Web browsing
We have already seen the nice results of Phenom II X4 in this group of tests. And those of Phenom II X2, too. Now we see that lineup doesn't really matter, the clock rate is the key. L3 cache helps improve results too, but it doesn't help Phenom II X4 925 catch up with higher-clock-rate Athlon II. (Athlon X2/X3 models with the highest clock rates for their class are better than the newest Core i3 as well.) Anyway, even Pentium E6800 is faster than Phenom II X4 925. But that only matters in terms of numbers. In terms of real-life experience, any up-to-date budget processor will do fine in web browsing. And if you would spend at least $90 on a CPU, you won't have to worry at all.
Audio encoding
These benchmarks don't care about cache, but do favor the number of cores and their clock rates. So it's not a wonder that Athlon II X4 wins along with Phenom II X4 840. The results could've been better, but audio codecs seem to have a certain dislike for AMD processors, so those need an extra core to perform on a par with Intel models. Two, if Hyper-Threading is involved. But similarly to the previous group of tests, numbers are not important here, because not many people encode terabytes in one go. Even a Celeron or a low-end Athlon II X2 will be enough to encode some tunes, not to mention faster processors with more cores.
Video encoding
This task is similar to the previous in concept, but different in all other aspects. First, each performance point is more important, because video compression remains a tedious, resource-intensive task. Second, these benchmarks do not offer the same advantage to multi-core processors, because all depends on parallelizability. Nothing has changed on a global scale though — Propus-based CPUs are one of the best for this task in the budget segment. In turn, the lower-end full-fledged Phenoms lag behind Core i3 and are comparable to the higher-end Athlon II X3.
Games
Cache plays a very important role in games, and that leads to an interesting result: Phenom II X4 925 remains faster than other mainstream solutions from AMD. Higher-end Phenom II X3 models are also nice. All that even despite the purely computational Fritz Chess Benchmark.
But the noticeable leadership of Core i3-2100 proves it's high time AMD did something about this very sought after 100-150-dollar segment. Intel has been much more active in it over the last two years, and here's the result. The situation remains the same in the below $100 segment, but newer Pentiums may change it all right.
Final thoughts
One thing we've always criticized AMD for is offering too many processors. It was fine at first, but then the clock rates of higher-end models increased, and lower-end CPUs stopped selling. As a result, the company accumulated too many offerings in the 100-120-point range (on our diagrams, that is). No wonder AMD has finally decided to introduce some proper order. Athlon II X2 remains the low-end family, competing nicely with Celerons and Pentiums. Athlon II X3 processors are one step higher, also replacing older Phenom II X3. In turn, Athlon II X4 transforms into a lower-end subseries of Phenom II X4, because higher-end Athlon II X4 are at least as good as the older Phenom II X4 810/820 and Phenom II X4 925. It's somewhat strange that AMD still develops the Phenom II X2 series. Well, they just like it, we guess. As for Phenom II X4 900 models that still remain, they undoubtedly belong to the mid-end class, where you don't save on L3 cache. At least that's approximately how the current processors are positioned. They are a bit easier to sort out now.
Our scepticism about Phenom II X4 840 was not completely justified. Its high clock rate allows it to replace Phenom II X4 925 in terms of performance at low cost price. In fact, this is one of the reasons AMD should cancel Phenom II X3 — just so it doesn't get in the way.
Anyway, AMD must not relax, because lower-end Sandy Bridge DC processors debuted quite well. The older Clarkdale often lost to Propus and only slightly outperformed Rana, but now the only means of competition is pricing. And since Sandy Bridge DC is based on a finer process technology (32nm versus 45nm), prices for it can be cut more aggressively. As for performance competition, it would require the participation of full-fledged Deneb CPUs which are even more expensive.
In other words, AMD really needs to update its lineups — all of them. And the company is doing just that. All we need to do is wait until Llano becomes widely availble, and see how well AMD has solved this task.
We express gratitude to Gigabyte and Crucial for providing equipment for the testbeds.
Write a comment below. No registration needed!