GPU Programming in Toronto Message Board › Future Meetup Ideas

Future Meetup Ideas

Rick E.
RickEyre
Toronto, ON
Post #: 1
Hi all,

Chris mentioned that he'd like us all to post the ideas we put so I thought I'd start this thread.

My idea was to possibly collaborate on a project in the future when we know more about GPUs and the problems/needs of the industry.
Adrian M.
user 12787712
Toronto, ON
Post #: 5
Maybe everyone can post ideas of a problems they would like to solve.

I'd like to see if there are any application for big data analysis.

Hi all,

Chris mentioned that he'd like us all to post the ideas we put so I thought I'd start this thread.

My idea was to possibly collaborate on a project in the future when we know more about GPUs and the problems/needs of the industry.

Brendan
hovercraft
Toronto, ON
Post #: 76
How about a study group of some sort? This could be separate from the main events, which I think should be more social.
Chris S.
user 70172432
Toronto, ON
Post #: 3
Good idea! Any suggestions? The groups might be small to start with. Ideas that come to mind include, ask Greg to talk about his own work on FPGAs; ask someone to talk about OpenMP; groups on specific software topics; we could build a few groups around the individual interests of our members or new members.
Brendan
hovercraft
Toronto, ON
Post #: 80
I'm the organizer of another programming group (NSCoder Toronto) and we are starting a book club at the end of the month. We picked a book on a topic (Learning Cocos2D) that seemed of interest to a lot of members and went from there.

I'd be interested in an intro to OpenCL or OpenGL type course myself.
Greg
user 51378412
Toronto, ON
Post #: 2
Perhaps some hands on sessions where everyone brings a CUDA/OpenCL capable GPU equipped laptop or desktop.

Not everyone has such a machine so we can work in teams.

Here are some thoughts:

1. an intro "hello world" session on CUDA/OpenCL,
install SDK, run manufactures demos, write and test "hello world example
(Linux or Microsoft)

2. On Nvidia platforms, after doing the above with CUDA, do it again with the OpenCL SDK

3. network everyone's machine together to build an MPI GPU based cluster farm
(trivial under Linux, many permissions/firewall issues to be overcome with Microsoft)

4. CUDA/OpenCL interoperability with OpenGL examples

5. solve a GPU problem proposed by the group.

6. Demonstrate the CUDA to OpenCL swan and cu2cl conversion tools
Chris S.
user 70172432
Toronto, ON
Post #: 5
Excellent suggestions.

How about starting separate session sub-groups: one on MPI, one on CUDA, one on OpenCL.

I find getting started with CUDA quite straightforward. I think that OpenCL is easier to learn once the CUDA terminology is familiar. I think that MPI could be started independently at any time.

A former member
Post #: 2
Perhaps some hands on sessions where everyone brings a CUDA/OpenCL capable GPU equipped laptop or desktop.

Not everyone has such a machine so we can work in teams.

Here are some thoughts:

1. an intro "hello world" session on CUDA/OpenCL,
install SDK, run manufactures demos, write and test "hello world example
(Linux or Microsoft)

2. On Nvidia platforms, after doing the above with CUDA, do it again with the OpenCL SDK

3. network everyone's machine together to build an MPI GPU based cluster farm
(trivial under Linux, many permissions/firewall issues to be overcome with Microsoft)

4. CUDA/OpenCL interoperability with OpenGL examples

5. solve a GPU problem proposed by the group.

6. Demonstrate the CUDA to OpenCL swan and cu2cl conversion tools

Rather than expect everyone to have a CUDA enabled laptop, how about we set up a remote server with CUDA capabilities? Account names/passwords could be issued to workshop attendees. Example code could be set up in advance.
This way people could come to the workshop with a cheap netbook with an ssh client installed. That's all they would need.
At the end of the workshop, users could zip up their examples and email the archive to themselves and the owner of the server could delete all the guest accounts.
Adrian M.
user 12787712
Toronto, ON
Post #: 6
Perhaps some hands on sessions where everyone brings a CUDA/OpenCL capable GPU equipped laptop or desktop.

Not everyone has such a machine so we can work in teams.

Here are some thoughts:

1. an intro "hello world" session on CUDA/OpenCL,
install SDK, run manufactures demos, write and test "hello world example
(Linux or Microsoft)

2. On Nvidia platforms, after doing the above with CUDA, do it again with the OpenCL SDK

3. network everyone's machine together to build an MPI GPU based cluster farm
(trivial under Linux, many permissions/firewall issues to be overcome with Microsoft)

4. CUDA/OpenCL interoperability with OpenGL examples

5. solve a GPU problem proposed by the group.

6. Demonstrate the CUDA to OpenCL swan and cu2cl conversion tools

I'd be interested in this.
Chris S.
user 70172432
Toronto, ON
Post #: 6
How about study sessions on specific classes of parallel algorithms - one day sorting - one day searching - one day merging - etc. Two volunteers for each session who would prepare the material. Different volunteers each time. A starting point could be http://www.cs.cmu.edu...­
Powered by mvnForum

Our Sponsors

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy