Java
Applications, written in the interpreted programming language Java, load the system in various ways. So AMD processors, better balanced in computing units and most importantly in communications with memory, easily outperform their rivals. Athlon II not only catches up with the E7400, but formally outperforms it. By the way, this group of tests is of more interest for budget processors than for expensive powerful CPUs.
Archiving
AMD takes the lead again. But this group of tests shows that a fast cache sometimes cannot be replaced even with fast memory access, so Athlon II is significantly outperformed by the more expensive model. Cache access latencies still differ much from memory access latencies, and that's what really matters in archivers.
Audio encoding
Audio encoding tasks, integer by their nature, are performed faster by Intel processors. But from the practical point of view, it must be noted that the difference between all four processors in this group of tests is not big. They are all approximately twice as slow as Core 2 Quad Q9300 (which results are taken for 100% for CPU scores in our test procedure). Hence the practical advice to users, who often encode audio: take a closer look at multi-core processors. At least when audio is encoded with such shells as dBpoweramp, that can use several cores even for single-thread codecs by encoding several tracks simultaneously.
Video encoding
It's an unexpected situation, even irrational. We cannot explain it from the logical point of view operating with CPU performance terms -- Phenom II X2 550 is not weaker than Athlon II X2 250 in any parameter, and it has just as many cores.
|
Athlon II X2 250 (DDR2) |
Athlon II X2 250 (DDR3) |
Phenom II X2 550 (DDR2) |
Phenom II X2 550 (DDR3) |
Pentium E5300 |
Core 2 Duo E7400 |
ProCoder |
0:04:14 |
0:04:06 |
0:03:49 |
0:03:46 |
0:05:28 |
0:04:59 |
DivX |
0:06:16 |
0:06:08 |
0:06:02 |
0:05:54 |
0:06:14 |
0:05:43 |
VC-1 |
0:12:15 |
0:11:58 |
0:11:28 |
0:11:23 |
0:13:19 |
0:11:54 |
x264 |
0:20:00 |
0:19:42 |
0:19:50 |
0:19:55 |
0:21:41 |
0:19:15 |
XviD |
0:05:12 |
0:05:00 |
0:06:57 |
0:07:01 |
0:05:20 |
0:04:40 |
If you take a look at detailed results, which were also duplicated in a table, this happens because of the strange result demonstrated in the XviD test. No matter how many cores, all Phenom II processors used to demonstrate much worse results in this encoder than their competitors from Intel. It was unusual, but there was nothing criminal in it. To be more exact, we could explain it with optimizations or specific load on arithmetic units. In fact, it's much more primitive -- to all appearances, it's a bug in the current version of the codec, some optimization error that does not allow the Phenom II to process data in the optimal way. But Athlon II unexpectedly avoids this problem, and the processor demonstrates results appropriate for a Phenom II adjusted for more cores, frequency, and other aspects that improve performance (in this code). That is at least on a par with its direct Core 2 Duo/Quad competitors.
We can grumble about XviD, as an Open Source project, and "too many cooks spoiling the broth". But on the other hand, Open Source programs can theoretically recover from bugs faster than commercial products. We'll keep tabs on this situation, of course.
What concerns results in the other codecs, we can mention only the E5300 lagging behind. The other processors are subjectively fast.
Games
Progressive architecture of Phenom II/Athlon II processors again proves its worth in games. As a result, a clear-cut victory and a nice diagram showing good performance gains that the new processors can squeeze from DDR3.
Conclusions
There is probably no need to mention that the processors under review are quite interesting offers in their target segment. The tests also prove that the upgrade to the 45nm process technology and elimination of bugs that hadn't allowed K10 to show its full potential in the first Phenoms gave AMD an opportunity for symmetric response (dual-core versus dual-core in this case).
Judging by results of Athlon II, this AMD core can do fine with a small cache, which is very handy for inexpensive processors, because such dice can be designed smaller and cheaper. What's important, this processor does not totally depend on memory performance despite its reduced cache size. It outperforms its rival even with DDR2-800. However, upgrading to DDR3 is quite beneficial in a number of tests, so support for the new memory type is not added for marketing reasons only. It should also be mentioned that Athlon II supports AMD-v virtualization technology, while Pentium Dual Core lacks this functionality. Support for such technologies appeared in desktop processors long ago, but it would tell nothing to most users. However, Windows 7 uses these features to emulate Windows XP.
What concerns Phenom II X2 550, this processor will probably be more popular among home users than in the corporate segment, although it generally performs on a par with the E7400. AMD did a good job as well by providing an unlocked multiplier for convenient overclocking, the more promising platform AM2+/AM3 that will see a lot of processors in future, unlike LGA775 (it's important for gradual upgrades), higher performance in games. Objectively, this processor will hardly be manufactured on such a great scale as Athlon II, because it uses the same dice as those in Phenom II processors with three or four cores. And as the process technology improves, there will be less rejects with only two working cores.
We express our gratitude to Corsair Memory Russia for a contribution to our testbeds.
Write a comment below. No registration needed!