Use tools, don't be one

One of the most important things that can be done when looking at performance is to analyze the workload with tools. Oftentimes, we have an idea of what the bottleneck is but frequently, we are wrong. The only way we can tell is through trial and error fixes. Even then, making a change and seeing the expected improvement does necessarily mean our guess was right. Towards that end, I wrote an article about how I used tools to look at Ticker Tape and find and fix problems. Many guides and documentation for various tools provide instruction and examples of how to gather information but leave off at that point. What comes next? How do I interpret that info? I hoped to try and provide some examples of how to do that; to go past the information gathering stage into the interpretation and finally application of a solution. The article is available on the web on the tickertape homepage. Although the examples presented use Ticker Tape as the workload they should be applicable to any program.
Reportez-vous à notre Notice d'optimisation pour plus d'informations sur les choix et l'optimisation des performances dans les produits logiciels Intel.