Porém...
I'd definitely take the E4500, Core 2 Duo; without hesitations. And here's why.
I'd bet one hundred bucks that if you were given two processors with the same cache size but different clocks speed, say, one with 2.2GHz and the other with 2.4GHz you wouldn't notice any difference in real-world applications and herein including games and processing/rendering apps. I am not saying that the processor with the higher clock speed won't act faster, it certainly will, and this makes sense architecturally, however, you as a normal human being wouldn't notice the difference in these days between those two modern CPUs.
The times when moving from a K6@300MHz to K6-2@400MHz was considered a huge step are behind us.
Now as an extension to my previous scenario, on the other hand, if you were given two processors running on the exactly same clock speed frequencies, say, 2.2GHz, but with different cache sizes, one having 1MB and the other 2MB, I'd bet that amount of money that you are going to notice the difference in real-world applications starting from multi-tasking in windows up to rendering/processing and gaming.
As a British guy posted on a random forum, the following analogy applies the most: “If the clock speed is the engine and the cache is the fuel tank. With a small fuel tank, all the speed in the world won't help much if you have to keep stopping for fuel every few miles, a large fuel tank means you can floor the pedal and appreciate the speed for quite a while.”
Therefore, doubling the amount of L2 cache is damn worth those $30 bucks. Unfortunately, nowadays, the cache is between the most underrated factors when choosing processors. It surely matter, it's all about cache size. Ask any hard drive, how does it ran with 8MB compared to 16MB. Doubling the amount of information that can be stored means less overall disk access, ultimately, less "time" is spent on redundant tasks that can be saving up thanks to the higher amount of cache. The same applies with your processor. The cache is ranked on the 2nd top level of the memory hierarchy. As a result, they are the most often used after the registers.
The L2 cache is entirely built up by transistors (SRAM). As a result they are very expensive but they are damned faster. Just to think of it, the access time of the L1 cache (on-chip) is 2-8 ns (nanosecond) and of the L2 cache (off-chip) is 5-12 ns. How do they compare with the main memory (DRAM) that sports 10-60 ns (sometimes even more) access time or with hard disk drives having 3,000,000 - 10,000,000 ns access time? Now I hope you do understand why is it critical to have large amounts of cache storage because it certainly doesn't matter how much it takes to fill it up with new stuff. And always remember, 2MB is 2x1MB, thus, it is double. As a result, you can place twice as much data than you could with the 1MB. And all of that for just an extra 30 bucks.
The CPU doesn't like to wait doing nothing (idle) until the memory I/O is done. As a result, it does multiple things at once (superscalar execution, pipelining, and all that). Since it is impossible to fit everything in the cache, the processors requires data thousands of hundreds of times each second. We call these cache misses, it tries to locate something in the registers, they don't have it, then it checks the cache, it doesn't have either, so it slowly gets loaded from the RAM (which was loaded from the HDD).
Now, after this usage, sometimes the cache must be also cleared. The frequency of clearing the cache (emptying) happens much often with a smaller cache (hence top priority tasks have access/storing priority). Then in case of redundant tasks (usually anything is redundand excluding streaming media) chances are the data that was located once in the cache might be required later on, but since the cache is small, it was wiped off in order to make space for the new data to come in, but the processors needs the previous data again, so it gets refilled, again much I/O access time, and so forth. In case of a larger cache due to branch prediction, parallel processing, and other hi-tech microprocessor architecture technologies, the data may stay in much longer in the cache. As a result, you save up large amount of I/O time.
All in all, I'd personally say that doubling the cache amount is worth much more than those extra 200 MHz and here we haven't mentioned the other microprocessor architectural benefits coming from the E4500 which aren't in E2220. In pure raw mathematical calculations, FFTs, calculating PI with 2^32 digit precision, perhaps the higher clocked processor might outperform the another. But in everyday usage, general purpose, multitasking, gaming, basically anything that you can think of, the extra cache will make a notable difference.
Remember, millions of those extra nanoseconds do add up...
É MELHOR ESPERAR MAIS UM POUCO!