[llvm-dev] PGO is ineffective for Rust - but why?
Michael Woerister via llvm-dev
As part of my work for Mozilla's Low Level Tools team I've
implemented PGO in the Rust compiler. The feature is
available since Rust 1.37 [1]. However, so far we have not
seen any actual performance gains from enabling PGO for
Rust code. Performance even seems to drop 1-3% with PGO
enabled. I wonder why that is and I'm hoping that someone
here might have experience debugging PGO effectiveness.
PGO in the Rust compiler
------------------------
The Rust compiler uses IR-level instrumentation (the
equivalent of Clang's `-fprofile-generate`/`-fprofile-use`).
This has worked pretty well and even enables doing PGO for
mixed Rust/C++ codebases when also using Clang.
The Rust compiler has regression tests that make sure that:
- instrumentation shows up in LLVM IR for the `generate` phase,
and that
- profiling data is actually used during the `use` phase, i.e.
that cold functions get marked with `cold` and hot functions
get marked with `inline`.
I also verified manually that `branch_weights` are being set
in IR. So, from my perspective, the PGO implementation does
what it is supposed to do.
However, as already mentioned, in all benchmarks I've seen so
far performance seems to stay the same at best and often even
suffers slightly. Which is suprising because for C++ code
using Clang's version of IR-level instrumentation & PGO brings
signifcant gains (up to 5-10% from what I've seen in
benchmarks for Firefox).
One thing we noticed early on is that disabling the
pre-inlining pass (`-disable-preinline`) seems to consistently
improve the situation for Rust code. Doing that we sometimes
see performance wins of almost 1% over not using PGO. This
again is very different to C++ where disabling this pass
causes dramatic performance loses for the Firefox benchmarks.
And 1% performance improvement is still well below
expectations, I think.
So my questions to you are:
- Has anybody here observed something similar while
wokring on or with PGO?
- Are there certain known characteristics of LLVM IR code
that inhibit PGO's effectiveness and that IR produced by
`rustc` might exhibit?
- Does anybody know of a good source that describes how to
effectively debug a problem like this?
- Does anybody know of a small example program in C/C++
that is known to profit from PGO and that could be
re-implemented in Rust for comparison?
Thanks a lot for reading! Any help is appreciated.
-Michael
[1] https://blog.rust-lang.org/2019/08/15/Rust-1.37.0.html#profile-guided-optimization
_______________________________________________
LLVM Developers mailing list
llvm...@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
Teresa Johnson via llvm-dev
| Teresa Johnson | | Software Engineer | | tejo...@google.com | |
Teresa Johnson via llvm-dev
I just have a couple suggestions off the top of my head:- have you tried using the new pass manager (-fexperimental-new-pass-manager)? That has access to additional analysis info during inlining and is able to make more precise PGO based inline decisions.
Xinliang David Li via llvm-dev
Philip Reames via llvm-dev
On 9/12/19 2:18 AM, Michael Woerister via llvm-dev wrote:
> Hi everyone,
>
> As part of my work for Mozilla's Low Level Tools team I've
> implemented PGO in the Rust compiler. The feature is
> available since Rust 1.37 [1]. However, so far we have not
> seen any actual performance gains from enabling PGO for
> Rust code. Performance even seems to drop 1-3% with PGO
> enabled. I wonder why that is and I'm hoping that someone
> here might have experience debugging PGO effectiveness.
>
>
> PGO in the Rust compiler
> ------------------------
>
> The Rust compiler uses IR-level instrumentation (the
> equivalent of Clang's `-fprofile-generate`/`-fprofile-use`).
> This has worked pretty well and even enables doing PGO for
> mixed Rust/C++ codebases when also using Clang.
>
> The Rust compiler has regression tests that make sure that:
>
> - instrumentation shows up in LLVM IR for the `generate` phase,
> and that
>
> - profiling data is actually used during the `use` phase, i.e.
> that cold functions get marked with `cold` and hot functions
> get marked with `inline`.
>
> I also verified manually that `branch_weights` are being set
> in IR. So, from my perspective, the PGO implementation does
> what it is supposed to do.
One thing missing here is profile guided devirtualization. That's super
significant for Java; it might be highly relevant for Rust as well.
However, I'd still expect to see *some* positive delta with what you've
got, so don't start here. Your immediate problem is likely something else.
>
> However, as already mentioned, in all benchmarks I've seen so
> far performance seems to stay the same at best and often even
> suffers slightly. Which is suprising because for C++ code
> using Clang's version of IR-level instrumentation & PGO brings
> signifcant gains (up to 5-10% from what I've seen in
> benchmarks for Firefox).
>
> One thing we noticed early on is that disabling the
> pre-inlining pass (`-disable-preinline`) seems to consistently
> improve the situation for Rust code. Doing that we sometimes
> see performance wins of almost 1% over not using PGO. This
> again is very different to C++ where disabling this pass
> causes dramatic performance loses for the Firefox benchmarks.
> And 1% performance improvement is still well below
> expectations, I think.
>
> So my questions to you are:
>
> - Has anybody here observed something similar while
> wokring on or with PGO?
>
> - Are there certain known characteristics of LLVM IR code
> that inhibit PGO's effectiveness and that IR produced by
> `rustc` might exhibit?
Have you checked to make sure *all* of your branches have weights?
Including the ones which don't directly correspond to Rust
conditionals? If you left off branch weights from range checks or
something (i.e something with a ton of occurrences) that might be
confusing the heuristics enough to explain your results.
Michael Woerister via llvm-dev
Michael Woerister via llvm-dev
confirmed that `rustc` indeed uses `-ffunction-sections` and
`-fdata-sections` on all platforms except for macOS. When trying out
different linkers for a small test case [1], however, I found that
there were rather large differences in execution time:
ld (no PGO) = 172 ms
ld (PGO) = 196 ms
gold (no PGO) = 182 ms
gold (PGO) = 141 ms
lld (no PGO) = 193 ms
lld (PGO) = 171 ms
So `gold` and `lld` both profit from PGO quite a bit, while `ld`
linked programs are slower with PGO. I then noticed that branch
weights for `ld` were missing from most branches, while the counts for
the other linkers are correct. All of this suggests to me that
something goes wrong when `ld` tries to link in the profiling runtime.
I'll be investigating further.
[1] https://github.com/michaelwoerister/rust-pgo-test-programs/tree/master/branch_weights
On Thu, Sep 12, 2019 at 6:31 PM Teresa Johnson <tejo...@google.com> wrote:
>
>
>
Teresa Johnson via llvm-dev
Xinliang David Li via llvm-dev
Michael Woerister via llvm-dev
> By ld do you mean GNU ld?
> I know GNU ld does "work" with LLVM's gold plugin, but it's an untested combination and not recommended.
Michael Woerister via llvm-dev
Yes, good observation! Whether the bug occurs depends only on the
linker used for creating the instrumented binary. The linker used
during the "use" phase makes no difference.
> If that is the case, try dump the sections of the resulting binary and compare __llvm_prf_** sections.
For the final instrumented executable, it looks like the
`__llvm_prf_data` section is 480 bytes large when using GNU ld, while
it is 528 bytes for gold and lld. The size difference (48 bytes)
incidentally is exactly the size of the `__llvm_prf_data` section in
the object file containing the code that is later missing branch
weights. It looks like the GNU linker loses the `__llvm_prf_data`
section from that object file?
> Also check the arguments passed to the linker. It should have -u__llvm_profile_runtime to force the profile runtime to be linked in.
`-u__llvm_profile_runtime` is properly passed to the linker,
regardless of which linker it is.
Michael Woerister via llvm-dev
Clang 8 does not have any problems with GNU ld: The `__llvm_prf_data`
section is the same size for all three linkers. It must be something
specific to the Rust compiler that's going wrong here.
[1] https://github.com/michaelwoerister/rust-pgo-test-programs/tree/master/cpp_branch_weights
Teresa Johnson via llvm-dev
> By ld do you mean GNU ld?
Yes, GNU ld version 2.31.1 on Fedora 30.
> I know GNU ld does "work" with LLVM's gold plugin, but it's an untested combination and not recommended.
That's good to know! However, in this case no linker plugin is involved. All of LLVM is executed within the Rust compiler and the linker only ever gets to see regular object files.
Xinliang David Li via llvm-dev
Hiroshi Yamauchi via llvm-dev
Hi everyone,
As part of my work for Mozilla's Low Level Tools team I've
implemented PGO in the Rust compiler. The feature is
available since Rust 1.37 [1]. However, so far we have not
seen any actual performance gains from enabling PGO for
Rust code. Performance even seems to drop 1-3% with PGO
enabled. I wonder why that is and I'm hoping that someone
here might have experience debugging PGO effectiveness.
PGO in the Rust compiler
------------------------
The Rust compiler uses IR-level instrumentation (the
equivalent of Clang's `-fprofile-generate`/`-fprofile-use`).
This has worked pretty well and even enables doing PGO for
mixed Rust/C++ codebases when also using Clang.
The Rust compiler has regression tests that make sure that:
- instrumentation shows up in LLVM IR for the `generate` phase,
and that
- profiling data is actually used during the `use` phase, i.e.
that cold functions get marked with `cold` and hot functions
get marked with `inline`.
I also verified manually that `branch_weights` are being set
in IR. So, from my perspective, the PGO implementation does
what it is supposed to do.
Michael Woerister via llvm-dev
Yes, they are present in the IR generate during the `profile-use`
phase. For my test case you can take a look at the IR generated here:
generate-phase:
https://github.com/michaelwoerister/rust-pgo-test-programs/blob/master/branch_weights/outputs/opt_lib_gen.lld.ll
use-phase: https://github.com/michaelwoerister/rust-pgo-test-programs/blob/master/branch_weights/outputs/opt_lib_use.lld.ll
source code: https://github.com/michaelwoerister/rust-pgo-test-programs/blob/master/branch_weights/opt_lib.rs
However, when using the right linker and thus not running into the GNU
ld bug mentioned earlier, I'm seeing proper speedups with PGO for this
test case. For other (larger) test cases I still don't see speedups so
I'll need to take a closer look at those.
Michael Woerister via llvm-dev
Down the road we definitely will want to combine PGO with ThinLTO.
Michael Woerister via llvm-dev
- I've been further investigating and found an issue [1] with the
Cargo build tool that most Rust projects use. This issue prevents all
projects using Cargo from properly using PGO because it causes symbol
names to be different between the generate and the use phase. With
this issue fixed the number of "No profile data available for
function" warnings goes down from 92369 to 1167 for the Firefox
codebase.
- I also found that the potential GNU ld bug mentioned above
apparently does not affect Firefox. The number of "No profile data
available for function" warnings is exactly the same for GNU ld and
LLD. I don't know yet where the remaining 1167 warnings come from
though.
- Unfortunately, even with all of the above fixes applied, my medium
sized benchmark still performs worse with PGO than without it. For my
tiny example [2] PGO reduces the number of branch misses by more than
50%. For the medium sized benchmark, however, the PGO version has
slightly *more* branch misses. This seems to indicate that there is
still something wrong.
I will further investigate.
[1] https://github.com/rust-lang/cargo/issues/7416
[2] https://github.com/michaelwoerister/rust-pgo-test-programs/tree/master/branch_weights/
Michael Woerister via llvm-dev
to the conclusion that, with the previously mentioned Cargo issue [1]
fixed, profile-guided optimization now works as expected with Rust. I
have a number of reasons to think so:
- I did some semi-automated investigation of benchmarks that did not
show much of a speedup and was not able to find any missing branch
weights or function call counts. The concrete branch weights that are
easy to predict (error paths in code that does not error during
instrumentation runs) also looked correct to me. I subsequently added
regression tests to the Rust compiler which make sure that branch
weights are correct in a number of basic cases.
- I also investigated indirect call promotion and it seems that
idiomatic Rust code just contains very few indirect calls. I added
regression tests that make sure that indirect call promotion is
correctly performed for the two most common cases, calling through a
function pointer and doing a dynamically dispatched method call.
- Someone brought forth the hypothesis that Rust's much coarser
compilation unit granularity might (partly) explain the difference of
PGO effectiveness compared to C/C++ [2] -- and indeed my experiments
seem to back this hypothesis up. When compiling Rust code for maximum
performance, one usually lets the compiler generate a single object
file per crate, which is equivalent to having a single object file per
static library in C/C++. With this setup, PGO was only able to achieve
an average 0.3% performance improvement in my benchmarks. However,
increasing the number of object files to (roughly) one per source file
led to an average performance improvement of 1.2%, that is, PGO made 4
times as much of a difference. Reducing ThinLTO's import-instr-limit
to 10 magnified the effect even more, making the PGO version about 4%
faster than the non-PGO version, which is well within the range of
improvement that one can expect from PGO. Interestingly, this last
configuration with the stricter import limit was the most performant
one, being also ~3% faster than the single compilation unit setup both
with and without PGO.
In conclusion, (1) there is no evidence that the implementation is
broken and (2) there are a number of cases and configurations that
demonstrate that PGO *can* make as much of a difference as can be
expected from it.
[1] https://github.com/rust-lang/cargo/issues/7416
[2] https://internals.rust-lang.org/t/profile-guided-optimization-how-well-does-it-work-for-you/11108/11
On Tue, Sep 24, 2019 at 5:15 PM Michael Woerister
Teresa Johnson via llvm-dev
Hiroshi Yamauchi via llvm-dev
Michael Woerister via llvm-dev
> Interesting. Does PGO mean PGO + ThinLTO here?
The compiler treats PGO orthogonally, so in those cases where ThinLTO is the default, PGO is combined with ThinLTO.