You're not quite there yet. You should be able to crank your 2500K up a bit higher. Intel's 32 Nanometer processors are OK up to 1.35 Vcore with Core temperatures in the upper 70's at 22C Ambient.
Although Prime95
Large FFT's states "maximum heat, power consumption, some RAM tested", it's very misleading. That statement actually refers to
TOTAL power consumption shared between the processor and memory.
As such, if you observe Large FFT's thermal signature in the "Graphs" in the utility "SpeedFan", you will see that Large FFT's is a fluctuating workload, and does not produce consistent maximum processor temperatures because it is not a steady-state 100% workload.
Large FFT's instead cycles between processor workload, and RAM workloads
shared with the processor. "Blend" which is Prime95's default test for memory stability is even more cyclic, and is seen in SpeedFan's "Graphs" as a thermal signature with large fluctuations that looks like a bad day of ups and downs on the Stock Market graph.
Small FFT's is the standard for thermal tesing because it's a steady-state 100% workload that's
not shared between other components. Again, this can be seen in SpeedFan's Graphs as a thermal signature which is a smooth and straight line that clearly indicates a steady-state 100% workload.
Other utilities such as LINPACK and OBT (Intel Burn Test is
NOT written by Intel) are useful for stability testing, but are
not suitable for thermal testing because they have segments that load all registers with all one's, which produces a fluctuating 115% workload. Although the "Performance" Tab in Windows Task Manager can not indicate workloads above 100%, this is instead revealed in SpeedFan's Graphs as temperatures that spike 15% higher than Small FFT's.
Please read this Sticky in the CPU Forum:
Intel Temperature Guide -
http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html
Hope this helps,
Comp