llvm dev - Jun 2018 - [RFC] Abstract Parallel IR Optimizations

This is an RFC to add analyses and transformation passes into LLVM to
optimize programs based on an abstract notion of a parallel region.

  == this is _not_ a proposal to add a new encoding of parallelism =
We currently perform poorly when it comes to optimizations for parallel
codes. In fact, parallelizing your loops might actually prevent various
optimizations that would have been applied otherwise. One solution to
this problem is to teach the compiler about the semantics of the used
parallel representation. While this sounds tedious at first, it turns
out that we can perform key optimizations with reasonable implementation
effort (and thereby also reasonable maintenance costs). However, we have
various parallel representations that are already in use (KMPC,
GOMP, CILK runtime, ...) or proposed (Tapir, IntelPIR, ...). 

Our proposal seeks to introduce parallelism specific optimizations for
multiple representations while minimizing the implementation overhead.
This is done through an abstract notion of a parallel region which hides
the actual representation from the analysis and optimization passes. In
the schemata below, our current five optimizations (described in detail
here [0]) are shown on the left, the abstract parallel IR interface is
is in the middle, and the representation specific implementations is on
the right.

         Optimization          (A)nalysis/(T)ransformation         Impl.
   ---------------------------------------------------------------------------
     CodePlacementOpt \  /---> ParallelRegionInfo (A) ---------|->
KMPCImpl (A)
       RegionExpander -\ |                                     |   GOMPImpl (A)
   AttributeAnnotator -|-|---> ParallelCommunicationInfo (A) --/   ...
   BarrierElimination -/ |
VariablePrivatization /  \---> ParallelIR/Builder (T) ----------->
KMPCImpl (T)


In our setting, a parallel region can be an outlined function called
through a runtime library but also a fork-join/attach-reattach region
embedded in an otherwise sequential code. The new optimizations will
provide parallelism specific optimizations to all of them (if
applicable). There are various reasons why we believe this is a
worthwhile effort that belongs into the LLVM codebase, including:

  1) We improve the performance of parallel programs, today.
  2) It serves as a meaningful baseline for future discussions on
     (optimized) parallel representations.
  3) It allows to determine the pros and cons of the different schemes
     when it comes to actual optimizations and inputs.
  4) It helps to identify problems that might arise once we start to
     transform parallel programs but _before_ we commit to a specific
     representation.

Our prototypes for the OpenMP KMPC library (used by clang) already shows
significant speedups for various benchmarks [0]. It also exposed a (to
me) prior unknown problem between restrict/noalias pointers and
(potential) barriers (see Section 3 in [0]).

We are currently in the process of cleaning the code, extending the
support for OpenMP constructs and adding a second implementation for a
embedded parallel regions. Though, a first horizontal prototype
implementation is already available for review [1].

Inputs of any kind are welcome and reviewers are needed!

Cheers,
  Johannes


[0] http://compilers.cs.uni-saarland.de/people/doerfert/par_opt18.pdf
[1] https://reviews.llvm.org/D47300


P.S.
  Sorry if you received this message multiple times!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 228 bytes
Desc: Digital signature
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20180607/55cab9f6/attachment.sig>

Hi Johannes,

apologies in advance if the questions following are silly or don't
make sense. I lack a bit of context here and I'm not sure to fully
understand your proposal.

Currently clang (and flang) are lowering OpenMP when building LLVM IR
(this is because LLVM IR can't express the parallel/concurrent
concepts of OpenMP so they have to be lowered first). So, can I assume
that your proposal starts off in a context where that lowering is not
happening anymore in the front end but it'd happen later in a LLVM IR
pass? If so, then you'd be assuming that there is already a way of
representing OpenMP constructs in the LLVM IR, is my understanding
correct here? I think that the Intel proposal [1] could be one way
(not necessarily the one) to do this (disregarding the fact that it is
tailored for OpenMP), does this still make sense?

If this is the case, and given that you explicitly state that this is
not a Parallel IR of any sort, is your suggestion to improve
optimisation of OpenMP code, based on a "side-car"/ancillary
representation built on top of the existing IR, which as I understand
should already be able to represent OpenMP? But then this looks a bit
redundant to me. So I'm pretty sure one of my assumptions is
incorrect. Unless your auxiliar representation is more an alternative
to the W-regions [1].

Or, maybe I am completely wrong here: you didn't say anything about
the FE lowering, which would still happen, and then your proposal
builds on top of that. I don't think you meant that, given that your
proposal mentions KMP and GOMP (and the current lowering done by clang
targets only KMP).

Thank you very much,
Roger

[1] https://dl.acm.org/citation.cfm?id=3148191

2018-06-07 12:25 GMT+02:00 Johannes Doerfert via llvm-dev
<llvm-dev at lists.llvm.org>:
> This is an RFC to add analyses and transformation passes into LLVM to
> optimize programs based on an abstract notion of a parallel region.
>
>   == this is _not_ a proposal to add a new encoding of parallelism =>
> We currently perform poorly when it comes to optimizations for parallel
> codes. In fact, parallelizing your loops might actually prevent various
> optimizations that would have been applied otherwise. One solution to
> this problem is to teach the compiler about the semantics of the used
> parallel representation. While this sounds tedious at first, it turns
> out that we can perform key optimizations with reasonable implementation
> effort (and thereby also reasonable maintenance costs). However, we have
> various parallel representations that are already in use (KMPC,
> GOMP, CILK runtime, ...) or proposed (Tapir, IntelPIR, ...).
>
> Our proposal seeks to introduce parallelism specific optimizations for
> multiple representations while minimizing the implementation overhead.
> This is done through an abstract notion of a parallel region which hides
> the actual representation from the analysis and optimization passes. In
> the schemata below, our current five optimizations (described in detail
> here [0]) are shown on the left, the abstract parallel IR interface is
> is in the middle, and the representation specific implementations is on
> the right.
>
>          Optimization          (A)nalysis/(T)ransformation         Impl.
>   
---------------------------------------------------------------------------
>      CodePlacementOpt \  /---> ParallelRegionInfo (A) ---------|->
KMPCImpl (A)
>        RegionExpander -\ |                                     |   GOMPImpl
(A)
>    AttributeAnnotator -|-|---> ParallelCommunicationInfo (A) --/   ...
>    BarrierElimination -/ |
> VariablePrivatization /  \---> ParallelIR/Builder (T) ----------->
KMPCImpl (T)
>
>
> In our setting, a parallel region can be an outlined function called
> through a runtime library but also a fork-join/attach-reattach region
> embedded in an otherwise sequential code. The new optimizations will
> provide parallelism specific optimizations to all of them (if
> applicable). There are various reasons why we believe this is a
> worthwhile effort that belongs into the LLVM codebase, including:
>
>   1) We improve the performance of parallel programs, today.
>   2) It serves as a meaningful baseline for future discussions on
>      (optimized) parallel representations.
>   3) It allows to determine the pros and cons of the different schemes
>      when it comes to actual optimizations and inputs.
>   4) It helps to identify problems that might arise once we start to
>      transform parallel programs but _before_ we commit to a specific
>      representation.
>
> Our prototypes for the OpenMP KMPC library (used by clang) already shows
> significant speedups for various benchmarks [0]. It also exposed a (to
> me) prior unknown problem between restrict/noalias pointers and
> (potential) barriers (see Section 3 in [0]).
>
> We are currently in the process of cleaning the code, extending the
> support for OpenMP constructs and adding a second implementation for a
> embedded parallel regions. Though, a first horizontal prototype
> implementation is already available for review [1].
>
> Inputs of any kind are welcome and reviewers are needed!
>
> Cheers,
>   Johannes
>
>
> [0] http://compilers.cs.uni-saarland.de/people/doerfert/par_opt18.pdf
> [1] https://reviews.llvm.org/D47300
>
>
> P.S.
>   Sorry if you received this message multiple times!
>
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>


-- 
Roger Ferrer Ibáñez

Hi Roger,

On 06/12, Roger Ferrer Ibáñez wrote:
> apologies in advance if the questions following are silly or don't
> make sense. I lack a bit of context here and I'm not sure to fully
> understand your proposal.
No worries, I'm glad if people ask questions!
> Currently clang (and flang) are lowering OpenMP when building LLVM IR
> (this is because LLVM IR can't express the parallel/concurrent
> concepts of OpenMP so they have to be lowered first). So, can I assume
> that your proposal starts off in a context where that lowering is not
> happening anymore in the front end but it'd happen later in a LLVM IR
> pass? If so, then you'd be assuming that there is already a way of
> representing OpenMP constructs in the LLVM IR, is my understanding
> correct here? I think that the Intel proposal [1] could be one way
> (not necessarily the one) to do this (disregarding the fact that it is
> tailored for OpenMP), does this still make sense?
My proposal does _not_ assume we change clang in any way, though it does
also not require it. However, the initial patch [1] will only work with
the OpenMP lowering used by clang right now.

The idea is as follows:

  We have different representation of parallelism in the IR, for example
  the KMP runtime library calls emitted by clang or the Intel parallel
  IR you mentioned. For each of them we write a piece of code that (1)
  extracts domain specific information and (2) allows to modify the
  parallel representation. This is the only piece of code that has to be
  adapted for each parallel representation we want to optimize. On top
  of this are abstract interfaces that expose the information and
  modification options to parallel optimization passes. The patch [1]
  only contains the attribute annotator but we have more as explained in
  the paper [0]. The analysis/optimization logic is part of these passes
  and not aware of the underlying representation. We can consequently
  use the same passes to optimize code that was lowered to use different
  parallel runtime libraries (GOMP, KMP, Cilk runtime, TBB, ...) or into
  a native parallel IR (of any shape). This is especially useful as the
  native parallel IR might not always be usable. If that happens we have
  to fallback to early outlining, thus runtime library calls emitted by
  the front-end. Even if we at some point have a native parallel
  representation that is always used, we can simply remove the
  abstraction introduced by this approach but keep the
  analysis/optimizations around.

[0] http://compilers.cs.uni-saarland.de/people/doerfert/par_opt18.pdf
[1] https://reviews.llvm.org/D47300
> If this is the case, and given that you explicitly state that this is
> not a Parallel IR of any sort, is your suggestion to improve
> optimisation of OpenMP code, based on a "side-car"/ancillary
> representation built on top of the existing IR, which as I understand
> should already be able to represent OpenMP? But then this looks a bit
> redundant to me. So I'm pretty sure one of my assumptions is
> incorrect. Unless your auxiliar representation is more an alternative
> to the W-regions [1].
> 
> Or, maybe I am completely wrong here: you didn't say anything about
> the FE lowering, which would still happen, and then your proposal
> builds on top of that. I don't think you meant that, given that your
> proposal mentions KMP and GOMP (and the current lowering done by clang
> targets only KMP).
I'm not sure if these paragraphs are still relevant. Does the above
"explanation" answers you questions already? If not, please continue
asking!

Cheers,
  Johannes
> Thank you very much,
> Roger
> 
> [1] https://dl.acm.org/citation.cfm?id=3148191
> 
> 2018-06-07 12:25 GMT+02:00 Johannes Doerfert via llvm-dev
> <llvm-dev at lists.llvm.org>:
> > This is an RFC to add analyses and transformation passes into LLVM to
> > optimize programs based on an abstract notion of a parallel region.
> >
> >   == this is _not_ a proposal to add a new encoding of parallelism
=> >
> > We currently perform poorly when it comes to optimizations for
parallel
> > codes. In fact, parallelizing your loops might actually prevent
various
> > optimizations that would have been applied otherwise. One solution to
> > this problem is to teach the compiler about the semantics of the used
> > parallel representation. While this sounds tedious at first, it turns
> > out that we can perform key optimizations with reasonable
implementation
> > effort (and thereby also reasonable maintenance costs). However, we
have
> > various parallel representations that are already in use (KMPC,
> > GOMP, CILK runtime, ...) or proposed (Tapir, IntelPIR, ...).
> >
> > Our proposal seeks to introduce parallelism specific optimizations for
> > multiple representations while minimizing the implementation overhead.
> > This is done through an abstract notion of a parallel region which
hides
> > the actual representation from the analysis and optimization passes.
In
> > the schemata below, our current five optimizations (described in
detail
> > here [0]) are shown on the left, the abstract parallel IR interface is
> > is in the middle, and the representation specific implementations is
on
> > the right.
> >
> >          Optimization          (A)nalysis/(T)ransformation        
Impl.
> >   
---------------------------------------------------------------------------
> >      CodePlacementOpt \  /---> ParallelRegionInfo (A)
---------|-> KMPCImpl (A)
> >        RegionExpander -\ |                                     |  
GOMPImpl (A)
> >    AttributeAnnotator -|-|---> ParallelCommunicationInfo (A) --/  
...
> >    BarrierElimination -/ |
> > VariablePrivatization /  \---> ParallelIR/Builder (T)
-----------> KMPCImpl (T)
> >
> >
> > In our setting, a parallel region can be an outlined function called
> > through a runtime library but also a fork-join/attach-reattach region
> > embedded in an otherwise sequential code. The new optimizations will
> > provide parallelism specific optimizations to all of them (if
> > applicable). There are various reasons why we believe this is a
> > worthwhile effort that belongs into the LLVM codebase, including:
> >
> >   1) We improve the performance of parallel programs, today.
> >   2) It serves as a meaningful baseline for future discussions on
> >      (optimized) parallel representations.
> >   3) It allows to determine the pros and cons of the different schemes
> >      when it comes to actual optimizations and inputs.
> >   4) It helps to identify problems that might arise once we start to
> >      transform parallel programs but _before_ we commit to a specific
> >      representation.
> >
> > Our prototypes for the OpenMP KMPC library (used by clang) already
shows
> > significant speedups for various benchmarks [0]. It also exposed a (to
> > me) prior unknown problem between restrict/noalias pointers and
> > (potential) barriers (see Section 3 in [0]).
> >
> > We are currently in the process of cleaning the code, extending the
> > support for OpenMP constructs and adding a second implementation for a
> > embedded parallel regions. Though, a first horizontal prototype
> > implementation is already available for review [1].
> >
> > Inputs of any kind are welcome and reviewers are needed!
> >
> > Cheers,
> >   Johannes
> >
> >
> > [0] http://compilers.cs.uni-saarland.de/people/doerfert/par_opt18.pdf
> > [1] https://reviews.llvm.org/D47300
> >
> >
> > P.S.
> >   Sorry if you received this message multiple times!
> >
> > _______________________________________________
> > LLVM Developers mailing list
> > llvm-dev at lists.llvm.org
> > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
> >
> 
> 
> 
> -- 
> Roger Ferrer Ibáñez
-- 

Johannes Doerfert
PhD Student / Researcher

Compiler Design Lab (Professor Hack) / Argonne National Laboratory
Saarland Informatics Campus, Germany / Lemont, IL 60439, USA
Building E1.3, Room 4.31

Tel. +49 (0)681 302-57521 : doerfert at cs.uni-saarland.de / jdoerfert at
anl.gov
Fax. +49 (0)681 302-3065  : http://www.cdl.uni-saarland.de/people/doerfert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 228 bytes
Desc: Digital signature
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20180612/b6824780/attachment.sig>