- Jun 3, 2005
- 1,080
- 0
- 0
Due to popular request, I'm doing a very small pilot with pre-testing some CPUs.
The basic methodology is to set the vcore/VTT to 1.3 as a reasonably conservative setting. The normal things like C1E, Powerstep, etc are shut off, but otherwise things are left stock. I gradually boost base clock and run Prime for 1 hour looking for crashes, and keep doing so until it crashes. Then I back off and keep testing until I can pass 4 hours of prime with no errors.
The big caveat here is that this is *not* supposed to be comprehensive testing. From an economics standpoint, the time required to do that would be prohibitive. I also intentionally want to leave plenty of options for people to tweak, to leave some of the surprise and fun in trying to squeeze a chip as best as possible.
The point is just to get some basic comparison going on. So, with that said, under these circumstances, what would you consider a good, average, and poor speed result?
The basic methodology is to set the vcore/VTT to 1.3 as a reasonably conservative setting. The normal things like C1E, Powerstep, etc are shut off, but otherwise things are left stock. I gradually boost base clock and run Prime for 1 hour looking for crashes, and keep doing so until it crashes. Then I back off and keep testing until I can pass 4 hours of prime with no errors.
The big caveat here is that this is *not* supposed to be comprehensive testing. From an economics standpoint, the time required to do that would be prohibitive. I also intentionally want to leave plenty of options for people to tweak, to leave some of the surprise and fun in trying to squeeze a chip as best as possible.
The point is just to get some basic comparison going on. So, with that said, under these circumstances, what would you consider a good, average, and poor speed result?