ok, i need to get this sorted once and for all! the other day i installed F@H on a dual core system running windows, according to task manager only 50% of the CPUs resources where being used (roughly half of each core, based on the graph). How can i get F@H running at 100% on both cores without having to run two instances? also, the same is happening on my machine (kubuntu)... i run this machine 24/7 so getting this sorted will make a big difference!
There is a SMP client for Linux (one that will run on both cores) but it is only available as a AMD64 binary. Im having to run 2 instances! Should F@H really be closed source?
um, if you think about it, you'll end up getting the same score, whether you get a unit done faster or you get two units done slower, it has the same result. Also with information as sensitive and all that as this, yes, i believe it should most definitely be closed source.
security through obscurity is no security at all... i would argue the opposite point, with data as important as this we should be able to verify that its being done correctly and be able to suggest improvements. if the whole world can see your code, then both sides can see it (good and bad if you will...). If only you can see your code, then most likely only the bad will be motivated enough to reverse engineer it. also, the faster you return work units the more points you get for them (i think)
on the second part about the score thats true, but what i'm saying is that you get on work unit in with a high score, or you get two work units in that are cumulatively greater than the single score only slower. So by the time your on your third or fourth work unit it would probably work out to be exactly the same ether way. Granted that is just theory.
Swansen, only 50% of his total CPU capacity was being used. If he runs 2 instances, or uses an SMP client he'll be using close to 100% capacity. I'm sure the scored will account for the time and computational complexity of the WUs. I don't see what is so sensitive about the information F@H sends back. Oh noes...some bad person will find out the result of a particular protein folding sequence. He world is in his hands....
Lol, Addis hates me, well i didn't realize he was only running one instance, not two. Anyways its their work that they have put countless hours into and i can understand, they have a lot of pride it what they are doing. They are doing all of this for no money, so it makes sense to me that they wouldn't want to give out everything they worked so hard for, but thats just my opinion. Secondly, some one could alter the code, or do something funny and they would have to back track, and it is very sensitive, in a way. How proteins fold explains many diseases and some cancers. So if any little thing is off they loose all their credibility, or release some false results that do more harm that good.
What? There is something seriously wrong with that statement but i dont even know where to start... You can do that without the source code, windows is an amazing example of that... most OSS kicks the :swear: out of proprietry software! (It doesnt have to be under a GPL style lisence, we dont need to be able to modify the SC!)
Nah I don't have anything against you Swansen. At most, I have a hard time understanding your posts, so you'll excuse me if I come across in a less than friendly manner. When the results of the WU are sent back, I think they are validated with identical WUs to check they're ok. So if anyone did send back bogus results, it would be dropped.
You cant trust a machine that you don't control, so yes... the results must be validated somehow (multiple participants processing the same WU, or there is a quick and dirty way to verify the result).
