I know this is going to be hard to believe.... but it's actually pretty normal to see higher temps while gaming than while running P95 on an FX-6300 with turbo core and APM active. The reason for this has to do with the way the chip's clocks and voltage scale with different workloads in order to maintain operation with the specified TDP envelope.
When you run P95, all the cores get hammered, and APM pulls back the voltage and clocks (to as low as 3.2ghz depending on the specific chip and your implementation of APM, which could theoretically vary a bit from board to board), and pulls back the voltage (to as low as ~1.15V). When running like this, the ~95W being dissipated by the chip is very spread out, evenly across the chip, the "peak" temp anywhere in the chip is low.
When gaming, the workload really only hammers on 1-3 cores hard at a time, so the chip ramps up to ~1.425V and pushes for 4.1ghz turbo speeds. This causes much higher localized temperatures even though the overall dissipation is likely no higher than it was running P95... This is why you're seeing 45C peak temps in gaming, and only ~30C peak temps in P95. This is normal...
Turn off APM/Turbo, and the phenomenon will "go away" because the chip will run at a more "fixed" voltage under a load at 3.5ghz all the time regardless.
-----------
Now.... it's important to understand that there is actually no such thing as a core temp reading from a piledriver chip. Try not to fuss too much over whether the numbers seem too high or too low in a particular piece of software because they are all WRONG. However, WRONG is fine, as long as you understand the way in which it is wrong.
On a properly calibrated scale, idle temps can read as low as 0C on piledriver, and the maximum supported continuous operating temp should read as 70C. This "range" is roughly the inverted output from AMD thermal margin, which is a very precise, but very inaccurate "calculated" temperature monitoring system. I do not know what "scale" HWMonitor is calibrated to, so I can't tell you where you are really at... but you can find out...
Use AMD Overdrive to observe thermal margin. This is the most precise way to observe your CPU temps.
------
Oh.. and on a final note... It would not surprise me at all for a properly mounted 212 EVO to give a stock 6300 some pretty HUGE thermal margin readings (very low temps). The 212 EVO's heatpipes are "tuned" to produce chart topping heat flux characteristics at loads in the ~100W range. It's not a great cooler for overclocking at 200W+, but down under 150W is probably the best heat-pipe unit out there.