How to benchmark Raku? - benchmarking

I regularly follow the GitHub Rakudo repository to see what changes are going on in the Rakudo compiler.
I sometimes see commits where the individual functions are sped up by certain percentage, and times like the image below.
What is the workflow to evaluate this? I am keen to learn this so one can know how your functions are performing and accordingly optimize further and can contribute to Rakudo development.
I read for help here, here. I googled but could not find this information. I also learnt about MoarVM profiler via command line --profile option here which generates html output. What is it to look for?
I am not a formal computer science guy. I understand famous quote of Tony Hoare — “Premature optimization is the root of all evil” but with passage of time, once the code is written correct, one wishes to optimize it. So the question.

I use --profile to get a better idea of where the bottlenecks are. The produced profile is a good start, but not very good for CPU usage when the differences become very small. It is however pretty good at tracking allocations of objects, and fewer object allocations at least can mean less memory churn (not always though, if the object are very short-lived). And the keeping track of stuff with --profile has its effects on optimizations as well, so Heisenberg's uncertainty principle definitely applies here.
Once I have a piece of code of before / after, I run it either as a script or as a one liner with time. I have a bunch of handy aliases that help me with that:
alias r='time raku -e'
alias rp='raku --profile -e'
The reason I do it as separate processes with at least a few seconds inbetween, is that:
running multiple benchmarks in the process tend to heat up the CPU, which will then get downthrottled, making the later benchmark worse.
if both benchmarks share some code in the core, the later benchmark may benefit from that code having been inlined / JITted by the earlier benchmark.
I then run each of the before and after code 3 to 5 times, and a Nil loop to find out the overhead. So e.g.:
$ r 'my $a = "42"; Int($a) for ^100000'
real 0m0.244s
$ r 'my $a = "42"; $a.Int for ^100000'
real 0m0.178s
$ r 'my $a = "42"; Nil for ^100000'
real 0m0.154s
And then calculate the difference:
$ r 'say (244 - 154) / (178 - 154)'
So it's about 3.75x as fast to use $a.Int than Int($a). Which of course could start another --profile cycle finding out why Int($a) is so much slower. Also, when I see differences in speed I cannot explain, I use a --profile to find out if it's really doing the things I think it's doing. Specifically unexpected constant-folding can sometimes make you think you found the optimal optimization, when in fact you reduced your code to doing basically nothing.


Profiling a Single Function Predictably

I need a better way of profiling numerical code. Assume that I'm using GCC in Cygwin on 64 bit x86 and that I'm not going to purchase a commercial tool.
The situation is this. I have a single function running in one thread. There are no code dependencies or I/O beyond memory accesses, with the possible exception of some math libraries linked in. But for the most part, it's all table look-ups, index calculations, and numerical processing. I've cache aligned all arrays on the heap and stack. Due to the complexity of the algorithm(s), loop unrolling, and long macros, the assembly listing can become quite lengthy -- thousands of instructions.
I have been resorting to using either, the tic/toc timer in Matlab, the time utility in the bash shell, or using the time stamp counter (rdtsc) directly around the function. The problem is this: the variance (which might be as much as 20% of the runtime) of the timing is larger than the size of the improvements I'm making, so I have no way of knowing if the code is better or worse after a change. You might think then it's time to give up. But I would disagree. If you are persistent, many incremental improvements can lead to a two or three times performance increase.
One problem I have had multiple times that is particularly maddening is that I make a change and the performance seems to improve consistently by say 20%. The next day, the gain is lost. Now it's possible I made what I thought was an innocuous change to the code and then completely forgot about it. But I'm wondering if it's possible something else is going on. Like maybe GCC doesn't yield a 100% deterministic output as I believe it does. Or maybe it's something simpler, like the OS moved my process to a busier core.
I have considered the following, but I don't know if any of these ideas are feasible or make any sense. If yes, I would like explicit instructions on how to implement a solution. The goal is to minimize the variance of the runtime so I can meaningfully compare different versions of optimized code.
Dedicate a core of my processor to run only my routine.
Direct control over the cache(s) (load it up or clear it out).
Ensuring my dll or executable always loads to the same place in memory. My thinking here is that maybe the set-associativity of the cache interacts with the code/data location in RAM to alter performance on each run.
Some kind of cycle accurate emulator tool (not commercial).
Is it possible to have a degree of control over context switches? Or does it even matter? My thinking is the timing of the context switches is causing variability, maybe by causing the pipeline to be flushed at an inopportune time.
In the past I have had success on RISC architectures by counting instructions in the assembly listing. This only works, of course, if the number of instructions is small. Some compilers (like TI's Code Composer for the C67x) will give you a detailed analysis of how it's keeping the ALU busy.
I haven't found the assembly listings produced by GCC/GAS to be particularly informative. With full optimization on, code is moved all over the place. There can be multiple location directives for a single block of code dispersed about the assembly listing. Further, even if I could understand how the assembly maps back into my original code, I'm not sure there's much correlation between instruction count and performance on a modern x86 machine anyway.
I made a weak attempt at using gcov for line-by-line profiling, but due to an incompatibility between the version of GCC I built and the MinGW compiler, it wouldn't work.
One last thing you can do is average over many, many trial runs, but that takes forever.
EDIT (RE: Call Stack Sampling)
The first question I have is, practically, how do I do this? In one of your power point slides, you showed using Visual Studio to pause the program. What I have is a DLL compiled by GCC with full optimizations in Cygwin. This is then called by a mex DLL compiled by Matlab using the VS2013 compiler.
The reason I use Matlab is because I can easily experiment with different parameters and visualize the results without having to write or compile any low level code. Further, I can compare my optimized DLL to the high level Matlab code to ensure my optimizations have not broken anything.
The reason I use GCC is that I have a lot more experience with it than with Microsoft's compiler. I'm familiar with many flags and extensions. Further, Microsoft has been reluctant, at least in the past, to maintain and update the native C compiler (C99). Finally, I've seen GCC kick the pants off commercial compilers, and I've looked at the assembly listing to see how it's actually done. So I have some intuition of how the compiler actually thinks.
Now, with regards to making guesses about what to fix. This isn't really the issue; it's more like making guesses about how to fix it. In this example, as is often the case in numerical algorithms, there is really no I/O (excluding memory). There are no function calls. There's virtually no abstraction at all. It's like I'm sitting on top of a piece of saran wrap. I can see the computer architecture below, and there's really nothing in-between. If I re-rolled up all the loops, I could probably fit the code on about one page or so, and I could almost count the resultant assembly instructions. Then I could do a rough comparison to the theoretical number of operations a single core is capable of doing to see how close to optimal I am. The trouble then is I lose the auto-vectorization and instruction level parallelization I got from unrolling. Unrolled, the assembly listing is too long to analyze in this way.
The point is that there really isn't much to this code. However, due to the incredible complexity of the compiler and modern computer architecture, there is quite a bit of optimization to be had even at this level. But I don't know how small changes are going to affect the output of the compiled code. Let me give a couple of examples.
This first one is somewhat vague, but I'm sure I've seen it happen a few times. You make a small change and get a 10% improvement. You make another small change and get another 10% improvement. You undo the first change and get another 10% improvement. Huh? Compiler optimizations are neither linear, nor monotonic. It's possible, the second change required an additional register, which broke the first change by forcing the compiler to alter its register allocation algorithm. Maybe, the second optimization somehow occluded the compiler's ability to do optimizations which was fixed by undoing the first optimization. Who knows. Unless the compiler is introspective enough to dump its full analysis at every level of abstraction, you'll never really know how you ended up with the final assembly.
Here is a more specific example which happened to me recently. I was hand coding AVX intrinsics to speed up a filter operation. I thought I could unroll the outer loop to increase instruction level parallelism. So I did, and the result was that the code was twice as slow. What happened was there were not enough 256 bit registers to go around. So the compiler was temporarily saving results on the stack, which killed performance.
As I was alluding to in this post, which you commented on, it's best to tell the compiler what you want, but unfortunately, you often have no choice and are forced to hand tweak optimizations, usually via guess and check.
So I guess my question would be, in these scenarios (the code is effectively small until unrolled, each incremental performance change is small, and you're working at a very low level of abstraction), would it be better to have "precision of timing" or is call stack sampling better at telling me which code is superior?
I've faced a similar problem some time ago but that was on Linux which made it easier to tweak. Basically the noise introduced by OS (called "OS jitter") was as big as 5-10% in SPEC2000 tests (I can imagine it's much higher on Windows due to much bigger amount of bloatware).
I was able to bring deviation to below 1% by combination of the following:
disable dynamic frequency scaling (better do this both in BIOS and in Linux kernel as not all kernel versions do this reliably)
disable memory prefetching and other fancy settings like "Turbo boost", etc. (BIOS, again)
disable hyperthreading
enable high-performance process scheduler in kernel
bind process to core to prevent thread migration (use core 0 - for some reason it was more reliable on my kernel, go figure)
boot to single-user mode (in which no services are running) - this isn't as easy in modern systemd-based distros
disable ASLR
disable network
drop OS pagecache
There may be more to it but 1% noise was good enough for me.
I might put detailed instructions to github later today if you need them.
-- EDIT --
I've published my benchmarking script and instructions here.
Am I right that what you're doing is making an educated guess of what to fix, fixing it, and then trying to measure to see if it made any difference?
I do it a different way, which works especially well as the code gets large.
Rather than guess (which I certainly can) I let the program tell me how the time is spent, by using this method.
If the method tells me that roughly 30% is spent doing such-and-so, I can concentrate on finding a better way to do that.
Then I can run it and just time it.
I don't need a lot of precision.
If it's better, that's great.
If it's worse, I can undo the change.
If it's about the same, I can say "Oh well, maybe it didn't save much, but let's do it all again to find another problem,"
I need not worry.
If there's a way to speed up the program, this will pinpoint it.
And often the problem is not just a simple statement like "line or routine X spends Y% of the time", but "the reason it's doing that is Z in certain cases" and the actual fix may be elsewhere.
After fixing it, the process can be done again, because a different problem, which was small before, is now larger (as a percent, because the total has been reduced by fixing the first problem).
Repetition is the key, because each speedup factor multiplies all the previous, like compound interest.
When the program no longer points out things I can fix, I can be sure it is nearly optimal, or at least nobody else is likely to beat it.
And at no point in this process did I need to measure the time with much precision.
Afterwards, if I want to brag about it in a powerpoint, maybe I'll do multiple timings to get smaller standard error, but even then, what people really care about is the overall speedup factor, not the precision.

Nussinov Parallel

I am a biologist and I am trying to study computer languages. But, when I was trying to learn about the lpthread library, it seems odd as the result was lower than the sequential version.
In fact I am still reading the Tanenbaum book. But my main focus is to learn the basics of the calculations of the secondary structure of RNAs. So I found the explanation to the nussinov algorithm in a book and did indeed implement it. But when I tried to make a parallel version I believe that I might be missing the whole point, as this is my first contact with parallel implementations.
My questions are:
1. How should I implement a data-parallelism version for this algorithm ?
2. Why is my implementation slightly slower than the sequential one?
The code is available on: (each file is a different version parallel/sequential)
there are two ways to make a parallel version of an algorithm/program.
You study the algorithm and write the serial program. Afterwards, you start profiling the program to see where you can obtain speed gains. Those are the places where parallelism might come in handy (might, not will). I call this method the "desparate man's tool". This method is useful (!), but most of the times, the method beneath can provide better performance gains. This way of doing the optimisation method only takes programming and user experience into account.
You take the algorithm and try to figure out an other algorithm that permits parallel handling of the problem. Are there independent calculations or steps in the algorithm, are there parts of the algorithm that can be done before other parts completely finish, ... This could be called "the theoretical approach". Keep in mind that every thread has its overhead, and you don't want the overhead to be bigger than the gain you wish to obtain.
In fact, a combination of both is the best way to go (if parallelism is really necessary): first concentrate on method 2 (optimise the algorithm so that is stays scientifically correct, but can be treated in multi threading). Then look at the critical thread (can be found while profiling) and start optimising that thread.
As Kerrek SB already told: parallel programming is a very complex topic, with lots of possible pitfalls. And at the end of the road, you should ask yourself: is it worth the effort. After all: loosing weeks of study and programming time to gain some minutes is not worth your while.
On the other hand, if your program will run thousands of times, frustrating users due to long waiting times or a lack of responsiveness, than maybe, it could be useful to make a more performant version after all. But again: can't you reach the same goal by optimising a sequential version without the parallel clutter? Lot's of algorithms are of order O(exp(x)) or worse and can be reduced to O(x) or even O(log(x)).
Kind regards,

How much faster is C than R in practice?

I wrote a Gibbs sampler in R and decided to port it to C to see whether it would be faster. A lot of pages I have looked at claim that C will be up to 50 times faster, but every time I have used it, it's only about five or six times faster than R. My question is: is this to be expected, or are there tricks which I am not using which would make my C code significantly faster than this (like how using vectorization speeds up code in R)? I basically took the code and rewrote it in C, replacing matrix operations with for loops and making all the variables pointers.
Also, does anyone know of good resources for C from the point of view of an R programmer? There's an excellent book called The Art of R Programming by Matloff, but it seems to be written from the perspective of someone who already knows C.
Also, the screen tends to freeze when my C code is running in the standard R GUI for Windows. It doesn't crash; it unfreezes once the code has finished running, but it stops me from doing anything else in the GUI. Does anybody know how I could avoid this? I am calling the function using .C()
Many of the existing posts have explicit examples you can run, for example Darren Wilkinson has several posts on his blog analyzing this in different languages, and later even on different hardware (eg comparing his high-end laptop to his netbook and to a Raspberry Pi). Some of his posts are
the initial (then revised) post
another later post
and there are many more on his site -- these often compare C, Java, Python and more.
Now, I also turned this into a version using Rcpp -- see this blog post. We also used the same example in a comparison between Julia, Python and R/C++ at useR this summer so you should find plenty other examples and references. MCMC is widely used, and "easy pickings" for speedups.
Given these examples, allow me to add that I disagree with the two earlier comments your question received. The speed will not be the same, it is easy to do better in an example such as this, and your C/C++ skills will mostly determines how much better.
Finally, an often overlooked aspect is that the speed of the RNG matters a lot. Running down loops and adding things up is cheap -- doing "good" draws is not, and a lot of inter-system variation comes from that too.
About the GUI freezing, you might want to call R_CheckUserInterrupt and perhaps R_ProcessEvents every now and then.
I would say C, done properly, is much faster than R.
Some easy gains you could try:
Set the compiler to optimize for more speed.
Compiling with the -march flag.
Also if you're using VS, make sure you're compiling with release options, not debug.
Your observed performance difference will depend on a number of things: the type of operations that you are doing, how you write the C code, what type of compiler-level optimizations you use, your target CPU architecture, etc etc.
You can write basic, sloppy C and get something that works and runs with decent efficiency. You can also fine-tune your code for the unique characteristics of your target CPU - perhaps invoking specialized assembly instructions - and squeeze every last drop of performance that you can out of the code. You could even write code that runs significantly slower than the R version. C gives you a lot of flexibility. The limiting factor here is how much time that you want to put into writing and optimizing the C code.
The reverse is also true (duplicate the previous paragraph here, but swap "C" and "R").
I'm not trying to sound facetious, but there's really not a straightforward answer to your question. The only way to tell how much faster your C version would be is to write the code both ways and benchmark them.

what are the steps/strategy to analyze and improve performance of an embedded system

I will break down this question in to sub questions. I am confused if I should ask them separately or in one question. So I will just stick to one SO question.
What are generally the steps to analyze and improve performance of C applications?
Do these steps change if I am developing for an embedded system?
What tools are out there which can help me?
Recently I have been given a task to improve the performance of our product on ARM11 platform. I am relatively new to this field of embedded systems and need gurus here on SO to help me out.
simply changing compilers can improve your C performance for the same source code by many times over. GCC has not necessarily gotten better for performance over the years, for some programs gcc 3.x produces much tighter code than 4.x. Back when I had access to the tools, ARMs compiler produced significantly better code than gcc. As much as 3 or 4 times faster. LLVM has caught up to GCC 4.x and I suspect will pass gcc by in terms of performance and overall use for cross compiling embedded code. Try different versions of gcc, 3.x and 4.x if you are using gcc. Metaware's compiler and arms adt ran circles around gcc3.x, gcc3.x will give gcc4.x a run for its money with arm code, for thumb code gcc4.x is better and for thumb2 (which doesnt apply to you) gcc4.x also better. Remember I have not said a word about changing a single line of code (yet).
LLVM is capable of full program optimization in addition to infinitely more tuning knobs than gcc. Despite that the code generated (ver 27) is only just catching up to the current gcc 4.x in terms of performance for the few programs I tried. And I didnt try the n factoral number of optimization combinations (optimize on the compile step, different options for each file, or combine two files or three files or all files and optimize those bundles, my theory is do no optimization on the C to bc steps, link all the bc together then do a single optimization pass on the whole program, the allow the default optimization when llc takes it to the target).
By the same token simply knowing your compiler and the optimizations can greatly improve the performance of the code without having to change any of it. You have an ARM11 arr you compiling for arm11 or generic arm? You can gain a few to a dozen percent by telling the compiler specifically which architecture/family (armv6 for example) over the generic armv4 (ARM7) that is often chosen as the default. Knowing to use -O2 or -O3 if you are brave.
It is often not the case but switching to thumb mode can improve performance for specific platforms. Doesnt apply to you but the gameboy advance is a perfect example, loaded with non-zero wait state 16 bit busses. Thumb has a handful of a percent overhead because it takes more instructions to do the same thing, but by increasing the fetch times, and taking advantage of some of the sequential read features of the gba thumb code can run significantly faster than arm code for the same source code.
having an arm11 you probably have an L1 and maybe L2 cache, are they on? Are they configured? Do you have an mmu and is your heavy use memory cached? or are you running zero wait state memory and dont need a cache and should turn it off? In addition to not realizing that you can take the same source code and make it run many times faster by changing compilers or options, folks often dont realize that when you use a cache simply adding a single up to a few nops in your startup code (as a trick to adjust where code lands in memory by one, two, a few words) you can change your codes execution speed by as much as 10 to 20 percent. Where those cache line reads hit in heavily used functions/loops makes a big difference. Even saving one cache line read by adjusting where the code lands is noticeable (cutting it from 3 to 2 or 2 to 1 for example).
Knowing your architecture, both the processor and your memory environment is where the tuning if any would start. Most C libraries if you are high level enough to use one (I often dont use a C library as I run without an operating system and with very limited resources) both in their C code and sometimes add some assembler to make bottleneck routines like memcpy, much faster. If your programs are operating on aligned 32 or even better 64 bit addresses, and you adjust even if it means using a handful of bytes more memory for every structure/array/memcpy to be an integral multiple of 32 bits or 64 bits you will see noticeable improvements (if your code uses structs or copies data in other ways). In addition to getting your structures (if you use them, I certainly dont with embedded code) size aligned, even if you waste memory, getting elements aligned, consider using 32 bit integers for every element instead of bytes or halfwords. Depending on your memory system this can help (it can hurt too btw). As with the GBA example above looking at specific functions that either by profiling or intuition you know are not being implemented in a manner that takes advantage of your processor or platform or libraries you may want to turn to assembler either from scratch or compiling from C initially then disassembling and hand tuning. Memcpy is a good example you may know your systems memory performance and may chose to create your own memcpy specifically for aligned data, copying 64 or 128 or more bits per instruction.
Likewise mixing global and local variables can make a noticeable performance difference. Traditionally folks are told never to use globals, but in embedded this isnt necessarily true, depends on how deeply embedded and how much tuning and speed and other factors you are interested in. This is a touchy subject and I may get flamed for it, so I will leave it at that.
The compiler has to burn and evict registers in order to make function calls, plus if you use local variables a stack frame may be required, so function calls are expensive, but at the same time, depending on the code within a function that has now grown in size by avoiding functions, you may create the problem you were trying to avoid, evicting registers to re-use them. Even a single line of C code can make the difference between all the variables in a function fits in registers to having to start evicting a bunch of registers. For functions or segments of code where you know you need some performance gain compile and disassemble (and look at register usage, how often it fetches memory or writes to memory). You can and will find places where you need to take a well used loop and make it its own function even though the function call has a penalty because by doing that the compiler can better optimize the loop and not evict/reuse registers and you get an overall net gain. Even a single extra instruction in a loop that goes around hundreds of times is a measurable performance hit.
Hopefully you already know to absolutely not compile for debug, turn all of the compile for debug options off. You may already know that code compile for debug that runs without bugs doesnt mean it is debugged, compiling for debug and using debuggers hide bugs leaving them as time bombs in your code for your final compile for release. Learn to always compile for release and test with the release version both for performance and finding bugs in your code.
Most instruction sets do not have a divide function. Avoid using divides or modulo in your code as much as humanly possible they are performance killers. Naturally this is not the case for powers of two, to save the compiler and to mentally avoid divides and modulos try to use shifts and ands. Multplies are easier and more often found in instruction sets, but are still costly. This is a good case to write assembler to do your multiplies instead of letting the C copiler do it. The arm multiply is a 32bit * 32bit = 32 bit so to do accurate math without overflowing there has to be extra C code wrapped around the multiply, if you already know you wont overflow, burn the registers for a function call and do the multiply in assembler (for the arm).
Likewise most instruction sets do not have a floating point unit, with yours you might, even so avoid float if at all possible. If you have to use float that is a whole other pandora's box of performance issues. Most folks dont see the performance problems with code as simple as this:
float a,b;
a = b * 7.0;
The rest of the problem is not understanding floating point accuracy and how good or bad the C libraries are just trying to get your constants into floating point form. Again float is a whole other long discussion on performance problems.
I am a product of Michael Abrash (I actually have a print copy of zen of assembly language) and the bottom line is time your code. Come up with an accurate way to time the code, you may think you know where the bottlenecks are and you may think you know your architecture but trying different things even if you think they are wrong, and timing them you may find and eventually have to figure out the error in your thinking. Adding nops to start.S as a final tuning step is a good example of this, all the other work you have done for performance can be instantly erased by not having a good alignment with the cache, this also means re-arranging functions within your source code so that they land in different places in the binary image. I have seen 10 to 20 percent swings of speed increase and decrease as a result of cache line alignments.
Code Review:
What are good code review techniques ?
Static and dynamic analysis of the code.
Tools for static analysis: Sparrow, Prevent, Klockworks
Tools for dynamic analysis : Valgrind, purify
Gprof allows you to learn where your program spent its time and which functions called which other functions while it was executing.
Steps are same
Apart from what is listed is point 1, there are tools like memcheck etc.
There is a big list here based on platform
Phew!! Quite a big question!
What are generally the steps to
analyze and improve performance of C
As well as other static code analysers mentioned here there is a fairly cheap version called PC-Lint which has been around for ages. Sometimes throws up lots of errors and warnings for one error but by the end of it you'll be happy and know waaaaay more about C/C++ because of it.
With all code analysers some of the issues may be more structural to the code so best to start analysing it from day 1 of coding; running analysis on old software may swamp you with issues which may take a while to untangle, best to keep it clean from the beginning.
But code analysers will not catch all logical errors, i.e. it doesn't do what you want it to do! These are best done by code reviews first, then testing. Performance is often improved by by trying to keep the algorithms as simple as possible, keeping instructions in loops tight, possibly unrolling loops (your compiler optimisations may do this), use of fast caches when accessing data which is slow to get.
Code reviews can raise a lot of issues from lots of other peoples eyes looking at it. Don't get too many people, try to get 3 other people if possible, sometimes junior developers ask the most insightful questions like, "why are we doing this?".
Testing can be roughly split into two sections, automated and manual. Automated testing requires effort producing test handlers for functions/units but once run can be run again and again very quickly. Manual testing requires planning, self-discipline to perform them all to the required, imagination to think up of scenarios that may impair performance and you have to be observant (you may have passed the test but the 'scope trace has a bit of an anomaly before/after the test).
"Do these steps change if I am
developing for an embedded system?"
Performance ananlysis can be different on embedded systems to applications systems; with the very broad brush that "embedded" now covers it depends how hardware-centric you are. It can be done using profilers, if you want a more cheap and chearful method then use test output pins to measure sections of code, or measure them with breakpoints on simulators that come with the development environment.
Make sure that not just a typical length of task is measured but also a maximum, as that is where one task may start impeding on other tasks and your scheduled tasks are not completed in time.
What tools are out there which can
help me?
Simulators on the IDEs, static analysis tools, dynamic analysis tools, but most of all you and other humans getting the requirements right, decent reviewing (of code and testing) and thorough testing (automated and manual).
Good luck!
My experiences.
Function calls are slow, eliminate with macros or inlined methods. Look at the disassembler listing to see.
If using GCC, mark optimized sections with #pragma GCC optimize("O3") or compile them separately.
Play with different combinations of applying the inline attribute (basically find a balance between size and speed).
It is a difficult question to be answered shortly since various techniques have been proposed such as flowchart and state diagram,so you can take a look at some titles:
ARM System-on-Chip Architecture, 2nd Edition -- Steve Furber
ARM System Developer's Guide - Designing and Optimizing System Software -- Andrew N. Sloss, Dominic Symes, Chris Wright & John Rayfield
The Definitive Guide to the ARM Cortex-M3 --Joseph Yiu
C Programming for Embedded Systems --Kirk Zurell
Embedded C -- Michael J. Pont
Programming Embedded Systems in C and C++ --Michael Barr
An Embedded Software Primer --David E, Simon
Embedded Microprocessor Systems 3rd Edition --Stuart Ball
Global Specification and Validation of Embedded Systems - Integrating Heterogeneous Components --G. Nicolescu & A.A Jerraya
Embedded Systems: Modeling, Technology and Applications --Gunter Hommel & Sheng Huanye
Embedded Systems and Computer Architecture --Graham Wilson
Designing Embedded Hardware --John Catsoulis
You have to use a profiler. It will help you identify your application's bottleneck(s). Then focus on improving the functions you spend the most time in and the ones you call the most. Repeat this procedure until you're satisfied with your application performance.
No they don't.
Depending on the platform you're developing onto :
Windows : AMD Code Analyst, VTune, Sleepy
Linux : valgrind / callgrind / cachegrind
Mac : the Xcode profiler is quite good.
Try to find a profiler for the architecture you actually work on.

Profiling C code on Windows when using Eclipse

I know I can profile my code with gprof and kprof on Linux. Is there a comparable alternative to these applications on Windows?
Commercial software:
Rational Quantify (expensive, slow, but very detailed)
AQTime (less expensive, less slow, a bit detailed)
Free software:
Very sleepy (
Luke StackWalker (
These commercial alternatives change the compiled code by 'instrumenting' (adding instructions) to it and perform the timing withing the added instructions. This means that they cause your application to slow down seriously.
These free alternatives use sampling, meaning they are less detailed, but very fast. In practice I found that especially Very Sleepy is very good to have a quick look at performance problems in your application.
There's a MinGW port of gprof that works just about the same as the Linux variant. You can either get a full MinGW installation (I think gprof is included but not sure) or get gprof from the MinGW binutils package.
For Eclipse, there's TPTP but it doesn't support profiling C/C++ as far as I know.
Yes, you can profile code with Visual Studio
What's the reason for profiling? Do you want to a) measure times and get a call graph, or b) find things to change to make the code faster? (These are not the same.)
If (b) you can use this trick, using the Pause button in Eclipse.
Added: Maybe it would help to convey some experience of what performance problems are actually like, and where you can expect to find them. Here are some simple examples:
An insertion sort (order n^2) where the items being sorted are strings, and are compared by a string-compare function. Where is the hot-spot? in string-compare. Where is the problem? In the sort where string-compare is called. If n=10 it's not a problem, but if n=1000, suddenly it takes a long time. The point where string-compare is called is "cold", but that's where the problem is. A small number of samples of the call stack pinpoint it with certainty.
An app that loads plugins takes a long time to start up. A profiler says basically everything in it is "cold". Something that measures I/O time says it is almost all I/O time, which seems like what you might expect, so it might seem hopeless. But, stack samples show a large percentage of time is spent with the stack about 20 layers deep in the process of reading the resource part of plugin dlls for the purpose of translating string constants into the local language. Investigating further, you find that most of the strings being translated are not the the kind that actually need translation. They were just put in "in case" they might need translation, and were never thought to be something that could cause a performance problem. Fixing that issue brings a hefty time savings.
So it is common to think in terms of "hotspots" and "bottlenecks", but most programs, especially the larger ones, tend to have performance problems in the form of function calls requesting work that doesn't really need to be done. Fortunately they display themselves on the call stack during the time that they are spending.