Suggesting a better system for defining Top Computer or performance

Tahoe
Tahoe
Joined: 9 Mar 05
Posts: 12
Credit: 23841020
RAC: 0
Topic 189315

The performance aspect of crunching can't be ignored. There are several crunchers who are "addicted" to seeing their machines chew through data faster than the cruncher next to them. There is a sense of anticipation as you creep towards the next cruncher in the list and pure elation when the website gives you their spot on the ranking list. The recent average credit formula is adequate for defining work completed in a group validation environment like BOINC, but it isn't ideal for defining performance. If it is too much work to readjust how the top computer is defined, maybe we can add a performance metric that measures how fast your computer crunches a "defined" work load. This work unit would only be used as a metric and the faster you crunch it, the faster your machine is. I have seen machines at the top of the computer list that appear to be an anomaly. When work is reported shouldn't be a factor. I could be mistaken but is seems that if a machine bunched work units together and was fortunate enough to have fast machines validate the units, it gets a higher RAC number. I recommend we define the top computers with a more objective performance assesment. Think of all the vapor phase systems that will spring up when the top spot is objectively earned. I know I'm considering it... One tangible benefit to having the top computer list based solely on performance metrics and not factoring in other computers, is that Large firms like, Sun, Intel, HP might throw some of their heavy hardware at this for some bragging rights.