llvm dev - Oct 2013 - [LLVMdev] Open work items with a small scope

Dear Clang/LLVM community!

We (a group of 3 highly motivated CS grad students) are taking a class
about compilers and optimizations.
This class offers the option of doing a implementation project (instead of
reading/writing x papers) for its main part.
Of course hacking is much more fun than ... ;)

So my question is do you know of any open points, compiler passes,
optimizations in the LLVM / Clang realm that are small enough for a student
group to be tackled in a quarter (~2 months)?
We acknowledge that compilers are very complex software and that it is very
hard to do anything meaningful in this timeframe.
So we would appreciate any guidance you can offer.
This could also mean doing work in a LLVM / Clang related project, as LLVM
/ Clang themselves are pretty mature projects and doing anything
substantial would probably take much longer.
The only requirement of the project would be that it is in the realm of
compilers and optimizations.

Some areas/ideas we are especially interested in:
Adaption of Clang to support OpenMP (http://clang-omp.github.io/), but
haven't been able to contact one of the developers yet.

Profile-guided optimization:
https://docs.google.com/document/d/1b2XFuOkR2K-Oao4u5fR3a9Ok83IB_W4EJWVmNak4GRE/pub

We are glad for any pointers, suggestions and other help.
Also feel free to answer me directly if you like.
-- Julian
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20131006/d7073f93/attachment.html>

There's an open project to rewrite our GVN (and it seems like there's
already a pretty good consensus for the direction to take it). See this
thread <http://thread.gmane.org/gmane.comp.compilers.llvm.cvs/158097/>.

-- Sean Silva


On Sun, Oct 6, 2013 at 10:12 PM, Julian Lettner <julian.lettner at
uci.edu>wrote:
> Dear Clang/LLVM community!
>
> We (a group of 3 highly motivated CS grad students) are taking a class
> about compilers and optimizations.
> This class offers the option of doing a implementation project (instead of
> reading/writing x papers) for its main part.
> Of course hacking is much more fun than ... ;)
>
> So my question is do you know of any open points, compiler passes,
> optimizations in the LLVM / Clang realm that are small enough for a student
> group to be tackled in a quarter (~2 months)?
> We acknowledge that compilers are very complex software and that it is
> very hard to do anything meaningful in this timeframe.
> So we would appreciate any guidance you can offer.
> This could also mean doing work in a LLVM / Clang related project, as LLVM
> / Clang themselves are pretty mature projects and doing anything
> substantial would probably take much longer.
> The only requirement of the project would be that it is in the realm of
> compilers and optimizations.
>
> Some areas/ideas we are especially interested in:
> Adaption of Clang to support OpenMP (http://clang-omp.github.io/), but
> haven't been able to contact one of the developers yet.
>
> Profile-guided optimization:
>
>
https://docs.google.com/document/d/1b2XFuOkR2K-Oao4u5fR3a9Ok83IB_W4EJWVmNak4GRE/pub
>
> We are glad for any pointers, suggestions and other help.
> Also feel free to answer me directly if you like.
> -- Julian
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20131006/4e6e88bf/attachment.html>

On Sun, Oct 6, 2013 at 7:12 PM, Julian Lettner <julian.lettner at uci.edu>
wrote:
> Dear Clang/LLVM community!
>
> We (a group of 3 highly motivated CS grad students) are taking a class
about
> compilers and optimizations.
> This class offers the option of doing a implementation project (instead of
> reading/writing x papers) for its main part.
> Of course hacking is much more fun than ... ;)
>
> So my question is do you know of any open points, compiler passes,
> optimizations in the LLVM / Clang realm that are small enough for a student
> group to be tackled in a quarter (~2 months)?
If you can choose a frontend-related project, then you might want to
consider implementing a refactoring tool.  A few ideas are listed
here:

http://clang.llvm.org/docs/ClangTools.html

Dmitri

-- 
main(i,j){for(i=2;;i++){for(j=2;j<i;j++){if(!(i%j)){j=0;break;}}if
(j){printf("%d\n",i);}}} /*Dmitri Gribenko <gribozavr at
gmail.com>*/

On Sun, Oct 06, 2013 at 07:12:54PM -0700, Julian Lettner
wrote:
> So my question is do you know of any open points, compiler passes,
> optimizations in the LLVM / Clang realm that are small enough for a student
> group to be tackled in a quarter (~2 months)?
One of the long term points of interest for NetBSD would be a M68000
backend for LLVM. This is the CPU family used by the original Amiga and
Atari computers. It is still alive in various forms in the embedded
space. A possible project for your time frame would be the MC specific
parts -- TD descriptions of the instruction set, assembler parsing and
disassembly. I don't know if that fits your area though.

Joerg

Hi Julian,
> Some areas/ideas we are especially interested in:
> Adaption of Clang to support OpenMP (http://clang-omp.github.io/), but
haven't been able to contact one of the developers yet.

This project is owned by my team, and we certainly welcome contributions!
Have you tried to contact us via https://github.com/clang-omp/clang/issues ?

Either way, it is really hard to find a short-term "new development"
work
for your team in OpenMP -- as the implementation is more or less already
finished. Yes, it can benefit from some refactoring here and there, but I
doubt this is what you are looking for.

One area still missing, though, is capturing variables with variable-length
array type. There is a similar question (is it posted by one of you?) and
more elaborate answer: https://github.com/clang-omp/clang/issues/12

Yours,
Andrey Bokhanko
============Software Engineer
Intel Compiler Team


On Mon, Oct 7, 2013 at 6:12 AM, Julian Lettner <julian.lettner at
uci.edu>wrote:
> Dear Clang/LLVM community!
>
> We (a group of 3 highly motivated CS grad students) are taking a class
> about compilers and optimizations.
> This class offers the option of doing a implementation project (instead of
> reading/writing x papers) for its main part.
> Of course hacking is much more fun than ... ;)
>
> So my question is do you know of any open points, compiler passes,
> optimizations in the LLVM / Clang realm that are small enough for a student
> group to be tackled in a quarter (~2 months)?
> We acknowledge that compilers are very complex software and that it is
> very hard to do anything meaningful in this timeframe.
> So we would appreciate any guidance you can offer.
> This could also mean doing work in a LLVM / Clang related project, as LLVM
> / Clang themselves are pretty mature projects and doing anything
> substantial would probably take much longer.
> The only requirement of the project would be that it is in the realm of
> compilers and optimizations.
>
> Some areas/ideas we are especially interested in:
> Adaption of Clang to support OpenMP (http://clang-omp.github.io/), but
> haven't been able to contact one of the developers yet.
>
> Profile-guided optimization:
>
>
https://docs.google.com/document/d/1b2XFuOkR2K-Oao4u5fR3a9Ok83IB_W4EJWVmNak4GRE/pub
>
> We are glad for any pointers, suggestions and other help.
> Also feel free to answer me directly if you like.
> -- Julian
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20131008/d335e101/attachment.html>

If you're interested in a refactoring tool then clang-modernize might be up
your alley. There's a nice list of small things to whet your appetite at
http://cpp11-migrate.atlassian.net/. You're completely free to build your
own transform that we haven't thought of yet though. There are lots of ideas
out there and if you're keen on the C++ language and its new features it
would probably be a lot of fun.

From: cfe-dev-bounces at cs.uiuc.edu [mailto:cfe-dev-bounces at cs.uiuc.edu] On
Behalf Of Andrey Bokhanko
Sent: Tuesday, October 8, 2013 8:30 AM
To: Julian Lettner
Cc: cfe-dev; LLVM Developers Mailing List
Subject: Re: [cfe-dev] [LLVMdev] Open work items with a small scope

Hi Julian,
> Some areas/ideas we are especially interested in:
> Adaption of Clang to support OpenMP (http://clang-omp.github.io/), but
haven't been able to contact one of the developers yet.
This project is owned by my team, and we certainly welcome contributions! Have
you tried to contact us via https://github.com/clang-omp/clang/issues ?
Either way, it is really hard to find a short-term "new development"
work for your team in OpenMP -- as the implementation is more or less already
finished. Yes, it can benefit from some refactoring here and there, but I doubt
this is what you are looking for.

One area still missing, though, is capturing variables with variable-length
array type. There is a similar question (is it posted by one of you?) and more
elaborate answer: https://github.com/clang-omp/clang/issues/12

Yours,
Andrey Bokhanko
============Software Engineer
Intel Compiler Team

On Mon, Oct 7, 2013 at 6:12 AM, Julian Lettner <julian.lettner at
uci.edu<mailto:julian.lettner at uci.edu>> wrote:
Dear Clang/LLVM community!

We (a group of 3 highly motivated CS grad students) are taking a class about
compilers and optimizations.
This class offers the option of doing a implementation project (instead of
reading/writing x papers) for its main part.
Of course hacking is much more fun than ... ;)

So my question is do you know of any open points, compiler passes, optimizations
in the LLVM / Clang realm that are small enough for a student group to be
tackled in a quarter (~2 months)?
We acknowledge that compilers are very complex software and that it is very hard
to do anything meaningful in this timeframe.
So we would appreciate any guidance you can offer.
This could also mean doing work in a LLVM / Clang related project, as LLVM /
Clang themselves are pretty mature projects and doing anything substantial would
probably take much longer.
The only requirement of the project would be that it is in the realm of
compilers and optimizations.

Some areas/ideas we are especially interested in:
Adaption of Clang to support OpenMP (http://clang-omp.github.io/), but
haven't been able to contact one of the developers yet.

Profile-guided optimization:
https://docs.google.com/document/d/1b2XFuOkR2K-Oao4u5fR3a9Ok83IB_W4EJWVmNak4GRE/pub

We are glad for any pointers, suggestions and other help.
Also feel free to answer me directly if you like.
-- Julian

_______________________________________________
LLVM Developers mailing list
LLVMdev at cs.uiuc.edu<mailto:LLVMdev at cs.uiuc.edu>        
http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20131008/6a9ac19c/attachment.html>