jiffyrune / gpuocelot

Automatically exported from code.google.com/p/gpuocelot
0 stars 0 forks source link

Add an OpenCL front-end #17

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
Describe the New Feature:
1. Add an OpenCL front-end to ocelot.
2. Completely re-implement the OpenCL 1.0 specification using the open
source Nvidia Open64 compiler to generate PTX code.
3. Use as much of the existing Executive class as possible to implement
OpenCL functionality.

Which milestone does the feature belong to?
1.0.0

Which branch does the new feature go in?
Branch

Original issue reported on code.google.com by gregory....@gatech.edu on 5 Aug 2009 at 4:04

GoogleCodeExporter commented 9 years ago

Original comment by gregory....@gatech.edu on 5 Aug 2009 at 4:04

GoogleCodeExporter commented 9 years ago
Assigned this issue to benjamin.

Original comment by gregory....@gatech.edu on 9 Sep 2009 at 8:56

GoogleCodeExporter commented 9 years ago
Hi,
Version 2.1.1b of GPGPU-Sim has been uploaded. This version of GPGPU-
Sim includes several enhancements versus 2.1.0b including support for
OpenCL.
Perhaps you can reuse wrapper code..

Original comment by oscarbar...@gmail.com on 31 Oct 2009 at 10:17

GoogleCodeExporter commented 9 years ago
Hi,
I would like to use ocelot to create an openCL execution/emulation environment. 
Has there been any progress on this issue? If not, I will be willing to help 
create the openCL front end. 

Original comment by eugene.m...@gmail.com on 12 Jan 2011 at 4:56

GoogleCodeExporter commented 9 years ago
Unfortunately, I starting this but was unable to make much useful progress. My 
studies moved elsewhere and I forgot I was assigned this task. Anyone is free 
to pick up the slack.

Original comment by bnbeckw...@gmail.com on 24 Jan 2011 at 7:11

GoogleCodeExporter commented 9 years ago
Not a problem.  We've looked into this in detail and concluded that the amount 
of effort required would be greater than 1-2 people full time for a year.  In 
the meantime, OpenCL has become less interesting from our end as all of the 
functionality exists in CUDA, and at this point, CUDA applications seem to 
still dominate the market.

I'm going to leave this open in case anyone else is interested in picking it up.

Original comment by SolusStu...@gmail.com on 25 Jan 2011 at 12:52

GoogleCodeExporter commented 9 years ago
[deleted comment]
GoogleCodeExporter commented 9 years ago
Has anyone still work on adding an OpenCL frontend? I think it will be very 
useful since current OpenCL compilers are verdor specific, there is not an 
OpenCL compiler which can generate optimized code for NV GPU, AMD GPU and x86 
CPU. 
OpenCL is the future!

Original comment by juvesea...@gmail.com on 18 Apr 2011 at 1:50

GoogleCodeExporter commented 9 years ago
There is not currently anyone working on this.  We would love volunteers to 
take this up, particularly someone with experience developing compiler language 
frontends.  At this point though, it is a large engineering effort, and most of 
the contributors to Ocelot are PhD students who want to leverage the codebase 
for their research. CUDA serves that purpose well enough in most cases, and the 
lack of language features like templates and C++ support in OpenCL make it 
unlikely that any of the current Ocelot developers are going to spend time on 
this.  

Eventually I would like to move Ocelot towards a complete open source compiler 
for parallel processors (GPUs).  Specifically by moving beyond PTX to a new IR 
and adding in a language front-end and a generic framework for supporting 
additional backends.  We need people to step up and take ownership of these 
components though.

Original comment by gregory....@gatech.edu on 18 Apr 2011 at 3:52

GoogleCodeExporter commented 9 years ago
OpenCL frontend is coming! The ability of Clang/LLVM to parse OpenCL/C and 
generate PTX via the NVPTX code generator makes this more realistic. I am a 
research scientist working for Prof. Hyesoon Kim at Georgia Tech on an OpenCL 
variant called LCL ("Light" CL) for mobile and embedded platforms (sponsored by 
Samsung). Implementing basic OpenCL functionality via Ocelot is a 
stepping-stone for us towards LCL. We have a subset of the Platform and Runtime 
APIs working and are making efforts to extend the library of built-ins required 
by OpenCL/C available at LLVM.org which was initiated by Peter Colingbourne.

Original comment by phillip....@gmail.com on 8 Aug 2012 at 1:49

GoogleCodeExporter commented 9 years ago
Any progress on this?  It would be a way to implement OpenCL 1.2 and 2.0 
functionality on supporting NVidia hardware without waiting for NVidia to 
finally update its drivers.

Original comment by del...@gmail.com on 27 May 2014 at 1:37