Aye congratulations Lasis

and welcome to the new dude

/poke hey guys! looks like things been goin pretty good here. ^^ Work has got me bouncin around all over the country the last few months, and will probably continue that way.

toasted a motherboard on the amd 3400, got that back up and going. looking to replace that whole setup with a quad core SLi setup. Been playing the new Age of Conan game, and the graphics are killing my machine. hehe

the old celeron 667 is offline, probably replace with a cheap dual core barebones.

the celeron D in my brothers compy going to get a dual core upgrade soon, and his fiances comp with the dual p3 i'll change to my current amd 3400, and keep it around for file serving or something. those p3's seem to love the dddt. :)

but anyways, i ordered a laptop to take on the road with me, i don't expect it will be much of a cruncher, but, perfect for keeping connected :D

i'll pop in when i can, keep up the great work guys! ^^

Hi all!

So far, which among the projects have the best number of points per hour?

this is a hard to give a strait answer on this, but the general rule is the longer it takes to do the work unit, the more CPU power. This will pump the most and eliminate most platform dependencies. This means slower machines do best on DDDT and HCC. Multi-core machines do best on FAAH and HPF. Rice is new, so may have points problems while the project is breaking in, but will be in the FAAD an HPF category eventually.

Try this for awhile, see how it goes.

Sek,

Nice graph, but no score. To understand the problem more fully, you must look at the max/min values of each group in relationship to its platform. This means that each individual system's stats can vary some degree from average. What your graph shows well is the relative probability any unknown machine would profit from the current order of project scores as you show, as a best guess where to start more personalized testing.

What I have noticed is that by tuning each system on its own, project by project, I can yield a seeming 20% gain. But this may be because the stats change during the tests. So the best clear refinement of only about 70% of that can be expected. But there is refinement.

The real problem is that it is hard to answer meowpao's question either way you look at it. Whats the real answer though, is what combination of machine and project performs the best. To determine this requires some time and the baseline changes, which compounds the problem.

Another problem is that if I told you that you could get 10% [improved]gas mileage out of your car, you would think I was a conspirator. The one obvious conspiracy is the one that tries to make grid.org vets into the new n-people on the block.