Replies to post #40757 on Advanced Micro Devices Inc (AMD)

[sgolds]

upc, combjelly, morrowinder - let's address upc's point that they are different scales and are not meant to be equated (even on 32-bit apps on 32-bit OS).

Then I am against it.

If you are going to distill performance to a single number then different architectures must be comparible on like applications. To do otherwise will confuse the buyers and they will not believe any of the ratings. (Lessons from Cyrix, of course.)

However, I am not convinced that this is what AMD is doing. Let's look at the statement which was quoted here:

Q: Do the model numbers of the AMD Sempron processors represent performance relative to the AMD Athlon processors?
A: No, the third-party benchmarks for the AMD Sempron processor include leading productivity applications and benchmark suites. These benchmarks reflect real-world application performance for popular, every-day software applications. These benchmarks are intended to help consumers and PC manufacturers make educated decisions. The AMD Sempron processor is designed for everyday computing and employs a benchmark suite focused on popular productivity applications. The AMD Athlon 64 processor, designed for demanding multimedia applications and a cinematic computing experience, employs a broader benchmark suite encompassing advanced digital content creation and demanding 3-D gaming applications. For more information on the performance of the AMD Sempron processor, visit www.amd.com.

This does tell us that the benchmark suite for A64 is more comprehensive than the suite for Sempron. It leaves me wondering:

1. Is the Sempron performance suite a subset of the A64 suite?
2. How do application results compare in the places where both suites overlap?

If like-rated Semprons and A64s don't compare equivalently on the overlapping parts of the performance suite then AMD has a major PR disaster coming very soon!