Used to be, 10 years or so ago, that a bigger number at the end was better, and the increase in speed and power consumption was, if not linear, at least always increasing.
In the market these days are two competing companies with competing architectures, each with several apparantly distinct product lines, with new chips still coming out in 32 and 64 bits, single, dual and quad cores, FX and E and all manner of other lines which serve only to confuse the issue of which one is the most reasonable performance for the price. Each one of these new chips, and they seem to come out almost weekly, was independantly reviewed by a number of websites when it was released, but they’ve been compared to chips in similar price ranges rather than against any standard benchmark, or any arbitrary way that website or reviewer pleases. How can I be expected to make an intelligent purchasing decision when I can’t really even say for sure that I know what all of the product lines are, let alone how they compare to each other? Heck, I don’t even know in which dimensions or attributes I should compare them any longer, since clock speed now appears to not be relevant except when comparing two products in the same line.
It takes some work but it can be done. You mentioned already what seems to me to be the most important thing: Each chip has been tested and reviewed by several different websites. Go read the reviews.
The standards for review aren’t as arbitrary as you make them out to be. Pretty much always, you’ll see a series of charts showing the benchmarks for the various chips. You’ll note that there will be seperate benchmarks for, for example, video compression, for business-type data crunching, for archiving, for 3d rendering, and so on. This is important: It turns out different chips are better at different things. One may be better for games, worse for business apps, while the other might be better at business apps, worse for games.
So you have to ask yourself, basically, two questions: What do I want to use this sucker for, and, how much money am I willing to spend?
Then look for benchmarks for chips at your price point, for applications you’re going to use the chip for, and the decision should be pretty easy once you’ve looked at three or four good review sites.
I just went through this process myself. AMD X2 3800+ is the cheapest AMD dual-core chip right now, and the E6300 is the cheapest intel dual-core. In fact it turns out the E6300 is a little cheaper when you consider combined chip/mobo cost, so that actually helped me make my decision. But even pretending they cost the same, what I discovered after googling and investigating three seperate review sites was that the AMD chip is comprable to the E6300 for non-game purposes, but much much worse than the E6300 for game purposes. Since a major reason I would like a new computer is to play new games on high settings, the choice is made easy for me. I don’t care how fast I can zip files. I want my computer to show me stuff in 3d that looks pretty. And through investigation, I’ve seen that its the E6300 what does this.
It is difficult. “Benchmarks” can be all over the place and it is not unheard of for manufacturers to pick benchmarks favorable to themselves to “prove” their product is faster. There has been a somewhat ongoing debate over a set of reliable and fair benchmark standards. Due to architecture and a few other reasons Intel chips may be faster for some applications while AMD are faster for others. Add in differences in the other hardware (memory, motherboard, disk drives, graphics cards and so on) and the end result of a system’s performance can be all over the place from one to the next.
I would recommend either Tom’s Hardware or AnandTech as two sites that are independant and try to fairly compare various products.
If you’d rather skip all the reading and are looking for a simple, one-stop check of various processors you could check the graphs on this site. Note I am not making a judgement on how fair or accurate that site’s benchmarking software is but for government work I expect it is close enough for most purposes.