[LLVMdev] Inline hint for methods defined in-class

Hal Finkel hfinkel at anl.gov
Thu Jul 9 18:46:22 PDT 2015


----- Original Message -----
> From: "Richard Smith" <richard at metafoo.co.uk>
> To: "Hal Finkel" <hfinkel at anl.gov>
> Cc: "cfe commits" <cfe-commits at cs.uiuc.edu>, "List" <llvmdev at cs.uiuc.edu>, "Xinliang David Li" <davidxl at google.com>
> Sent: Thursday, July 9, 2015 8:08:38 PM
> Subject: Re: [LLVMdev] Inline hint for methods defined in-class
> 
> On Thu, Jul 9, 2015 at 4:30 PM, Hal Finkel < hfinkel at anl.gov > wrote:
> 
> 
> ----- Original Message -----
> > From: "Richard Smith" < richard at metafoo.co.uk >
> > To: "Xinliang David Li" < davidxl at google.com >
> > Cc: "cfe commits" < cfe-commits at cs.uiuc.edu >, "<
> > llvmdev at cs.uiuc.edu > List" < llvmdev at cs.uiuc.edu >
> > Sent: Thursday, July 9, 2015 3:40:54 PM
> > Subject: Re: [LLVMdev] Inline hint for methods defined in-class
> > 
> 
> 
> > On Tue, Jul 7, 2015 at 10:25 PM, Xinliang David Li <
> > davidxl at google.com > wrote:
> > 
> > 
> > On Tue, Jul 7, 2015 at 6:06 PM, Chandler Carruth <
> > chandlerc at gmail.com > wrote:
> > > On Tue, Jul 7, 2015 at 4:11 PM Easwaran Raman < eraman at google.com
> > > >
> > > wrote:
> > >> 
> > >> I'm reviving this thread after a while and CCing cfe-commits as
> > >> suggested by David Blaikie. I've also collected numbers building
> > >> chrome (from chromium, on Linux) with and without this patch as
> > >> suggested by David. I've re-posted the proposed patch and
> > >> performance/size numbers collected at the top to make it easily
> > >> readable for those reading it through cfe-commits.
> > > 
> > > 
> > > First off, thanks for collecting the numbers and broadening the
> > > distribution. Also, sorry it took me so long to get over to this
> > > thread.
> > > 
> > > I want to lay out my stance on this issue at a theoretical and
> > > practical
> > > level first. I'll follow up with thoughts on the numbers as well
> > > after that.
> > > 
> > > I think that tying *any* optimizer behavior to the 'inline'
> > > keyword
> > > is
> > > fundamentally the wrong direction.
> > 
> > Chandler, thanks for sharing your thought -- however I don't think
> > it
> > is wrong, let alone 'fundamentally wrong'. Despite all the analysis
> > that can be done, the inliner is in the end heuristic based. In
> > lack
> > of the profile data, when inlining two calls yield the same static
> > benefit and size cost, it is reasonable for the inliner to think
> > the
> > call to the function with inline hint to yield more high
> > dynamic/runtime benefit -- thus it has a higher static size budget
> > to
> > burn.
> > 
> > >We have reasons why we have done this
> > > historically, and we can't just do an immediate about face, but
> > > we
> > > should be
> > > actively looking for ways to *reduce* the optimizer's reliance on
> > > this
> > > keyword to convey any meaning whatsoever.
> > 
> > yes those additional things will be done, but they are really
> > orthogonal.
> > 
> > > 
> > > The reason I think that is the correct direction is because, for
> > > better or
> > > worse, the 'inline' keyword in C++ is not about optimization, but
> > > about
> > > linkage.
> > 
> > It is about both optimization and linkage. In fact the linkage
> > simply
> > serves as an implementation detail. In C++ standard 7.1.2,
> > paragraph
> > 2 says:
> > 
> > "A function declaration (8.3.5, 9.3, 11.3) with an inline specifier
> > declares an inline function. The inline specifier indicates to the
> > implementation that inline substitution of the function body at the
> > point of call is to be preferred to the usual function call
> > mechanism.
> > An implementation is not required to perform this inline
> > substitution
> > at the point of call; however, even if this inline substitution is
> > omitted, the other rules for inline functions defined by 7.1.2
> > shall
> > still be respected."
> > 
> > 
> > 
> > This describes Clang's current behavior, not the behavior you
> > propose.
> > 
> > 
> > Note that there is a difference between an "inline specifier"
> > (meaning that the inline keyword appeared in the declaration, see
> > the grammar description in paragraph 1 of 7.1.2, and the
> > introductory text in 7.1) and a function being an "inline function"
> > (meaning that multiple definitions are permitted in different
> > translation units, see 3.2/4 and /6).
> > 
> > 
> > The above wording clearly says that the inline *specifier* is a
> > hint
> > that we should inline the function, and...
> 
> Is it really that clear?
> 
> 1.3p3 says, "Terms that are used only in a small portion of this
> International Standard are defined where they
> are used and italicized where they are defined.".
> 
> 7.1.2p2 says, "A function declaration (8.3.5, 9.3, 11.3) with an
> inline specifier declares an inline function." and "inline function"
> is italicized there. Thus, "inline function", as a term, is
> *defined* to be a function declared with an inline specifier.
> 
> 
> We "misuse" italics in this way in various places,
> where we define a
> term and then later give some exceptions. (See 8.4.3/1 for example;
> there are deleted functions that do not have that syntactic form.)
> 
> 
> 7.1.2p3 then also says that a "function defined within a class
> definition is an inline function." I would read this to mean, "as if
> it had an inline specifier", because that is how the term is defined
> in the paragraph above it.
> 
> Reading the words in that way results in rejecting valid code.
> Consider this:
> 
> 
> struct X {
> 
> constexpr int f();
> };
> 
> 
> 7.1.2/3 says "If the inline specifier is used in a friend
> declaration, that declaration shall be a definition or the function
> shall have previously been declared inline."
> 
> 
> Now, a function declared with the constexpr specifier is implicitly
> an inline function, but it is not declared with the inline specifier
> , so this program is well-formed.
> 

I disagree. The standard simply provides a set of constructs that are also inline functions as if they had been declared with an inline specifier.

> 
> And, frankly, I don't think the intent of the drafting was to create
> a bifurcated system. If we're to read "inline function" as something
> other than, "a function intended to be inlined", it seems the
> standard would have used a different term to indicate the linkage
> effect from the hinting.
> 
> 
> It does; the terms are "declared with an inline specifier" and
> "inline function"
> 

To be pedantic, "declared with an inline specifier" does not seem to occur in the standard. The closest thing I see is 7.1.2p2, "A function declaration (8.3.5, 9.3, 11.3) with an inline specifier declares an inline function.", and that *defines* the term "inline function".

> 
> In any case, the point is: it is not reasonable to use the standard's
> wording to justify when to provide an inline hint.

It is perfectly reasonable; the standard contains a facility designed explicitly for this purpose, and it happens to be the facility under discussion.

> Even if we agreed
> that it said that any inline function is intended to undergo inline
> substitution, that is only a non-binding suggestion due to the as-if
> rule, and we are under no obligation to base our inlining decision
> on it.
> 

Agreed.

> 
> 
> 
> > Developers see those and rely on those to give compiler the hints.
> 
> 
> 
> Likewise here, different developers have different expectations, so I
> don't think we can use this argument to make the decision.
> 
> 
> There seem to be two relevant factors that should affect our
> decision:
> 
> 
> 1) What signals best indicate that inlining is beneficial for
> existing code? That is, which heuristics make us optimize better?
> and

It is not quite that simple. We also need to give a high-quality user experience, and that means taking advantage of language facilities designed to provide optimization hints in a consistent way. Now we can debate what consistent is... ;)

> 2) Which signals are reasonable for us to use, given the current and
> expected future state of the C++ language? We don't want people
> contorting their code in order to get it well-optimized (such as
> moving functions out of their class definition because they turn out
> to be somewhat non-trivial, and inlining them is harmful),
> and we
> want to allow inline hints to be given in ways that are orthogonal
> to program semantics (the inline specifier is not good for this,
> because it also carries real language semantics, not just an
> optimization hint).
> 

I agree this is an unfortunate combination of properties for 'inline' to have. But it was not done by accident, and the intent of the language seems to be not to give us the option of treating them separately and still provide a consistent user experience. We certainly *can* and still have a conforming implementation... we could also only consider the first 100 inline specifiers for hinting and ignore the rest. But, in my experience, users have expectations that inline functions are (at least more likely) to be inlined than other functions of similar size, and this includes functions defined in the class definition. Furthermore, the data seems to indicate that this is the right approach.

I also don't want users to burden their code with the 'inline' keyword on nearly every (or every) in-class function definition because, even though defining it there is supposed to make it an inline function, that will make it a 'really inline' inline function.

Thanks again,
Hal

> 
> 
> 
> 
> > Most importantly, paragraph 3 says:
> > 
> > "A function defined within a class definition is an inline
> > function.
> > 
> > 
> > 
> > ... the same is *not* true for a function definition that appears
> > within a class definition. That is merely an inline function (that
> > is, it can be defined in multiple translation units).
> > 
> > 
> > The inline specifier shall not appear on a block scope function
> > declaration.93 If the inline specifier is used in a friend
> > declaration, that declaration shall be a definition or the function
> > shall have previously been declared inline."
> > 
> > Here we can see regardless of how optimizer will honor the hint and
> > to
> > what extent, and based on what analysis,
> > it is basically incorrect to drop the attribute on the floor for
> > in-class function definitions. Eswaran's fix is justified with this
> > reason alone. The side effect of changing inliner behavior is
> > irrelevant.
> > 
> > > It has a functional impact and can be both necessary or
> > > impossible
> > > to use to meet those functional requirements. This in turn leaves
> > > programmers in a lurch if the functional requirements are ever in
> > > tension
> > > with the optimizer requirements.
> > 
> > Not sure what you mean. Performance conscious programmers use it
> > all
> > the time.
> > 
> > > 
> > > We're also working really hard to get more widely deployed
> > > cross-module
> > > optimization strategies, in part to free programmers from the
> > > requirement
> > > that they put all their performance critical code in header
> > > files.
> > > That
> > > makes compilation faster, and has lots of benefits to the
> > > factoring
> > > and
> > > design of the code itself. We shouldn't then create an incentive
> > > to
> > > keep
> > > things in header files so that they pick up a hint to the
> > > optimizer.
> > 
> > > 
> > > Ultimately, the world will be a better place if we can eventually
> > > move code
> > > away from relying on the hint provided by the 'inline' keyword to
> > > the
> > > optimizer.
> > > 
> > 
> > While I would like to see that happen some day, I do think it is an
> > independent matter.
> > 
> > > 
> > > That doesn't mean that the core concept of hinting to the
> > > optimizer
> > > that a
> > > particular function is a particularly good candidate for inlining
> > > is without
> > > value.
> > 
> > yes.
> > 
> > >While I think it is a bad practice that we shouldn't encourage in
> > > code (especially portable code)
> > 
> > yes -- there are indeed programmers who use this casually without
> > considering performance.
> > 
> > > I can see the desire to at least have *some*
> > > attribute which is nothing more or less than a hint to the
> > > optimizer to
> > > inline harder[1].
> > 
> > yes -- there are programmers who use the attribute consciously.
> > 
> > > It would help people work around inliner bugs in the short
> > > term, and even help debug inliner-rooted optimization problems.
> > 
> > I think it is a good hint to the compiler even in the longer term.
> > With PGO, we should minimize the reliance on the hint though.
> > 
> > >Codebases
> > > with strong portability requirements could still (and probably
> > > should)
> > > forbid or tightly control access to this kind of hint. I would
> > > want
> > > really
> > > strong documentation about how this attribute *completely voids*
> > > your
> > > performance warranty (if such a thing exists) as from version to
> > > version of
> > > the compiler it may go from a helpful hint to a devastatingly bad
> > > hint.
> > 
> > Why? If the compiler becomes smarter and smarter, the inline hint
> > will
> > become more and more irrelevant and eventually has no effect -- why
> > would the performance warranty be voided? If the compiler is not
> > yet
> > smart enough, why would the compiler refuse to take the hint and
> > forbid developer provide the hint?
> > 
> > > But
> > > I think I could be persuaded to live with such a hint existing.
> > > But
> > > I'm
> > > *really* uncomfortable with it being tied to something that also
> > > impacts
> > > linkage or other semantics of the program.
> > 
> > For consistent with standard, we should pass the attribute. Linkage
> > is
> > not affected in anyway.
> > 
> > > 
> > > [1]: Currently, the only other hint we have available is pretty
> > > terrible as
> > > it *also* has semantic effects: the always_inline attribute.
> > > 
> > > 
> > >> 
> > >> The proposed patch will add InlineHint to methods defined inside
> > >> a
> > >> class:
> > >> 
> > >> --- a/lib/CodeGen/CodeGenFunction.cpp
> > >> +++ b/lib/CodeGen/CodeGenFunction.cpp
> > >> @@ -630,7 +630,7 @@ void
> > >> CodeGenFunction::StartFunction(GlobalDecl
> > >> GD,
> > >> if (const FunctionDecl *FD = dyn_cast_or_null<FunctionDecl>(D))
> > >> {
> > >> if (!CGM.getCodeGenOpts().NoInline) {
> > >> for (auto RI : FD->redecls())
> > >> - if (RI->isInlineSpecified()) {
> > >> + if (RI->isInlined()) {
> > >> Fn->addFnAttr(llvm::Attribute::InlineHint);
> > >> break;
> > >> }
> > >> 
> > >> Here are the performance and size numbers I've collected:
> > >> 
> > >> 
> > >> - C++ subset of Spec: No performance effects, < 0.1% size
> > >> increase
> > >> (all size numbers are text sizes returned by 'size')
> > >> - Clang: 0.9% performance improvement (both -O0 and -O2 on a
> > >> large
> > >> .ii
> > >> file) , 4.1% size increase
> > > 
> > > 
> > > FWIW, this size increase seems *really* bad. I think that kills
> > > this
> > > approach even in the short term.
> > 
> > Re. size and performance trade-off -- 0.9% performance improvement
> > should greatly win the size cost. Besides among all programs see,
> > only
> > clang sees this size increase with all the others seeing negligible
> > size increase.
> > 
> > This is not a short term vs long term situation. It is basically a
> > bug
> > fix that FE drops the attribute. If it exposes inliner heuristic
> > bug,
> > that should be fixed/tuned separately. With the hint correctly
> > passed
> > in, Easwaran will do further tuning including time based analysis.
> > 
> > > 
> > >> 
> > >> - Chrome: no performance improvement, 0.24% size increase
> > >> - Google internal benchmark suite (geomean of ~20 benchmarks):
> > >> ~1.8%
> > >> performance improvement, no size regression
> > > 
> > > 
> > > I'm also somewhat worried about the lack of any performance
> > > improvements
> > > outside of the Google benchmarks. That somewhat strongly suggests
> > > that our
> > > benchmarks are overly coupled to this hint already. The fact that
> > > neither
> > > Chrome, Clang, nor SPEC improved is... not at all encouraging.
> > 
> > Other than Google benchmarks, we do see Clang improve performance.
> > Besides, current inliner needs to be further tuned in order to get
> > more performance benefit. Passing the hint through is simply an
> > enabler. Also remember that most of SPEC benchmarks are C programs.
> > C++ programs with heavy use of virtual functions may not benefit a
> > lot
> > either.
> > 
> > David
> > 
> > 
> > > 
> > > -Chandler
> > _______________________________________________
> > cfe-commits mailing list
> > cfe-commits at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
> > 
> > 
> > _______________________________________________
> > cfe-commits mailing list
> > cfe-commits at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
> > 
> 
> 
> 
> --
> Hal Finkel
> Assistant Computational Scientist
> Leadership Computing Facility
> Argonne National Laboratory
> 
> 

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory



More information about the cfe-commits mailing list