Err, "GCC does this optimization, because strlen is a "built-in function:"
No.
I wrote the optimization pass that does this (GVN PRE with SCC based value numbering).
It does it to any pure/const function. Const ones no matter what, and pure ones if it can prove the global memory state does not otherwise change between calls in a way that impacts that pure call (IE there are times it can prove the calls that happen in between don't matter, and will still do the elimination).
You don't usually have to mark the functions const/pure if they are visible to GCC, it does interprocedural analysis to prove they are pure/const and mark them for you.
It will even do it through function pointers if the value numbering or alias analysis can prove what they point to, or that they don't change in between the calls.
double cos (double) __attribute__ ((const));
double sin (double) __attribute__ ((const));
double f(double a)
{
double b;
double c,d;
double (*fp) (double) __attribute__ ((const));
/* Partially redundant call */
if (a < 2.0)
{
fp = sin;
c = fp (a);
}
else
{
c = 1.0;
fp = cos;
}
d = fp (a);
return d + c;
}
In this example, it will eliminate the unconditional fp(a) call at the end by reusing the value of c in the first if block, and storing the result of calling cos(a) in the else block.
IE:
double cos (double) __attribute__ ((const));
double sin (double) __attribute__ ((const));
double f(double a)
{
double b;
double c,d;
double (*fp) (double) __attribute__ ((const));
/* Partially redundant call */
if (a < 2.0)
{
fp = sin;
c = fp (a);
temp = c
}
else
{
c = 1.0;
fp = cos;
temp = cos(a)
}
d = temp;
return d + c;
}
(attribute pure is ignored on function pointers, so you can't just s/const/pure/ and expect it to work)
Loop code hoisting is actually a special case of this partial redundancy elimination.
double sin (double) __attribute__ ((const));
double f (double a)
{
int i;
double c, d;
double (*fp) (double) __attribute__ ((const));
fp = sin;
for (i = 0; i < 50; i++)
{
c = fp (a);
}
d = fp (a);
return d + c;
}
The call to fp(a) in the loop will be hoisted above the loop through redundancy elimination, the call d = fp(a) will be eliminated in favor of that hoisted value.
Basically, the optimization is a lot more powerful than he describes.
OP here. Thanks for the great description of loop hoisting of invariant code, and it's awesome to hear from the guy who wrote it.
However, what I wrote was not wrong. strlen is not marked as pure or const on OS X, and yet it is hoisted. glibc does happen to mark strlen as pure in string.h, but the optimization occurs even if you do not include that header - even if there is no declaration for strlen in scope at all!
So this optimization really does proceed because strlen is a builtin function, and does not depend on strlen being marked as pure or const.
Under the hood, it may be that strlen is rewritten as __builtin_strlen, which is marked as pure and therefore hoisted like any pure expression. Or it may be due to some of the other magic, like the pass that optimizes strlen("foo") to 3. I don't know which optimization pass is responsible, and it didn't seem important, so I kept my mouth shut on that aspect.
"So this optimization really does proceed because strlen is a builtin function, and does not depend on strlen being marked as pure or const."
No. It's exactly the other way around.
We could not give a crap whether it is a built in function, only whether it is pure or const.
We do not special case built-in functions anywhere near this optimization.
The special casing you see in the first part is trying to constant fold a few built-in calls in a utility function, and trying to see through memcpys for memory state.
The reason the optimization proceeds is because strlen gets marked as pure by the compiler if there is no non-pure definition that overrides it.
Basically, the compiler defines a function named "strlen" that is pure and nothrow behind your back, but you can override it by providing your own definition.
This is unrelated to whether it is a builtin (because the builtin version is __builtin_strlen)
"strlen gets marked as pure by the compiler if there is no non-pure definition that overrides it"
That's what I meant when I wrote that "the compiler recognizes strlen, and optimizes it specially." It was not my intent to say that only builtins or all builtins may be hoisted in this way, although now I see how someone could interpret that. That was bad wording on my part.
The point I was trying to communicate is that gcc can do special optimizations on functions that it recognizes as builtins - for example, replacing printf() with fputs(). One illustration is how strlen is treated as pure, even if it is not marked as pure. As someone with far more gcc expertise than me, would you agree with that point?
"Basically, the compiler defines a function named "strlen" that is pure and nothrow behind your back, but you can override it by providing your own definition. This is unrelated to whether it is a builtin"
Well, the optimization is defeated by -fno-builtin, so I assumed that the underlying mechanism is that a call to strlen is replaced by a call to the builtin. Was this wrong?
It is wrong, but only because of the weirdness of how this works.
The call to strlen is not replaced with builtin_strlen, it defines both a builtin_strlen and a strlen.
Both are marked pure and nothrow.
Due to some wonderful oddness around freestanding environments, if you use -fno-builtin, it will define neither.
I've done this many times. This is special case that happens if you test with values inside the code. The compiler that I use, Visual C++, is smart enough to know that the result can be known at compile time, so it does just that in (Release).
What you need to do is to read the arguments from a file or standard input.
I don't spend a lot of time looking at complier outputs, but the optimization is quite obvious in the first case. I don't have a clue how it would optimize it out in the second case though...
In the second case the divide function isn't being exported from the compilation unit and the compiler can tell[1], so the compiler is free to do whatever it wants with the calling convention and in this case it will choose to inline. And once it's inlined finding the correct answer at compile time is still obvious.
[1] Normally you have to declare a function as 'static' to tell the compiler that the function isn't being exported form the compilation unit, since the compiler isn't supposed to be able to tell the difference between function declarations in the .c file itself and function declarations that were included from .h files. But in this case the function is just defined and never declared, so I presume that's the same as a static declaration.
As mpyne explains, the function still is externally visible.
However, the compiler is perfectly free to both compile it as an external function and inline it wherever it wants. Yes, setting breakpoints on such a function would be 'funny', but that's nothing of real concern to the compiler.
No, it's still an extern declaration. However the compiler is still allowed to "peek within" an extern function as long as it has the definition available.
You can see this by inspecting the disassembly yourself, you'll see that printf never actually calls divide, and yet the divide function is still defined within the executable output.
I did some similar experiments myself a long time ago. Something mildly surprising for me at that time was that this got optimized to a single roll instruction:
I haven't managed to make GCC generate a non-constant roll (i.e. (x << c) | (x >> (32-c)) ), probably because it couldn't infer that c is in the [0,31] range.
Question 4 points at a huge surprise. With a scientific computing project, I got (if I recall correctly), correct and expected output at no optimization and -O1, and patently incorrect output at -O2 and -O3.
Cost: one sleepless night in college."What do you mean, optimization doesn't just make it faster?!?"
Really? Wow, I thought optimisations had to provably avoid affecting the correctness of your program. Hence:
"This need to avoid imprecise intermediate results often prevents compilers from exploiting mathematical properties like associativity when optimizing FP arithmetic."
Not really; there's no rounding problem with x << 1. x * 2, x+x, x<<1 are all equivalent. The real reason it doesn't use the shift instruction on x86 is because it takes 3 bytes to encode the instruction whereas adding a register to itself takes two. But like it says, for powerpc it does use a shift as there's no difference in instruction lengths on a RISC.
random aside: gcc usually uses the LEA instruction for this sort of thing, which lets you compute any combination of {1,2,4,8} * x + y + n in one cycle where x and y are registers and n is a constant number. So even with the LEA instruction you can use either lea eax, [eax*2] or lea eax, [eax+eax] to compute this. And so on. I think add eax, eax is most likely what it does though.
Compilers for x86 prefer adds to shifts because there are processors (x86 exists for so long) on which shift costs more cycles. LEAs are always fast as they are needed for always used addressings.
Sometimes there are "canonical forms" for these operations depending on what chip is being targeted, where the hardware can automatically break data dependencies and improve hardware-level parallelism, as long as the instruction is encoded in the right form.
I don't know that this is necessarily the reason here, but it's one possible explanation.
What an interesting read. Thanks OP. I got stumbled on the floating point question. I remember a few weeks ago in class our instructor taught us about floating point precision loss and I thought the compiler might take care into that and not optimize it, but I guess I was wrong.
Would you explain a bit more on that? I'm really interested! Thanks!
No. I wrote the optimization pass that does this (GVN PRE with SCC based value numbering).
It does it to any pure/const function. Const ones no matter what, and pure ones if it can prove the global memory state does not otherwise change between calls in a way that impacts that pure call (IE there are times it can prove the calls that happen in between don't matter, and will still do the elimination). You don't usually have to mark the functions const/pure if they are visible to GCC, it does interprocedural analysis to prove they are pure/const and mark them for you.
It will even do it through function pointers if the value numbering or alias analysis can prove what they point to, or that they don't change in between the calls.
In this example, it will eliminate the unconditional fp(a) call at the end by reusing the value of c in the first if block, and storing the result of calling cos(a) in the else block. IE: (attribute pure is ignored on function pointers, so you can't just s/const/pure/ and expect it to work)Loop code hoisting is actually a special case of this partial redundancy elimination.
The call to fp(a) in the loop will be hoisted above the loop through redundancy elimination, the call d = fp(a) will be eliminated in favor of that hoisted value.Basically, the optimization is a lot more powerful than he describes.