AMD X4 975 vs. Intel i5-2400 [closed]

Question

Take these two processors. They are roughly the same price (AMD one just slightly cheaper, if you factor in the motherboard as well).

I'm trying to decide between these two and the reason why I'm having trouble is the "Intel is more powerful" rule of thumb. Because from a completely objective point of view, the AMD chip kicks Intel's microbutt:

1. More L1 cache (4x64kb vs. 4x32kb)
2. More L2 cache (4x512kb vs. 4x256kb)
3. Faster clock (3.6Ghz vs. 3.1Ghz)
4. Unlocked (overclocking friendly) vs. locked

The only places where AMD loses are:

1. 45nm vs. 32nm. manufacturing process
2. 125W vs. 95W TDP
3. No GPU integrated -- which I don't need anyway

My questions are:

1. Are there any hard specs that I overlooked that would make the i5 better? If yes, what are those?
2. If the answer to 1. is "no", then what makes the i5 better in real-world use? Is it the 32nm architecture? Is it the TDP (less power wasted on heat = more power used for computing)?

Here they compare the X4 975 with the i5-2500k, which is slightly more expensive than the 2400 (which, in turn, is slightly more expensive than the X4 975). However, solely specs-wise, the 2500k still seems worse than the 975 (it only knocks off number 4. in the first list). They say that the i5 "simply crushes" the X4. What makes it do that?

Answer 1

Clock speeds are only a good basis of comparison between chips in the same architecture. Different architectures need different numbers of clock cycles to do the same work. If the AMD has 15% more clock cycles per second, but needs 40% more to finish the same processing, it will take longer.

Also, you forgot TurboBoost. The more power-efficient Core i5 processor can dynamically overclock, because it uses less power at stock speeds than the AMD. This is even more true when you're not using the IGP, since the thermal capacity it's not using can be used for more overclocking.

And you missed something in the L1 cache comparison. These chips have two kinds of L1 cache, instruction and data, yet you've given only a single number (per core).

Answer 2

Generally current generation Intel CPU's are slightly more powerful than the AMD equivalent, this is can be seen in benchmarks; for example using PassMark we can see that the i5-2400 scores 6,109 whereas the AMD (Phenom II?) X4 975 scores only 4,365. While this is only a single synthetic benchmark, the general trend holds true if you look at Tom's Hardware graphs. For example looking at the per-core comparison for Cinebench, while the graph doesn't have the exact same CPUs, a roughly equivalent comparison would be i5-2500K scoring 1.20 vs AMD X4 980 scoring 0.88.

These trends tend to hold true across most synthetic benchmarks and some game benchmarks as well, however today most CPU's are by far and a way powerful enough and you'll find that they are almost never the performance bottleneck. So at the end of the day, if you prefer AMD because of the non-performance related features (such as chip-set support, better GPU integration, more unlocked processors and the possibility to actually over-clock locked CPUs) then run with it. Much better gains can be made to a modern system (avert bottlenecks) by adding Solid-State Drives or more powerful Graphics Cards and in some cases more physical memory (even if programs aren't using it, your OS will use it for caching data from disk).

As to why things are this way, you get down very quickly into low-level architecture design and how these differ between AMD and Intel, for example this includes things such as: Instruction Pipeline depth, Accuracy of the Branch Predictor, Supported Instruction Sets, Available compute resources (FPU, ALU), Design of the Bus and so many other things. It can also come down to things as simple as how many transistors have been spent per core. Wikipedia has a selection of very good articles on CPUs and their design.

Disclaimer: I use an AMD Phenom II x6 1055T at home and an Intel i5-2400 at work.