Among similar chips, more GHz is usually better. For example, a 3.0GHz P4 is usually slower than a 3.4GHz P4. Similarly, the 2.0GHz 3200+ is slower than the 2.2GHz 3500+. However, a 2.2GHz 3500+ is faster than a 3.2GHz P4, so you can't compare dissimilar architectures. It's like comparing cars to planes or boats using only horsepower. The fact is, AMD's architecture is much more efficient and well-rounded than Intel's netburst architecture (the basis of the P4). Without going into too much detail, suffice it to say that Intel took the path of creating a CPU with a long series of very small steps, which allows for a much higher top speed. However, this architecture is vulnerable to sudden changes that frequently occur in processing, causing the entire process to stop and start over. Also, it needs high-bandwidth access to things like memory in order to keep the machine fed. AMD, on the other hand, has chosen a shorter series of steps, and each step does more work. If something gets disrupted, you don't have to backtrack as much. Access to things like memory and the rest of the system is speedy and convenient. All this means that Intel has the theoretical advantage in terms of maximum work that can be done, but AMD is much more likely to perform near its maximum all the time, and that is why it is faster than Intel all around. Think of it like this: Intel is a ferrari, while AMD is a Hummer. On straight road, with no obstacles or speed limit, the ferrari will win. This is why Intel wins in media encoding benchmarks. As long as both machines have a steady stream of inputs, and there are no wrenches thrown into the operation, the one with more raw power will win. However, in general use, this is not the case. Imagine driving through the mountains in a storm. The ferrari is too busy adjusting to take advantage of its speed advantage, and the constant twists and turns keep it from running at a steady speed. The hummer has no problem adjusting, and can make its way through the course at a good pace while staying in control. This is why AMD wins in games and in general usage. Games are complicated, balancing AI, Textures, Physics, etc, all of which occur in various amounts at different times.
As for comparing, AMD's Performance Rating system provides a rough comparison to a P4 of the same grade. The A64 3000+ competes mainly with the 3.0GHz P4, while the 3400+ competes with the 3.4Ghz P4, etc. Just remember these tips:
1) AMD's ratings are conservative at the low end, i.e. 3500+ or below, while being a bit liberal at the high end, i.e. 3800+ or 4000+.
2) AMD adds points for things like Dual channel memory, higher clockspeed, and larger cache. This means that sometimes different CPUs have the same rating. In general, dual channel memory matters slightly more than an extra 200MHz, while the extra 200MHz matters more than larger cache. For example, there are 3 3200+ processors. The fastest is the 2.0GHz dual channel version with 512KB L2 cache (socket 939). The next fastest is the 2.2GHz single channel with 512KB L2 (socket 754). The slowest is the 2.0Ghz single channel with 1MB L2 cache (socket 754). This is the same for other models, as well.
3)AMD is usually substantially faster in games, and moderately faster in general usage. In fact, an A64 2800+ can beat a 3.0Ghz P4 in many games, and even the 3.2GHz P4 in a few cases.
4) Intel is generally faster in media encoding. However, this lead is much more pronounced against the socket 754 single channel versions of the A64. The socket 939 dual channel version is much closer to comparable P4 performance.
5)the nm (i.e. 90 or 130) represents the length of a single transistor in the CPU. As technology increases, the size of transistors can shrink, allowing more complex CPUs to be built from in the same size. 90nm (for nanometers) is currently the smallest process used for chips. When looking for an Intel CPU, remember that 130nm chips run much cooler and slightly faster than their 90nm siblings. However, 130nm chips top off at 3.4Ghz, while the 90nm ones go to 3.8 and overclock better (although with much more heat.) Also, the new 90nm parts are on a newer platform, which supports PCI-express and DDRII, which are future technologies, although DDRII is pretty marginal right now. For AMD, their 90nm parts perform better, run cooler, and overclock better than their 130nm cousins, so if you can, try to get one of those.
Hopefully this cleared some things up for you
Seriously, all you need to know is the mhz wars are over for now and performance in actual games and apps is what is important.
Facebook
Twitter