SuperPi Since ages, the SuperPI program is one of the favorite benchmarks among overclockers. It can be used to test for quick stability (1M) and 24/7 stability (32M). As it's quite old it does not matter if the CPU is single, dual or quad core. In fact, most benchmarkers turn off cores to gain higher speeds. Cache on the other hand is very important, the more the better.

As you can see, the E2160 is definitely slower than the E6300 at stock, as the lower cache decreases the performance ~13%. Even at 3.3Ghz, the E2160 does not manage to surpass the E6300 at 3Ghz, the gap is closed to 6%.

Same conclusion in the 32M benchmark: the E2160 is completely outrun by an equally clocked E6300 (22% difference at stock and 20% at 3ghz)
Sisoft SandraThe Sisoft Sandra application measurers system performance with different synthetic benchmarks, in the chart below you find the results of the CPU benchmark (Dhrystone/Whetstone) and memory bandwidth benchmark (int/buff’ed). Higher value is better:

At stock speeds there is a small gap in pure CPU power, this difference is gone once both models run at the exact same speed. The theoretical memory bandwidth for the E6300 is consistently higher as it run at a higher FSB speeds (266vs200Mhz).
3DMark06 CPU TestThe CPU benchmark subtest from Futurmark’s 3DMark06 is excellent performance gauge, it’s pushes a detailed scene, normally running at 30-60FPS on a VGA card, to render in software mode, this results in an average FPS of 1 or below on most systems. This CPU test is multithreaded and you will notice a higher score if your CPU has multiple cores.

While the SuperPi benchmark favored the extra L2 cache of the E6300 we see different outcome here, the results with both processors is pretty much on par, L2 cache doesn’t matter much here, 1Mb or 2Mb, the results are almost on par.