Performance, Analysis, Threading and Intel Developer Forum

Performance, Analysis, Threading and Intel Developer Forum - What do all these have in common? Keep reading and I will answer this in my blog.

First let me introduce myself. I joined Intel back in 1992. In my years at Intel I have always been interested in parallel programming and parallel computing. My first position at Intel was in the Supercomputing Systems Division . My work in those early days on parallel programming was based on message passing. Even though my emphasis back then was message passing, I was still working with threads. I created one of the first classes in the Supercomputer Systems Division on Posix Threads (some of those coming to my class thought the p in pthreads stood for parallel threads rather than Posix threads). Now most of my focus is on threading, not on message passing. The basic concepts of parallelism are the same. After working in the Supercomputer systems division I joined Intel's ISV performance lab and worked with ISVs to tune software for Intel platforms. About 2002, I transferred to the Software Developers Division to work with the Threading Analysis tools. I first participated in IDF back when it was held in Palm Springs, I taught labs and lectures at Intel Developer Forum many times. Today I want to write about the Intel Developer Forum coming up in August in San Francisco. I noticed that when we teach software labs at IDF we receive very high marks. I also noticed that when we run a 3 hour instructor led lab some participants could move through the lab faster than others, but we need to keep everyone at the same stage. I also noticed that we can seat a lot more people in the lectures. So this year I suggested a new idea. I suggested that we offer lectures to cover the basic information and then offer an open lab as a follow on to the lecture. The IDF team liked it and so did the Software products development group. So at IDF next week there will be 3 lectures on Intel Software Development products and an open lab that corresponds to the lectures. In the open lab you can come in and work at your own pace - as fast or as slow as you want. Each system in the lab will have programming exercises on them that reinforce the ideas taught in the lectures and allow you to try out Intel Software Development engineers. We will have Intel software engineers on hand to answer any of your questions about our software products.

The lectures and open labs will be taught twice - Wednesday 20 August and Thursday 21 August. I will personally be in the open lab on Thursday. So come on by and say hello and let me know what you think of this new format. If you like it we may repeat it again in future forums. If we don't get good feedback, we will be looking for new ideas. If you already use Intel Software products or have evaluated them and want to tell me of your experiences I look forward to hearing from you. For those carefully planning out their schedules, the Software Development Products open lab is SFTL001. The three lectures about Intel Software Development Products are SFT006, SFT007, SFT008. So now I answer the question I asked at the top of the blog. These sessions at IDF will cover performance, analysis and threading.

I will begin blogging on a regular basis. I now manage a team of technical consulting engineers in the Performance, Analysis and Threading Lab - so yep - you guessed it, most of my blogs will probably be about performance, analysis or threading. In addition to teaching at IDF I conducted a few web seminars and sometimes write a paper now and then. Some of you may have attended one of my web seminars. I will list a few of the links here (a couple of my webcasts are archived and still available for replay) - There is one here on threading a sequential application. Another web seminar by me here is on maintaining correctness while developing threaded software: A paper I wrote earlier this year was on using libraries.

I hope to meet some of you Thursday 21 August at Intel Developer Forum.
-David Mackay

Para obter informações mais completas sobre otimizações do compilador, consulte nosso aviso de otimização.