I've been trying to use C in Keil to write some test code for my TM4C123G, which uses an ARM microcontroller. I have no clue about ARM assembly, but I have written some assembly code for an AVR microcontroller in the past.
Where are the values of the variables stored, if we declare a variable in C as global, as opposed to declaring it in main?
Are there general guidelines as to whether we should declare a variable global as opposed to in main (when it comes to writing C for a microcontroller) ?
Globals in ARM cause an offset to be placed in the calling function's "pool" area. In general, each global needs its own offset. If you do decided to use globals, place them into a single structure so that all the variables can be accessed via a singel pool offset.
Variables defined at the function level live on the stack. These, within reason, can be accessed with a simpler offset from the stack register and tend to be a touch more efficient opcode wise, and perhaps cache wise, depending on what kind of system you are using.
When to use either? Outside of the global vs. local holy wars of maintainability, it comes down to what else your code wants to do. If some variable is being used in a ton of places and you don't want to pass it around as a parameter (a sign that perhaps your design needs work...) then a global would be the way to go. If you need to access this variable across multiple assembler packages (that desing thing again...) then perhaps a global is also fine.
Most of the time though, I would go with local variables. They are much more thread safe, and from a above, tend to be more efficient. They are easier to maintain as well, because there scope is much more confined - which also helps compilers do there jobs better as well.
The compiler will put global variables in the "data" or "bss" memory segment. This allocation is permanent, so the variable will never lose its value. Function-local variables are allocated dynamically on the stack, and disapears when the function returns.
Use global variables when the variable must keep its value between function calls, and must be accessible by code in multiple functions and/or files. Use function-local variables for everything else.
There are also "static" variables. They function the same way as global variables, but have a more limited namespace. They are used as file-local variables and function-local variables which keep their value between function calls.
Globals are fine they have a place especially in embedded things like microcontrollers where you are extremely tight on resources. Globals make it easy to manage your resources where locals are very dynamic and difficult at best to not get into trouble.
But... As far as assembly goes, there are no rules, there is not really a notion of local versus global that is strictly a higher level language thing. Now there may be implementations of assembly languages that allow/force such things, understand that C is a standard (as others) that cross platforms and are not specific to one. But assembly not only does not have a standard, but is also processor specific. the assembly language is defined by the assembler (the program that parses it and turns it into machine code). And anyone and their brother can bang out an assembler and make up whatever language or rules they want.
In general if you are trying to hand implement C code in assembly instead of having the compiler do it (or instead of just having the compiler do it and at least see what it does). Your globals unless optimized out are going to get a home in ram. Locals may or may not depending on optimization get a place on the stack, they may just live temporarily in registers, depends on the processor number of registers available the trade off between preserving registers that contained something else to preserving the local in favor of keeping the something else in a register.
You should just take some simple functions (not necessarily complete programs) and see what the compiler does.
if you want to write in pure assembly and not have the notion of converting C you would still have the same dilemma of do I keep something in a register for a long time, do I put it on the stack for the duration of this chunk of code or do I assign it an address where it lives forever.
I suggest you be open minded and try to understand the whys and why nots, a lot of these kinds of rules, no globals, small functions, etc are preached not because of facts but just because of faith, someone I trust told me so I preach it as well, not always but sometimes you can dig in and find that the fear is real or the fear is not real, or maybe it applied 30 years ago but not anymore, and so on. An alternative to a global for example is a local at a main() or top level that you keep passing down as you nest in, which basically means it is a global from a resource perspective. In fact depending on the compiler (an extremely popular one in particular) that one main level local that is passed down, actually consumes resources at each nesting level consuming a significantly higher quantity of ram than if it had just been declared a global. The other side if it is not inefficient memory consumption but access, who can mess with that variable and who cant, locals make that quite easy, a fair amount of laziness, you can be messy. Have to take care with globals to not mess them up. Note static locals are also globals from a resource perspective they sit in the same .data space for the duration of the program.
There is a reason, many reasons why C cannot die. Nothing has come along that can replace it. There is a reason why it is basically the first compiler for each new processor/instruction set.
Write some simple functions, compile them, disassemble, see what is produced. take some of your embedded/bare metal applications for these platforms, disassemble, see what the compiler has done. Globals and static locals that cant be optimized out get an address in .data. Sometimes some or all of the incoming parameters get stack locations and sometimes some or all of the local variables consume stack as well, has to do with the compiler and optimization if you chose to optimize. Also depends on the architecture a CISC that can do memory and register or memory to memory operations doesnt have to move stuff in and out of registers all the time RISC often does have to exclusively use registers or often use registers, but also often has a lot more available. So the compiler knows that and manages the home for the variables as a result.
Everyone is going to put globals and static locals in ram (unless they can optimize out). traditional x86 brings the parameters in on the stack anyway and is going to put the locals there too and just access them there. a mips or arm is going to try to minimize the amount of ram used, and lean on registers, trying to keep some of the variables in registers exclusively and not consume stack.
An assembly programmer, pre-compilers, used a lot of the equivalent of globals, pick an address for each variable. now post compilers, you can choose to just do it that way setup all of your variables and hardly use the stack other than returning from calls, or you can program as if you were a compiler and make some rules or simply follow the compilers convention and have some disposable registers and others preserved and lean on the stack for preservation and local to the function items.
At the end of the day though assembly does not have a notion of global vs local any more than it has a notion of signed vs unsigned vs pointer vs array or anything like that that a high level language would have.
Related
Is this possible to tell the compiler to put a certain global variable in a register? Thus effectively blocking this register from use everywhere else. I understand this might be difficult because even a simple call to malloc() will require spilling it temporarily.
I am thinking about it because I am looking for a way to efficiently implement a secondary stack for sort of VM and naturally it would be great to store the secondary stack pointer in another register.
In theory, this is possible: You could take any register that the calling conventions require to be preserved across function calls, and use that for your global variable.
However, there are some problems with this:
The effect is, that your functions will have one less register for local variables available. This means, more memory accesses on average.
Library functions will not preserve the registers value, they will save it on the stack as any other non-clobber register, use it as they please, and restore it before they return.
As such, it is impossible to pass a callback pointer to a library function, and access the register global from the callback. (But this is probably not your problem anyway.)
If you are implementing a VM stack as a global variable, you are doing something very wrong in the first place. A stack should be thread local by nature, it has no business being global.
Doing things right, keeping the stack pointer as a local variable within the VM emulator is likely to give you the best performance you can get.
Is this possible to tell the compiler to put a certain global variable in a register?
Not really. There is the register storage class, but this only means that the variable should be "as fast as possible". This keyword is mostly obsolete nowadays, it is from a time when compilers were trash.
Thus effectively blocking this register from use everywhere else
It's not possible to select a specific register. You will have to use inline assembler for that.
I am thinking about it because I am looking for a way to efficiently implement a secondary stack for sort of VM and naturally it would be great to store the secondary stack pointer in another register.
Sounds like you need to be writing assembler overall. It is not even possible to set the stack pointer from C. Or if by "stack" you don't mean the program memory, but rather some data type, don't go fiddle with pre-mature optimizations in the first place.
All,
I have C functions that are called many times a second as they are part of a control loop on a PIC18 board. These functions have variables that only need method scope, but I was wondering what if any overhead there was to constantly allocating these variables vs. using a global or at least higher scoped variable. (Thought of typedef'ing a struct to pass around from a higher scope to avoid global variable use if performance dictates not using method local varables)
There are some good threads on here that cover this topic, but I have yet to see a definitive answer as most preach best practices which I agree and would follow as long as there are not performance gains to be had as every microsecond counts.
One thread mentioned using file scoped static variables as a substitute for global variables, but I can't help wonder if even that is necessary.
What does everyone think?
Accessing a local variable requires doing something like *(SP + offset) (where SP is the stack-pointer), whereas accessing a static (which includes globals) requires something like *(address).
From what I recall, the PIC instruction set has very limited addressing modes. So it's very likely that accessing the global will be faster, at least for the first time it's accessed. Subsequent accesses may be identical if the compiler holds the computed address in a register.
As #unwind said in the comments, you should take a look at the compiler output, and profile to confirm. I would only sacrifice clarity/maintainability if you've proved that it's worthwhile in terms of the runtime of your program.
While I've not used every single PIC compiler in existence, there are two styles. The style I've used allocates all local variables statically by analyzing the program's call graph. If every possible call were in fact performed, the amount of stack memory consumed by locals would match what would be required by static allocation, with a couple of caveats (describing the behavior of HiTech's PICC-18 "standard" compiler--others may vary)
Variadic functions are handled by defining local-variable storage in the scope of the caller, and passing a two-byte pointer to that storage to the function being called.
For every different signature of indirect function pointer, the compiler generates a "pseudo-function" in the call graph; everything that calls a function of that signature calls the pseudo-function, and that pseudo-function calls every function with that signature that has its address taken.
In this style of compiler, consecutive accesses to local variables will be just as fast as consecutive accesses to globals. Other than global and static variables explicitly-declared as "near", however, which must total no more than 64-128 bytes (varies with different models of PIC), the global and static variables for each module are located separately from local variables, and bank-switching instructions are needed to access things in different banks.
Some compilers which I have not used employ the "enhanced instruction set" option. This option gobbles up 96 bytes of the "near" bank (or all of it, on PICs with less than 96 bytes) and uses it to access 96 bytes relative to the FSR2 register. This would be a wonderful concept if it used the first 16, or maybe 32, bytes as a stack frame. Using 96 bytes means giving up all of the "near" storage, which is a pretty severe limitation. Nonetheless, compilers which use this instruction set can access local variables on a stack just as fast, if not faster, than global variables (no bank-switch required). I really wish Microchip had an option to only set aside 16 bytes or so for the stack frame, leaving a useful amount of 'common bank' RAM, but nonetheless some people have good luck with that mode.
I would imagine that this depends a lot on which compiler you are using. I don't know PIC but I'm guessing some (all?) PIC compilers will optimize the code so that local variables are stored in CPU registers whenever possible. If so, then local variables will likely be equally fast as globals.
Otherwise if the local variable is allocated on the stack the global may be a bit faster to access (see Oli's answer).
I just to want to get an idea about how the register variables are handled in C program executables. ie in which location(or register) it exactly get stored in case of an embedded system and in a X86 machine(C program executable in a desktop PC)?
What about this view? (correct me if am wrong)
Suppose we have declared/initialized one variable inside a function as 'int' datatype. Normally it will go to the stack segment and it will be there in that section only at run time ,when the caller calls the callee containing the local variable. But if we declare above local variable as 'register int' then also it'll go to the stack segment. But on run time , the processor put that local variable from stack to its general purpose register locations(because of extra compiler inserted code due to 'register' keyword) and a fast access of the same from there.
That is the only difference between them is at run time access and there is no memory loading differences between them.
__Kanu
The register keyword in C (rarely ever seen anymore) is only a hint to the compiler that it may be useful to keep a variable in a register for faster access.
The compiler is free to ignore the hint, and optimize as it sees best.
Since modern compilers are much better than humans at understanding usage and speed, the register keyword is usually ignored by modern compilers, and in some cases, may actually slow down execution speed.
From K&R C:
A register variable advises the
compiler that the variable in question
will be heavily used. The idea is that
register variables are to be placed in
machine registers, which may result in
smaller & faster programs. But
compilers are free to ignore this
advice.
It is not possible to take the address of a register variable, regardless of whether the variable is actually placed in a register.
Hence,
register int x;
int *y = &x; // is illegal
So, you must weigh in the cons of not being able to get the address of the register variable.
In addition to crypto's answer (that has my vote) just see the name register for the keyword as a historical misnomer. It has not much to do with registers as you learn it in class e.g for the von Neumann processor model, but is just a hint to the compiler that this variable doesn't need an address.
On modern machines an addressless variable can be realized by different means (e.g an immediate assembler operator) or optimized away completely. Tagging a variable as register can be a useful optimization hint for the compiler and also a useful discipline for the programmer.
When a compiler takes its internal code and the backend turns it into machine/assembler for the target processor, it keeps track of the registers it is generating instructions for as it creates the code. When it needs to allocate a register to load or keep track of a variable if there is an unused working variable then it marks it as used and generates the instructions using that register. But if all the working registers have something in them then it will usually evict the contents of one of those registers somewhere, often ram for example global memory or the stack if that variable had a home. The compiler may or may not be smart about that decision and may evict a variable that is highly used. By using the register keyword, depending on the compiler, you may be able to influence that decision, it may choose to keep the register keyword variables in registers and evict non-register keyword variables to memory as needed.
which location(or register) it exactly get stored in case of an embedded system and in a > X86 machine(C program executable in a desktop PC)?
You don't know without opening up the assembly output, which will be liable to shift based on compiler choices. It's a good idea to check the assembly just for educational purposes.
If you need to read and write particular registers that precisely, you should write inline assembly or link in an assembly module.
Typically when using a standard C compiler for x86/amd64 (gcc, icc, cl), you can reasonably assume that the compiler will optimize sufficiently well for most purposes.
If, however, you are using a non-standard compiler, e.g., one cooked up for a new embedded system, it is a good idea to consider hand optimization. If the architecture is new, it might also be a good idea to consider hand optimization.
My function will be called thousands of times. If i want to make it faster, will changing the local function variables to static be of any use? My logic behind this is that, because static variables are persistent between function calls, they are allocated only the first time, and thus, every subsequent call will not allocate memory for them and will become faster, because the memory allocation step is not done.
Also, if the above is true, then would using global variables instead of parameters be faster to pass information to the function every time it is called? i think space for parameters is also allocated on every function call, to allow for recursion (that's why recursion uses up more memory), but since my function is not recursive, and if my reasoning is correct, then taking off parameters will in theory make it faster.
I know these things I want to do are horrible programming habits, but please, tell me if it is wise. I am going to try it anyway but please give me your opinion.
The overhead of local variables is zero. Each time you call a function, you are already setting up the stack for the parameters, return values, etc. Adding local variables means that you're adding a slightly bigger number to the stack pointer (a number which is computed at compile time).
Also, local variables are probably faster due to cache locality.
If you are only calling your function "thousands" of times (not millions or billions), then you should be looking at your algorithm for optimization opportunities after you have run a profiler.
Re: cache locality (read more here):
Frequently accessed global variables probably have temporal locality. They also may be copied to a register during function execution, but will be written back into memory (cache) after a function returns (otherwise they wouldn't be accessible to anything else; registers don't have addresses).
Local variables will generally have both temporal and spatial locality (they get that by virtue of being created on the stack). Additionally, they may be "allocated" directly to registers and never be written to memory.
The best way to find out is to actually run a profiler. This can be as simple as executing several timed tests using both methods and then averaging out the results and comparing, or you may consider a full-blown profiling tool which attaches itself to a process and graphs out memory use over time and execution speed.
Do not perform random micro code-tuning because you have a gut feeling it will be faster. Compilers all have slightly different implementations of things and what is true on one compiler on one environment may be false on another configuration.
To tackle that comment about fewer parameters: the process of "inlining" functions essentially removes the overhead related to calling a function. Chances are a small function will be automatically in-lined by the compiler, but you can suggest a function be inlined as well.
In a different language, C++, the new standard coming out supports perfect forwarding, and perfect move semantics with rvalue references which removes the need for temporaries in certain cases which can reduce the cost of calling a function.
I suspect you're prematurely optimizing, however, you should not be this concerned with performance until you've discovered your real bottlenecks.
Absolutly not! The only "performance" difference is when variables are initialised
int anint = 42;
vs
static int anint = 42;
In the first case the integer will be set to 42 every time the function is called in the second case ot will be set to 42 when the program is loaded.
However the difference is so trivial as to be barely noticable. Its a common misconception that storage has to be allocated for "automatic" variables on every call. This is not so C uses the already allocated space in the stack for these variables.
Static variables may actually slow you down as its some aggresive optimisations are not possible on static variables. Also as locals are in a contiguous area of the stack they are easier to cache efficiently.
There is no one answer to this. It will vary with the CPU, the compiler, the compiler flags, the number of local variables you have, what the CPU's been doing before you call the function, and quite possibly the phase of the moon.
Consider two extremes; if you have only one or a few local variables, it/they might easily be stored in registers rather than be allocated memory locations at all. If register "pressure" is sufficiently low that this may happen without executing any instructions at all.
At the opposite extreme there are a few machines (e.g., IBM mainframes) that don't have stacks at all. In this case, what we'd normally think of as stack frames are actually allocated as a linked list on the heap. As you'd probably guess, this can be quite slow.
When it comes to accessing the variables, the situation's somewhat similar -- access to a machine register is pretty well guaranteed to be faster than anything allocated in memory can possible hope for. OTOH, it's possible for access to variables on the stack to be pretty slow -- it normally requires something like an indexed indirect access, which (especially with older CPUs) tends to be fairly slow. OTOH, access to a global (which a static is, even though its name isn't globally visible) typically requires forming an absolute address, which some CPUs penalize to some degree as well.
Bottom line: even the advice to profile your code may be misplaced -- the difference may easily be so tiny that even a profiler won't detect it dependably, and the only way to be sure is to examine the assembly language that's produced (and spend a few years learning assembly language well enough to know say anything when you do look at it). The other side of this is that when you're dealing with a difference you can't even measure dependably, the chances that it'll have a material effect on the speed of real code is so remote that it's probably not worth the trouble.
It looks like the static vs non-static has been completely covered but on the topic of global variables. Often these will slow down a programs execution rather than speed it up.
The reason is that tightly scoped variables make it easy for the compiler to heavily optimise, if the compiler has to look all over your application for instances of where a global might be used then its optimising won't be as good.
This is compounded when you introduce pointers, say you have the following code:
int myFunction()
{
SomeStruct *A, *B;
FillOutSomeStruct(B);
memcpy(A, B, sizeof(A);
return A.result;
}
the compiler knows that the pointer A and B can never overlap and so it can optimise the copy. If A and B are global then they could possibly point to overlapping or identical memory, this means the compiler must 'play it safe' which is slower. The problem is generally called 'pointer aliasing' and can occur in lots of situations not just memory copies.
http://en.wikipedia.org/wiki/Pointer_alias
Using static variables may make a function a tiny bit faster. However, this will cause problems if you ever want to make your program multi-threaded. Since static variables are shared between function invocations, invoking the function simultaneously in different threads will result in undefined behaviour. Multi-threading is the type of thing you may want to do in the future to really speed up your code.
Most of the things you mentioned are referred to as micro-optimizations. Generally, worrying about these kind of things is a bad idea. It makes your code harder to read, and harder to maintain. It's also highly likely to introduce bugs. You'll likely get more bang for your buck doing optimizations at a higher level.
As M2tM suggests, running a profiler is also a good idea. Check out gprof for one which is quite easy to use.
You can always time your application to truly determine what is fastest. Here is what I understand: (all of this depends on the architecture of your processor, btw)
C functions create a stack frame, which is where passed parameters are put, and local variables are put, as well as the return pointer back to where the caller called the function. There is no memory management allocation here. It usually a simple pointer movement and thats it. Accessing data off the stack is also pretty quick. Penalties usually come into play when you're dealing with pointers.
As for global or static variables, they're the same...from the standpoint that they're going to be allocated in the same region of memory. Accessing these may use a different method of access than local variables, depends on the compiler.
The major difference between your scenarios is memory footprint, not so much speed.
Using static variables can actually make your code significantly slower. Static variables must exist in a 'data' region of memory. In order to use that variable, the function must execute a load instruction to read from main memory, or a store instruction to write to it. If that region is not in the cache, you lose many cycles. A local variable that lives on the stack will most surely have an address that is in the cache, and might even be in a cpu register, never appearing in memory at all.
I agree with the others comments about profiling to find out stuff like that, but generally speaking, function static variables should be slower. If you want them, what you are really after is a global. Function statics insert code/data to check if the thing has been initialized already that gets run every time your function is called.
Profiling may not see the difference, disassembling and knowing what to look for might.
I suspect you are only going to get a variation as much as a few clock cycles per loop (on average depending on the compiler, etc). Sometimes the change will be dramatic improvement or dramatically slower, and that wont necessarily be because the variables home has moved to/from the stack. Lets say you save four clock cycles per function call for 10000 calls on a 2ghz processor. Very rough calculation: 20 microseconds saved. Is 20 microseconds a lot or a little compared to your current execution time?
You will likely get more a performance improvement by making all of your char and short variables into ints, among other things. Micro-optimization is a good thing to know but takes lots of time experimenting, disassembling, timing the execution of your code, understanding that fewer instructions does not necessarily mean faster for example.
Take your specific program, disassemble both the function in question and the code that calls it. With and without the static. If you gain only one or two instructions and this is the only optimization you are going to do, it is probably not worth it. You may not be able to see the difference while profiling. Changes in where the cache lines hit could show up in profiling before changes in the code for example.
for a long time, I am thinking and studying output of C language compiler in assembler form, as well as CPU architecture. I know this may be silly to you, but it seems to me that something is very ineffective. Please, don´t be angry if I am wrong, and there is some reason I do not see for all these principles. I will be very glad if you tell me why is it designed this way. I actually truly believe I am wrong, I know the genius minds of people which get PCs together knew a reason to do so. What exactly, do you ask? I´ll tell you right away, I use C as a example:
1: Stack local scope memory allocation:
So, typical local memory allocation uses stack. Just copy esp to ebp and than allocate all the memory via ebp. OK, I would understand this if you explicitly need allocate RAM by default stack values, but if I do understand it correctly, modern OS use paging as a translation layer between application and physical RAM, when address you desire is further translated before reaching actual RAM byte. So why don´t just say 0x00000000 is int a,0x00000004 is int b and so? And access them just by mov 0x00000000,#10? Because you wont actually access memory blocks 0x00000000 and 0x00000004 but those your OS set the paging tables to. Actually, since memory allocation by ebp and esp use indirect addressing, "my" way would be even faster.
2: Variable allocation duplicity:
When you run application, Loader load its code into RAM. When you create variable, or string, compiler generates code that pushes these values on the top o stack when created in main. So there is actual instruction for do so, and that actual number in memory. So, there are 2 entries of the same value in RAM. One in form of instruction, second in form of actual bytes in the RAM. But why? Why not to just when declaring variable count at which memory block it would be, than when used, just insert this memory location?
How would you implement recursive functions? What you are describing is equivalent to using global variables everywhere.
That's just one problem. How can you link to a precompiled object file and be sure it won't corrupt the memory of your procedures?
Because C (and most other languages) support recursion, so a function can call itself, and each call of the function needs separate copies of any local variables. Also, on most current processors, your way would actually be slower -- indirect addressing is so common that processors are optimized for it.
You seem to want the behavior of C (or at least that C allows) for string literals. There are good and bad points to this, such as the fact that even though you've defined a "variable", you can't actually modify its contents (without affecting other variables that are pointing at the same location).
The answers to your questions are mostly wrapped up in the different semantics of different storage classes
Google "data segment"
Think about the difference in behavior between global and local variables.
Think about how constant and non-constant variables have different requirements when functions are called repeatedly (or as Mehrdad says, recursively)
Think about the difference between static and non static automatic variables again in the context of multiple or recursive calls.
Since you are comparing assembler and c (which are very close together from an architectural standpoint), I'm inclined to say that you're describing micro-optimization, which is meaningless unless you profile the code to see if it performs better.
In general, programming languages are evolving towards a more declarative style (i.e. telling the computer what you want done, rather than how you want it done). When you program in an imperative language (like assembly or c), you specify in extreme detail how you want the problem solved. This gives the compiler little room to make optimization decisions on your behalf.
However, as the languages become more declarative, the compilers are getting smarter, because we are giving them the room they need to make more intelligent performance optimizations.
If every function would put its first variable at offset 0 and so on then you would have to change the memory mapping each time you enter a function (you could not allocate all variables to unique addresses if you want recursion). This is doable, but with current hardware it's very slow. Furthermore, the address translation performed by the virtual memory is not free either, it's actually quite complicated to implement this efficiently.
Addressing off ebp (or any other register) costs having a mux (to select the register) and an adder (to add the offset to the register). The time taken for this can often be overlapped with other operations.
If you want to be able to modify the static value you have to copy it to the stack. If you don't (saying it's 'const') then a good C compiler will no copy it to the stack.