Well, in reading around, that processor has some history for running very warm with some motherboards. I note some folks have gone so far as to lower the voltages on the CPU to lower the temps - and thus fan noise. I have the 2.8Ghz i7 860 on a Gigabyte board pushed a little to 3.15Ghz using Gigabyte's EasyTune utilities. CoreTemp's tray icon typically sits around 51 - 52° when I am using my machine. That said, I am not a gamer but typically have several things busy on two monitors so while I may tax my CPU's resources to it's limits, it is typically in very short bursts - easy for my OEM cooler to handle. I may touch 60°C but that's typically a warning to me it's about time to clean the air filter of dog hair, dust and who knows what.
As far as TIM, as long as you have complete coverage across the die, there is no such thing as too thin of a layer. The best heat transfer occurs with direct metal to metal contact of the mating surfaces. The TIM just fills in (and pushes the air out) of the microscopic pits and valleys of those mating surfaces. Any extra is in the way and counterproductive. And as noted in that sticky, with some TIM, you can realize another 3 or degrees of coolness once the TIM has "cured" over several full heatup and cooldown cycles. So if you think you might have it too thick in some spot, you might try another layer of TIM one more time. And of course, make sure the fan assembly is properly fasten - necessary for even pressure.
That said, benchmarks don't typically, in spite of what their makers may say, reflect real-world computing. At best, they reflect the computing demands of what the makers think is the "typical" user. What's typical? There are 1 billion Windows machines out there and each and every one became unique the first time its new owner turned it on. Benchmarks were designed to measure benchmarks - the extreme limits of the design and configuration. If you are testing to see how long your CPU can run at full power before blowing up, use a benchmark program. For my machines, I don't care what Prime95 says. I care what CoreTemp says, and my own observations of performance. I like CoreTemp, BTW, and you can set it to display only the highest temperature of the CPU's cores.
If this were me, I would put the OEM fan back on, with a fresh new layer of TIM and see what happens.