As for now the most important thing is for the agent to fully support 64-bit CPUs. Home desktops are equipped rather with 1 x 64-bit than 2 x 32-bit. Anyway if the upgrade won't be made soon it will be a huge waste of computing power.

I can fairly confidently say that the client does not use any such chip-specific optimizations. One can hope the next version might, assuming the sort of calculations being run lend themselves to improvments using the extra processor commands. Meanwhile, the only *official* word on new versions is that we can expect a linux version sometime in 2005. Could be next week, could be a year away, no one knows at this point, besides, perhaps, Rick Alther and other WCG insiders.

As to HT, it DOES help... the rest of your system, although only in certain instances (which others have mentioned in this thread). Don't bother turning on or off HT just for WCG, unless you have computers dedicated SOLELY to running the client, because it really makes little difference for purposes of this project.

If you are convinced HT makes a big difference, and feel the burning need to prove it, try timing it. Pull down a new WU. Exit the prog, COPY the entire folder to new location, and restart the client. Time the WU to 99% completion, then exit the client again. Kill the entire WCG folder, and then restore the folder from the copy you made when the WU was new. This should force the same WU to start over in the exact same place, and make a valid comparison possible. Change your HT settings, and restart the client. Time it to 99% again, and compare the two numbers. Then you will know for certain! (Be sure to not use the computer for ANYTHING else during these tests, of course, else the results will skew.)