There is a device driver for a camera device provided to us as a .so library file by the vendor.
Only the header file with API's is available which provides the list of functions that we can work with the device. Our application is linked with the .so library file provided by the vendor and uses the interface functions provided for our objective.
When we wanted to measure the time taken by our application in handling different tasks, we have added GCC -pg flag and compiled+built our application.
But we found that using this executable built with -pg, we are observing random failure in the camera image acquire functions. Since we are using the .so library file, we do not know what is going wrong inside that function.
So in general I wanted to understand what could be the possible reasons of such a failure mode. Any pointers or documents that can help what goes inside profiling and its side effects is appreciated.
This answer is a helpful overview of how the gcc -pg flag profiler actually works. The take-home point is mostly to do with possible changes to timing. If your library has any kind of time-sensitivity in it, introducing profiler overheads might be changing the time it takes to execute parts of the code, and perhaps violating some kind of constraint.
If you look at the gprof documentation, it would explain the implementation details:
Profiling works by changing how every function in your program is
compiled so that when it is called, it will stash away some
information about where it was called from. From this, the profiler
can figure out what function called it, and can count how many times
it was called. This change is made by the compiler when your program
is compiled with the `-pg' option, which causes every function to call
mcount (or _mcount, or __mcount, depending on the OS and compiler) as
one of its first operations.
So the timing of your application would change quite a bit when you turn on -pg.
If you would like to instrument your code without significantly affecting the timings, you could possibly look at oprofile. It does not pose as significant an overhead as gprof does.
Another fairly recent tool that serves as a good lightweight profiling tool is perf.
The profiling tools are useful primarily in understanding the CPU bound pieces of your library/application and can help you optimize those critical pieces. Most of the time they serve to identify some culprit function/method which wastes CPU cycles. So do not use it as the sole piece for debugging any and all issues.
Most vendor libraries would also provide means to turn on extra debugging or dumping extra information during runtime. They include means such as environment variables, log files, /proc or /sys interfaces for drivers, etc. and sometimes even tools to increase debugging levels at runtime. See if you can leverage these.
If you have defined APIs in a library/driver, you should run unit-tests on them instead of trying to debug the whole application you've built.
If you find a certain unit-test fails, send the source code of the unit-test to your vendor, and ask them to fix the bug. If it is not a bug, your vendor would at least point you towards the right set of APIs or the semantics to use.
Related
We are trying to optimize HPC applications using OpenMP on a new hardware platform. These applications need precise placement/pinning of their cores or performance falls in half. Currently, we provide the user a custom GOMP_CPU_AFFINITY map for each platform, but this is cumbersome, because it's different on each hardware version, and even platforms with different firmware versions sometimes change their CoreID physical mappings - all things impossible for the user to detect on the fly.
It would be a great help if HPC applications could simply set GOMP_PROC_BIND to "close" and OpenMP would do the right thing for the given platform - but to make this possible, the hardware vendor would need to define what "close" means for each machine. We'd like to do this, but we can't tell how/where OpenMP gets CoreID lists to use for things like close, spread, etc. (For various external requirements, the CoreID spatial pattern on this machine would appear utterly random to a software writer.)
Any advice as to where/how OpenMP defines the CoreID lists for OMP_PROC_BIND so we could configure them? We are comfortable with the idea that we might need a custom version of OpenMP (with altered source code) for this platform if needed.
Thanks, everyone. :)
Jeff
Expanding on what #VictorEijkhout said...
You seem have invented an envirable that I can't find anywhere with Google (GOMP_PROC_BIND), with the OpenMP standard envirable (OMP_PROC_BIND). If GOMP_PROC_BIND exists the name suggests that it is a GNU feature. Note too that one of the two Google hits for GOMP_PROC_BIND says "Code that reads the setting is buggy. Setting is invalid and ignored at runtime." So, if you are setting that it is unsurprising that it has no effect!
I will therefore answer for the more general case of OMP_PROC_BIND.
The binding of OpenMP threads to logicalCPUs clearly has to be done at runtime, since, beyond its ISA, the compiler has no knowledge of the hardware on which the compiled code will run. Therefore you need to be looking at the runtime library code.
I have not looked at GNU's libgomp, but, where it can, LLVM's libomp uses the hwloc library to explore the machine hardware. Since hwloc also includes other useful tools for machine exploration (such as lstopo) it is likely that your effort is best invested in ensuring good hwloc support on your machine, at which point there will be no need to delve inside the OpenMP runtime.
In a tutorial I've encountered a new concept (for me), that I never thought is possible. Actually, I thought that compilation is an entirely pre-run-time process. This is the phrase from tutorial: "Compile time occurs before link time (when the output of one or more compiled files are joined together) and runtime (when a program is executed). In some programming languages it may be necessary for some compilation and linking to occur at runtime".
My questions are:
Is pre-run-time compilation and linking processes absolutely different from run-time compilation and linking? If yes, please explain the main differences.
How are code sections that need to be compiled(linked) during run-time marked and where is that information kept? (This may be different from language to language, if possible, please give a specific example).
Thank you very much for your time!
Runtime compilation
The best (most well known) example I'm personally aware of is the just in time compilation used by Java. As you might know Java code is being compiled into bytecode which can be interpreted by the Java Virtual Machine. It's therefore different from let's say C++ which is first fully (preprocessed) compiled (and linked) into an executable which can be ran directly by the OS without any virtual machine.
The Java bytecode is instead interpreted by the VM, which maps them to processor specific instructions. That being said the JVM does JIT, which takes that bytecode and compiles it (during runtime) into machine code. Here we arrive at your second question. Even in Java it can depend on which JVM you are using but basically there are pieces of code called hotspots, the pieces of code that are run frequently and which might be compiled so the application's performance improves. This is done during runtime because the normal compiler does not have (or well might not have) all the necessary data to make a proper judgement which pieces of code are in fact ran frequently. Therefore JIT requires some kind of runtime statistics gathering, which is done parallel to the program execution and is done by the JVM. What kind of statistics are gathered, what can be optimised (compiled in runtime) etc. depends on the implementation (you obviously cannot do everything a normal compiler would do due to memory and time constraints - guess this partly answers the first question? you don't compile everything and usually only a limited set of optimisations are supported in runtime compilation). You can try looking for such info but from my experience usually it's very badly documented and hard to find (at least when it comes to official sources, not presentations/blogs etc.)
Runtime linking
Linker is a different pair of shoes. We cannot use the Java example anymore since it doesn't really have a linker like C or C++ (instead it has a classloader which takes care of loading files and putting it all together).
Usually linking is performed by a linker after the compilation step (static linking), this has pros (no dependencies) and cons (higher memory imprint as we cannot use a shared library, when the library number changes you need to recompile your sources).
Runtime linking (dynamic/late linking) is actually performed by the OS and it's the OS linker's job to first load shared libraries and then attach them to a running process. Furthermore there are also different types of dynamic linking: explicit and implicit. This has the benefit of not having to recompile the source when the version number changes since it's dynamic and library sharing but also drawbacks, what if you have different programs that use the same library but require different versions (look for DLL hell). So yes those two concepts are also quite different.
Again how it's all done, how it's decided what and how should be linked, is OS specific, for instance Microsoft has the dynamic-link library concept.
I am developing an algorithm that uses ARM Neon instructions. I am writing the code using assembler file (.S and no inline asm).
My question is that what is the best way for debugging purpose i.e. viewing registers, memory, etc.
Currently, I am using Android NDK to compile and my Android phone to run the algorithm.
Poor man's debug solutions...
You can use gdb / gdbserver to remotely control execution of applications on an Android phone. I'm not giving full details here because they change all the time but for example you can start with this answer or make a quick search on Internet. Learning to use GDB might seem to have a high steep curve however material on web is exhaustive. You can easily find something to your taste.
Single-stepping an ARM core via software tools is hard that's why ARM ecosystem is full of expensive tools and extra HW equipment.
Trick I use is to insert BRK instructions manually in assembly code. BRK is Self-hosted debug breakpoint. When core sees this instruction it stops and informs OS about situation. OS then notifies debugger about the situation and passes control to it. When debugger gets control you can check contents of registers and probably even make changes to them. Last part of the operation is to make your process continue. Since PC is still at our break point instruction what you must do is to increase PC, set it to instruction after BRK.
Since you mentioned you use .S files instead of .s files you can utilize gcc to do preprocessing / macro work. This way enabling, disabling BRK might become less of an issue.
Big down side of this way of working is turnaround time. If there is a certain point that you want to investigate with gdb you must make sure there is a BRK instruction there and this will probably require another build/push/debug cycle.
I want to install a driver for Ros (robot operating system), and I have two options the binary install and the compile and install from source. I would like to know which installation is better, and what are the advantages and disadvantages of each one.
Source: AKA sourcecode, usually in some sort of tarball or zip file. This is RAW programming language code. You need some sort of compiler (javac for java, gcc for c++, etc.) to create the executable that your computer then runs.
Advantages:
You can see what the source code is which means....
You can edit the end result program to behave differently
Depending on what you're doing, when you compile, you could enable certain optimizations that will work on your machine and ONLY your machine (or one EXACTLY like it). For instance, for some sort of gfx rendering software, you could compile it to enable GPU support, which would increase the rendering speed.
You can create a version of an application for a different OS/Chipset (see Binary below)
Disadvantages:
You have to have your compiler installed
You need to manually install all required libraries, which frequently also need to be compiled (and THEIR libraries need to be installed, etc.) This can easily turn a quick 30-second command into a multi-hour project.
There are any number of things that could go wrong, and if you're not familiar with what the various errors mean, finding support online could be quite difficult.
Binary: This is the actual program that runs. This is the executable that gets created when you compile from source. They typically have all necessary libraries built into them, or install/deploy them as necessary (depending on how the application was written).
Advantages:
It's ready-to-run. If you have a binary designed for your processor and operating system, then chances are you can run the program and everything will work the first time.
Less configuration. You don't have to set up a whole bunch of configuration options to use the program; it just uses a generic default configuration.
If something goes wrong, it should be a little easier to find help online, since the binary is pre-compiled....other people may be using it, which means you are using the EXACT same program as them, not one optimized for your system.
Disadvantages:
You can't see/edit the source code, so you can't get optimizations, or tweak it for your specific application. Additionally, you don't really know what the program is going to do, so there could be nasty surprises waiting for you (this is why Antivirus is useful....although LESS necessary on a linux system).
Your system must be compatible with the Binary. For instance, you can't run a 64-bit application on a 32-bit operating system. You can't run an Intel binary for OS X on an older PowerPC-based G5 Mac.
In summary, which one is "better" is up to you. Only you can decide which one will be necessary for whatever it is you're trying to do. In most cases, using the binary is going to be just fine, and give you the least trouble. Sometimes, though, it is nice to have the source available, if only as documentation.
Since the dynamically linked libraries have to be resolved at run-time, are statically linked executables faster than dynamically linked executables?
Static linking produces a larger executable file than dynamic linking because it has to compile all of the library code directly into the executable. The benefit is a reduction in overhead from no longer having to call functions from a library, and anywhere from somewhat to noticeably faster load times.
A dynamically linked executable will be smaller because it places calls at runtime to shared code libraries. There are several benefits to this, but the ones important from a speed or optimization perspective are the reduction in the amount of disk space and memory consumed, and improved multitasking because of reduced total resource consumption (particularly in Windows).
So it's a tradeoff: there are arguments to be made why either one might be marginally faster. It would depend on a lot of different things, such as to what extent speed-critical routines in the program relied on calls to library functions. But the important point to emphasize in the above statement is that it might be marginally faster. The speed difference will be nearly imperceptible, and difficult to distinguish even from normal, expected fluctuations.
If you really care, benchmark it and see. But I advise this is a waste of time, and that there are more effective and more important ways to increase your application's speed. You will be much better off in the long run considering factors other than speed when making the "to dynamically link or to statically link" decision. For example, static linking may be worth considering if you need to make your application easier to deploy, particularly to diverse user environments. Or, dynamic linking may be a better option (particularly if those shared libraries are not your own) because your application will automatically reap the benefits of upgrades made to any of the shared libraries that it calls without having to lift a finger.
EDIT: Microsoft makes the specific recommendation that you prefer dynamic linking over static linking:
It is not recommended to redistribute
C/C++ applications that statically
link to Visual C++ libraries. It is
often mistakenly assumed that by
statically linking your program to
Visual C++ libraries it is possible to
significantly improve the performance
of an application. However the impact
on performance of dynamically loading
Visual C++ libraries is insignificant
in almost all cases. Furthermore,
static linking does not allow for
servicing the application and its
dependent libraries by either the
application's author or Microsoft. For
example, consider an application that
is statically linked to a particular
library, running on a client computer
with a new version of this library.
The application still uses code from
the previous version of this library,
and does not benefit from library
improvements, such as security
enhancements. Authors of C/C++
applications are strongly advised to
think through the servicing scenario
before deciding to statically link to
dependent libraries, and use dynamic
linking whenever possible.
It depends on the state of your disk and whether or not the DLLs might be used in other processes. A cold start happens when your program and its DLLs were never loaded before. An EXE without DLLs has a faster cold start since only one file needs to be found. You would have to have a badly fragmented disk that's almost full to not have this case.
A DLL can start to pay off when it is already loaded in another process. Now the code pages of the DLL are simply shared, startup overhead is very low and memory usage is efficient.
A somewhat similar case is a warm start, a startup where the DLL is still available in the file system cache from a previous time it was used. The difference between a cold and a warm start can be quite significant on a sluggish disk. The one reason that everybody likes a SSD.
No, I don't think so. in most of the cases only a copy of the library in memory per program makes the overall system less memory. suppose you have 100 programs using the libc library statically, and libc is ~2-3MB, so it makes the size of the program increase.
But same in a dynamic we can share stuff, so fewer bytes in the memory means more bytes in Caches, More bytes in cache means faster.
Though it has loading overhead, your overall system performance is faster.