What a difference of a few degrees makes...

Today my room temperature dropped 4C. Already I have gained 50mhz with the same voltage(2400 to 2446). Before I wouldn't boot at this speed with this voltage. Granted, I got a Prime95 error, but I think this speaks volumes to what is possible when I pipe in some cold air during the winter.

Also, this is useful information for those of you trying to 'get over that hump' of a few MHZ or so.

From my estimate, 4 C would translate into about 1.6% of CPU MHz (full load) for Tbred B. So at 2400 Mhz, that would be about 38 MHz. If temperature goes lower, it would be 38 MHz faster at full load, and vice versa.

hitechjb1 said:

...

Effect of die temperature on CPU clock frequency at a given Vcore
For a given Vcore, die temp does make a difference, can be a big difference.

E.g. for the same Vcore 1.9V on a Tbred B 1700+ DLT3C, roughly speaking,
- a difference of 10 C would enable one to clock about 4% faster, which is roughly 100 MHz at 2500 MHz.
- If there is a 50 C difference in die temp, e.g. one at 50 C on air, one at 0 C using extreme cooling, at same Vcore, the CPU at 0C would be able to run at about 20% faster, which is about 500 MHz more, compared to the one at 2.5 GHz on air.

Reason: electrons move faster at lower temperature under same voltage.

Click to expand...

Details:
Relationship of clock, die temperature and Vcore (update) (page 13)

It seems your estimates are pretty accurate. Very good work there.