Bug affecting older computers

[prime95]

A bug has been discovered in the 1M FFT size affecting x87 CPUs that do not support prefetching. This includes Pentium, Pentium Pro, Pentium 2, really old Celerons (less than 533 MHz?), and AMD K5, and maybe some early AMD K6 chips.

I'm not sure if SoB is using this FFT size right now. If it isn't, it is close. Until I have a fix available, run doublechecks which use much smaller FFT sizes.

[vjs]

FROM THE CREATOR TO THE SoB CLIENT



really old Celerons (less than 533 MHz?)
and AMD K5
and maybe some early AMD K6 chips


My notes...

early AMD K6 chips, This would include all socket 7 processors. I'm pretty sureall slot a's had prefect... but to safe lets move all those slot a's as well.

Second I believe 533 should be 566MHz, I know all 600's are the newer design.

Converting to secondpass on these processors is probably the best thing to do anyways. The solution is simple ...

First download the new client if you have not already done so...

http://www.seventeenorbust.com/download

Then determine your % complete, if your only a few % into the block stop the client. If you have not already upgraded reinstall over the existing client, your work should not be lost if you decide to continue with the present k/n pair.

If you decide to start fresh to avoid this problem simply change your username to.

usernameQQQsecondpass, choose yes to clear cache and it will download a new test. (example vjsQQQsecondpass ).

If you do not clear the cache your existing test will finish. Once finished a smaller n test will be downloaded avoiding the potential problem once the username is changed.


---------------------

IMHO, these processors are best suited for secondpass testing in the first place.