Originally posted by l008com:
And scott, why do you hate RC5? I think its the best DC program.
I hate it because it is inherently worthless. They did rc5-56. rc6-64 had 256 times the keyspace. rc5-72 has 256 times the keyspace of rc5-64. So what is the point? We know it will take up to 256 times longer to find the answer. That is all that it can possibly prove. However, we already know that. Therfore, the project is totally pointless and has no merit. It serves as nothing more then a drain of resources. Granted, it has the best client, but the project itself is useless.
If I'm doing something on my computer, i wanna make sure it uses up all the CPU i am giving it.
What project doesn't use all of your CPU? Do some projects only use 50% of the processor, and top shows 50% idle? I don't think so. All DC projects, by their very nature, use all CPU. I would argue that other projects, ones that make sure of say, both the integer and floating point units, use more of the processor, since they are making more of it work. Altivec is just a small part of the processor, and the rest is being left idle.
And RC5 is the only one that does thanks to its "cores hand optimized in assembly" :-)
Actually, by that argument, folding at home would be great, since it also has hand optimized cores. RC5 is very easy to test, so they can make code for each OS. Other projects, which actually have something to do, need to make sure their code is accurate, which is far more important then optimizations. A single codebase is far easier to maintain and ensure accuracy.
Hand optimized code doesn't always mean it is that great. For a project that really does stuff, it is not very easy to do. Scientific integrity is a much higher priority, which means the codebase is not distributed. RC5 just happens to do well on the G4 because of one specific operation that happens to exist in Altivec.
Be aware that I am not saying you can't run this project, just that I feel that it is utterly lacking in merit. We have a finite amount of processing power, and I would rather see our team put it towards projects that have real scientific merit, or who's goal is otherwise determined to be worth of our attention, and that are not corporations trying to use the popularity of DC projects to get free processing time for their own profit.
I will at some point be proposing a criteria for evaluating projects, as I would like to see this team have a bit of focus in what we run.
Does that make my view more clear?
Top
