Talk:Introduction to HPC

From CSI 702

Based on the discussion in the lecture (and the presentation), the first lecture should be comprised of three topics: history of HPC, computing hardware and architecture, and HPC challenges/applications. Is there anything else that we're missing?

Stefan Novak 14:01, 24 January 2010 (EST)

I added a set of main topics to the page, based on the list I proposed in the main discussion page. It's a bit more granular than the three topics you listed above but I think it will make it easier for three of us to edit. We can consolidate sections later, if that makes sense. I added "Supercomputing challenges", based on your comment above. While the main topic refers to HPC, most of Dr. Wallin's slides refer to supercomputers/supercomputing instead of HPC so I recommend we do the same.

-thomas Tboggs 21:01, 24 January 2010 (EST)

Looks good! I'll start chipping away at it tomorrow. As you mentioned, I think that it would be best to compile all the notes on this page here so that's it's easier for the three of us to collaborate. Once we're done, we can then migrate the content to separate pages if need be. Thanks for outlining the content!

Stefan Novak 21:33, 24 January 2010 (EST)

Looks good, however, couple of comments. Do we really need the section 'Why Scientists use Computers'? The two new distributed supercomputing environments that are gaining ground are Cluster Computing and Grid Computing. Do you think we need separate sections for these two since these are not conventional supercomputers?

Any preferences on the distribution of topics for us to work on? Shall I take 4, 5 and 6?

--Ash 12:47, 25 January 2010 (EST)

I still think that we should openly edit any of the sections rather than explicitly assigning different sections to each person. This allows for cross-pollination of notes/ideas and falls a bit more in line with the purpose of the wiki. (Granted that it's more difficult to manage accountability, it would be pretty easy to see who contributed what through the "history" tab.)

If we go this route, you just have to remember to submit changes fairly frequently so there aren't any major conflicts submitted at the same time.

Stefan Novak 15:53, 25 January 2010 (EST)

I think Cluster Computing and Grid Computing should probably be two different subsections under Supercomputing Architectures.

-thomas Tboggs 17:44, 25 January 2010 (EST)

Nice update, Thomas. I think the "Project Performance Development" chart would go better the "Trends in Supercomputing" section. I'll move it down there for now but if you think it should be up near the top, feel free to switch it back.

Stefan Novak 13:16, 26 January 2010 (EST)

My thinking was that the Trends section would focus more on trends in terms of what types of architectures are/were used for particular purposes or at particular periods (historically speaking). But if we want to have that include trends in computing power, power consumption, etc., that's fine by me.

-thomas Tboggs 15:38, 26 January 2010 (EST)