Wow! I just got the latest version of the NT command line executable and much to my pleasant surprise, it runs at least 40% faster than the old version. Although they say this latest version should take a little longer per unit, it's much faster for me...
Now here's why I think this is: My Celeron 566 at 867MHz was previously taking around 10 hours a unit, which is about as long as my PII 400@412. The reason is almost undoubtably the size of the L2 cache. Although the PII's clock speed and L2 cache run much slower than the Celeron's, It has 4 times the L2, which even in its PII form is much faster than system memory. v.2 obviously needed more L2 to run optimally than the poor Celeron had, so all its clock speed advantages were wasted when it had to hit the system memory. Thus, the two chips performed similarly despite the Celeron's usually huge power advantage.
But in v3.0, the executable must not need so much L2 to run at it's "full" speed, so my Celeron was free to haul ass like it should have been, and times dropped to about 6 hours a unit (while I was using the computer!). This measures up nicely to the mid-6 hour times I've seen 800/133 Pentium III's bringing in. My Celeron had its chains taken off and now it can fly (and not be embarrased by chips running at less than half the speed). 😀
So all us Celeron owners owe a big thanks to the SETI programers for giving us a big boost!
Any other Celeron owners seen similar changes? I'm curious.
Now here's why I think this is: My Celeron 566 at 867MHz was previously taking around 10 hours a unit, which is about as long as my PII 400@412. The reason is almost undoubtably the size of the L2 cache. Although the PII's clock speed and L2 cache run much slower than the Celeron's, It has 4 times the L2, which even in its PII form is much faster than system memory. v.2 obviously needed more L2 to run optimally than the poor Celeron had, so all its clock speed advantages were wasted when it had to hit the system memory. Thus, the two chips performed similarly despite the Celeron's usually huge power advantage.
But in v3.0, the executable must not need so much L2 to run at it's "full" speed, so my Celeron was free to haul ass like it should have been, and times dropped to about 6 hours a unit (while I was using the computer!). This measures up nicely to the mid-6 hour times I've seen 800/133 Pentium III's bringing in. My Celeron had its chains taken off and now it can fly (and not be embarrased by chips running at less than half the speed). 😀
So all us Celeron owners owe a big thanks to the SETI programers for giving us a big boost!
Any other Celeron owners seen similar changes? I'm curious.
