Whether you like it or not, AMD continues to expand its range of desktop dual-core processors for Socket AM2 combining just two parameters: core clock rate and L2 Cache size. This article is devoted to another combination of the well-known components — L2 Cache of this processor is of the same size as in Athlon 64 X2 4800+ (1 MB for each core), and its clock rate equals that in Athlon 64 X2 5000+ (2.6 GHz). Thus, having accumulated the key advantages of both previous models, it got the proud name of Athlon 64 X2 5200+ (standing ovation).
But as the desktop series is getting close to the FX series, the FX-62 has little time left to "fatten": the only model that AMD can launch without pushing it aside is a 2.8 GHz processor with halved L2 Cache. It's only natural to assume that it will be named Athlon 64 X2 5400+.
But our tests gave us a little surprise. Our constant readers should know that our permanent test procedure, consistently used from review to review, lets us test a processor only once. And then we just use the results for later articles. There are also some drawbacks in this approach, especially for new processors — the choice of "companions" for the first tests may not be the best, which will affect results among other things. It's easy to find out, as a rule: results of a previously tested faster processor suddenly conflict with results of a weaker processor, which was tested later (for example, they got worse). There can be only one solution in such alarming situations: re-test the CPU "in question". That's exactly what we have done with Athlon 64 FX-62. Thus, FX-62 results in this article are updated, even though we already reviewed this processor before.
Hardware and Software
* - "2 x ..." means per core
Essential foreword to charts
Our test procedure features two peculiarities of data representation: (1) all data types are reduced to one - integer relative score (performance of a given processor relative to that of Pentium D 805, given its performance is 100 points), and (2) detailed results are published in a Microsoft Excel table, while the article contains only summary charts by benchmark classes.
3D Modelling & Rendering
AMD processors lined up in a standard "ladder", only FX-62 is slightly out of this order. However, it outscores the A64 X2 5200+ in performance no more than in frequency, so we can assume that the results are within the measurement error.
CAD & CAE
Such comparisons as the 4800+ versus the 5000+ and then the 5000+ versus the 5200+ show that CAD applications react better to a frequency gain than to a cache increase. It's confirmed by the comparison between the 5200+ and the FX-62.
It's quite a standard situation. But like in 3D package tests, FX-62 is slightly out of the general trend.
CPU RightMark reacts to frequency and practically doesn't react to cache. Quite an illustrative situation.
The situation is similar to CAD's — performance grows with the increase in frequency as well as in cache size. But it grows noticeably better in the first case.
A brilliantly informative result! Large cache let the 4800+ catch up with the 5000+, and the FX-62 did not outperform the 5200+ that much, despite its advantage in frequency. Paying heed to the good old Occam's Razor, we can only assume that Socket AM2 is limited by memory performance (or by insufficient cache — though the increased cache can solve this problem for a very short period of time) in this class of applications.
If you remove the 5000+ with a small cache, you will see a perfect "ladder". The conclusion is obvious.
The 5000+ is practically on a par with the 5200+, the 5200+ is outperformed by the FX-62. Only the frequency makes the difference.
We have already seen something like that in CAD and 3D modeling tests.
The situation is similar to the previous test.
3D Shooter Games
Games prefer cache to frequency. At least on the AMD platform. By the way, it's indirectly proved by results of Intel processors, because in case of a single active thread, their core gets all 4 MB of the shared L2.
Efficiency per GHz
Well, we expected no surprises, and we saw no surprises. Updated test results of the Athlon 64 FX-62 eliminated all oddities due to its old test results, and general performance of all top dual-core processors from AMD lined up in a regular "ladder". Athlon 64 5200+ is no disappointment or a joy, it's just another step that used to be missing.
Memory modules for our testbeds are kindly provided by
Corsair Memory Russia
Stanislav Garmatiuk (firstname.lastname@example.org)
November 28, 2006
Write a comment below. No registration needed!