I have a non-Android Embedded system which uses Embedded Linux based OS. It has a Mali GPU and I need to profile GPU performance. Some options were DS5 Streamline and Mali Graphics debugger, but it seems tailored for Android (with adb). Is there any generic GPU profiling tool which I can use?
Streamline (DS version) does support Linux-based targets. On the Streamline documentation page you'll find near the bottom a link to the doc for setting up a Linux-based target.
Related
I want to start some Linux development for my research. Writing few simple scheduling algorithms and test them. I have few questions:
1) How do you develop for the linux kernel? IDE? How do you import the kernel files and see how they are related or connected?
2) Once you write your code, how do you simulate/debug it? I mean one can't just build the kernel for 20 - 30 minutes, make a new image and change boot.ini each time. This is a lenghty process plus you can't simulate or debug just observe if it works or not.
3) Is there A guide for starting developing in Linux. I find the lack of documentation surprising
I am developing for ARM-based boards
Excuse my ignorance.
Thanks
How do you develop for Linux kernel?
There are many components in the Linux kernel. Typically, kernel is divided into core and driver parts.
Core includes scheduling, MMU, memory management, process management etc
Drivers includes file system, networking, peripheral device drivers, USB etc
IDE is not a must to develop kernel code. For kernel veterans, VIM/nano is also OK. The development environment is up to you. If you are new to the kernel code, you want to build the function relationship views, some tools can be helpful:
Source Insight (Commercial)
vim + ctags (http://vim.wikia.com/wiki/Single_tags_file_for_a_source_tree)
How to debug it?
There are many Linux favors/distributions. You can use Software emulator or Hardware boards to debug the kernel. Android is based on Linux and there are many mobile phones or development boards that support Android. iOS is also derived from Linux and it is its own debug method.
Where to find the kernel documents?
For kernel part, there are many readme articles in kernel source tree. e.g. http://lxr.free-electrons.com/source/Documentation/debugging-via-ohci1394.txt
printk is powerful enough for newbies.
For ARM part, there are many articles in infocenter.arm.com
Debugging Linux kernels using DS-5
http://infocenter.arm.com/help/topic/com.arm.doc.den0024a/ch18s03s03.html?resultof=%22%6b%65%72%6e%65%6c%22%20
I found this link with drivers and runtimes.
And also as per this , OpenCL codebuilder is now part of INDE/Media server Studio and these are not free (Though they come with trial version)
So my ultimate question is how to get started with OpenCL with Intel HD Graphics?
There is a starter edition of the Intel INDE suite, which is free (and is not just a trial). This contains their core OpenCL SDK. The Media Server Studio may contain some additional tools to aid OpenCL development on Intel HD Graphics, but is certainly not required in order to develop or run OpenCL programs.
You don't actually need any OpenCL SDK to get started with OpenCL development. You need a driver/runtime in order to run OpenCL programs (these are freely available, as per your first link). To develop OpenCL programs you just need the headers and a library to link against, which are both also freely available (more info in this answer).
Which programming language would you like to use? In Java you can start with a simple Java SE application and integrate the JavaCL package. You do not need any further vendor specific drivers.
I've been experimenting with JavaFX, and I've found out that as of 2.2.0.b15, it uses GPU to provide improved font rendering. This makes a big difference for me. I intend to use it for visualization and animated UI features, so I'd like to know if GPU support will be provided in Linux.
I have come accross many statements that say that JavaFX will use my GPU if it is supported, and it does, but only under Windows. I have not seen anything about hardware acceleration for graphics under Linux, so I'm curious, can I expect to have this with 2.2 release?
I believe JavaFX for Linux already supports hardware acceleration for some features based on this statement from the JavaFX 2.1 Linux Develop Preview Release Notes:
3D features are supported for Nvidia cards (proprietary drivers only).
Not entirely conclusive, as it does not explicitly mention hardware acceleration, but I think use of hardware acceleration is likely if you have the appropriate Nvidia card and driver installed. Potentially, over time, hardware acceleration support under Linux for other hardware and driver configurations may be added.
I have a very simple Toshiba Laptop with i3 processor. Also, I do not have any expensive graphics card. In the display settings, I see Intel(HD) Graphics as display adapter. I am planning to learn some cuda programming. But, I am not sure, if I can do that on my laptop as it does not have any nvidia's cuda enabled GPU.
In fact, I doubt, if I even have a GPU o_o
So, I would appreciate if someone can tell me if I can do CUDA programming with the current configuration and if possible also let me know what does Intel(HD) Graphics mean?
At the present time, Intel graphics chips do not support CUDA. It is possible that, in the nearest future, these chips will support OpenCL (which is a standard that is very similar to CUDA), but this is not guaranteed and their current drivers do not support OpenCL either. (There is an Intel OpenCL SDK available, but, at the present time, it does not give you access to the GPU.)
Newest Intel processors (Sandy Bridge) have a GPU integrated into the CPU core. Your processor may be a previous-generation version, in which case "Intel(HD) graphics" is an independent chip.
Portland group have a commercial product called CUDA x86, it is hybrid compiler which creates CUDA C/ C++ code which can either run on GPU or use SIMD on CPU, this is done fully automated without any intervention for the developer. Hope this helps.
Link: http://www.pgroup.com/products/pgiworkstation.htm
If you're interested in learning a language which supports massive parallelism better go for OpenCL since you don't have an NVIDIA GPU. You can run OpenCL on Intel CPUs, but at best you can learn to program SIMDs.
Optimization on CPU and GPU are different. I really don't think you can use Intel card for GPGPU.
Intel HD Graphics is usually the on-CPU graphics chip in newer Core i3/i5/i7 processors.
As far as I know it doesn't support CUDA (which is a proprietary NVidia technology), but OpenCL is supported by NVidia, ATi and Intel.
in 2020 ZLUDA was created which provides CUDA API for Intel GPUs. It is not production ready yet though.
I would like to profile my embedded linux device for various actions (e.g. touch input on screen, flip a page in a news reading app etc). But the requirement is that the profiler should generate information for the entire software stack on the device both user space and kernel space. It would be best if the profiler does not include a significant overhead for logging purposes as memory and resources are very limited on the embedded linux device.
Any suggestions ?
Have you tried OProfile?
It supports several hardware architectures and I believe that it is included in the latest mainline kernels. It can also profile both the userspace and the kernel itself.