel in Analyze mode (formerly KernelAnalyzer), the ISA window shows ALU packing and ALU clauses as "(0.0)" for all Northern Islands (Radeon HD 6000 series) GPU Cards. (344059) On Linux with KDE, CodeXL becomes unresponsive after opening User Guide/ Quick-Start Guide / Forum, when as the root user. (406604) CPU Profiling on Linux does not always contain call stack information - Call Graph view shows empty parents table and call graph tree. (388188) Known Issues Debugging OpenCL kernels that use read-modify-write atomic operations is not supported. GPU Debugging on OpenCL Static C++ Kernels is not supported. (345852) read_only and write_only Image variables are not shown in locals during GPU Debugging. (345852) OpenCL 1.2 keyword printf and barriers are not supported during kernel debugging. Building kernels with OpenCL 1.2 clcreateprogramwithbinaries and cllinkprogram API prevents the display of source code when debugging these kernels. (369171) Performing CPU Profiling with Call-Stack Sampling (CSS) enabled, on systems with discrete graphics card (Radeon HD 5000, 6000 or 7000 series) and Linux kernel version 3.0 or lower, may result in Linux kernel panic. This kernel panic does not occur with Linux kernel version 3.2 onwards. (352399) CPU Profiling on Linux platforms - Limitations of PERF

o CPU profiling uses PERF which requires kernel 2.6.32 or later. CPU Profiling with Call Stack Sampling requires Linux kernel 3.0 or later. However, we recommend using kernel 3.2 and above which has shown to be more stable. o Call chain analysis on Linux currently depends on the call chain information provided by Linux PERF. This requires the profiled binaries to have stack frame pointer. (i.e. compiled with -fno-omit-frame-pointer). o For non-root users to run CodeXL CPU profiling, "/proc/sys/kernel/perf_event_paranoid" needs to be set to "-1". o Instruction-Based Profiling on Linux requires Linux kernel 3.5 and above, and must be used with system-wide profiling. o Call chain information (stack trace) for inline functions is not available. PERF call chains which contain call stacks across modules have shown to be truncated. This results in inaccurate "Deep Samples", "Downstream Samples", and "Call Path" analysis. Performing Time-Based-Profiling on Intel CPUs with Call-Stack Collection enabled may display an error message that no data was collected. (423985) If gdebugger 6.x is installed on the machine, mouse click doesn't start text fields editing in CodeXL Visual Studio Extension. Workaround: Navigate to the text fields using TAB or uninstall gdebugger before installing CodeXL. (344811) Menu items are present but not visible after minimization and restore of CodeXL in Ubuntu system using Unity theme. Workaround: Use Unity 2D theme instead of Unity theme. (353082) AMDTTeapot sample may crash while debugging OpenCL kernels after multiple step operations (45 or more). (357741) For applications that perform simultaneous asynchronous data transfer and kernel execution, the timeline shown in the Application Trace session view will not show these operations being overlapped. This is because the driver and hardware force these operations to be synchronous while profiling. (333981) CPU Profiling on Windows 8 shows two target applications in Profile Overview. The conhost.exe process is an actual executable. This process fixes a fundamental problem in the way previous versions of Windows handled console windows, which broke drag & drop in Vista. If CodeXL is installed in path that includes non-ascii Unicode characters, profiling does not work (365118). GPU Profiler Performance Counters show wrong values for DirectX samples on Intel Platform. (378477) On Linux platforms, CodeXL crashes when an OpenGL application's window is closed during kernel debugging and a step operation is later performed (363957) Segmentation Fault while launching CodeXL in root mode on Llano (A-3000 series APU) + Radeon HD 6870. Workaround: Rename the file /usr/lib64/libamdocl64.so. (401418) On Catalyst driver 13.11 Beta and later, APP SDK sample AESEncryptDecrypt crashes when debugged. (420499) APP SDK sample Concurrent Kernel crashes when debugged. (423041) CodeXL crashes when building APP SDK sample Concurrent Kernel in Analyze mode. (420158) Exiting CodeXL while GPU profiling is in progress doesn't close the profiled application. (423059)

On Linux KDE platforms, when GPU profiling the TeaPot sample on a remote target, the Kernel Occupancy view may fail to open. (424024) The GPU Debugger can't step into a kernel if blocks that contain a return statement. (422699) In GPU Debugger's buffer view, the Hovered Value and Selected Value fields are active only for the lower left quarter of the image and displays the wrong value. (420968) The kernel execution time reported by the GPU Profiler may not be reliable when running on an APU with active power management. (413379) During remote GPU Debugging, kernel source breakpoints from previous debug sessions may appear as active, but will not be hit. Workaround: Before starting a remote GPU Debugging session, close all active kernel source windows and delete all existing kernel source breakpoints. (422715) Support Please use our CodeXL Forum for support requests and our development team will contact you.