Our today's processor is interesting by itself, as its the first representative of Intel Core 2 Duo series that we review that has shared L2 cache halved to 2MB (FYI, both E6600/6700 and X6800 feature shared L2 of 4MB). However, due to a large clock rate gap between the lowest clock rate Core 2 Duo, for which we already have test results (i.e. E6600) and the Core 2 Duo E6300, it's hard to separate visually (judging by charts) the negative effect of smaller L2 from the effect of lower clock rate. Because of that we resorted to a little trick and expanded our charts with another "processor" existing purely in our imagination as a mathematical formula. Meet the "Prospective E6300". But what is that?
It's merely a result of a simplest extrapolation sequenting from the following assumption: any drops and increases of this series' performance are a function of clock rate. Given we know the results of a certain E6700 and E6600 benchmark as well as clock rates of E6700, E6600 and E6300, we can calculate a hypothetical result E6300 could get in that benchmark. It's obvious that our formula doesn't consider the halved cache of a real E6300. That's the point! Our virtual product seems to have the same L2 cache as its senior brothers. And we can compare its virtual performance with that of a perfectly real E6300 we have for tests.
As for its rivals from AMD camp, since we have no test results for a similarly clocked product for Socket AM2, we will provide results of the following couple: Athlon 64 X2 4000+ (lags behind E6300 as a rule) and Athlon 64 X2 4800+ (usually outperforms E6300). Judging by this lagging/outperforming we can, with a certain degree of probability, assume which of the models situated between X2 4000+ and X2 4800+ will be the closest to E6300 in terms of performance.
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 diagrams by benchmark classes.
3D modelling & rendering
Reality let down our expectations in this case: virtual E6300 got 143 points, while the real one just 138. We haven't expected any precise matches (though... well, let's not put the card before the horse). Perhaps, sometimes L2 cache size matters, after all. A fresh thought, isn't it? :) But this doesn't make it any better for AMD, since even Athlon 64 X2 4800+ is on par with the most junior of Core 2 Duo processors.
Nearly a "direct hit", our virtual CPU is even slightly slower. This is surely nice, especially considering that we are dealing with a rather serious benchmark group. Seemingly, 2MB of cache is not that bad for 2 cores. On the other hand, why should it be bad? AMD Athlon 64 X2 CPUs operate, don't they? AMD looks serious in this group: both outperformed E6300 by a minimum of 6% (A64 X2 4000+).
A rather strange result as a compiler should be critical about L2 cache (and the virtual CPU is even worse than a real Core 2 Duo E6300). AMD budget product performed rather nice, nearly without a lag.
And again an astonishingly precise match of assumptions to real test results. CPU RightMark has never been especially sensitive to L2 though (to a certain extent, of course - it still rejected the Celeron "Willamette-128"). Athlon 64 X2 4000+ (to a certain extent this one can be named a budget dual-core CPU as well as Intel Core 2 Duo E6300) demonstrates results good enough. And this is nice, because at least in the lowest price range A64 X2 is able to compete in some benchmarks with respective Intel processors.
Photo (raster image) processing
Well, there's bound to be a reason for a large L2 in Core 2 Duo processors. It would be rather strange, if developers were equipping senior models with 4MB of L2 "just to impress". We can also see for the first time as E6300 takes place not on the edge of the "AMD group", but nearly in the middle.
This is definitely not for budget solutions. E6700/6600 look so clearly victorious that results of E6300 (and both AMD CPUs) are out of discussion. There is some difference, but all of these processors lag behind winners.
Finally! It's the only group of benchmarks (archivers with large dictionaries) where virtual E6300 clearly outperformed the real one. Naturally due to latter's paired-down cache.
And here the considerable gap between the real and virtual E6300s is turned upside-down, and the real one is definitely faster. We see only one logical explanation to that: when Core 2 Duo clock rate increases, its efficiency per clock drops in some tasks.
And again Core 2 Duo E6300 is nearly on par with Athlon 64 X2 4800+ significantly outperforming AMD's budget dual-core solution.
It's just like in the previous benchmark.
In games Core 2 Duo E6300 performance is between those of A64 X2 4000+ and A64 X2 4800+, but still closer to the latter. This leads us to a conclusion, sad for AMD: even with cache paired-down, in games Intel's new core clearly outperforms AMD's older core. It's really high time to upgrade K8...
Efficiency per gigahertz of clock rate
As we expected, the junior Core 2 Duo showed the best efficiency per GHz among all Intel processors we tested.
Intel made such an interesting CPU. Performing nearly on par with Athlon 64 X2 4800+ (the latter scores 6% more in total), but seemingly positioned to compete with Athlon 64 X2 3800+/4000+ (of which it outperforms the latter by 9%, so between the 3800+ the gap is even higher). We can state that Intel's new low-end is, even by the most conservative estimates, on par with AMD's respective middle-end.
On the other hand, Intel's Core 2 Duo series is very small consisting just of 4 CPUs, while AMD offers twice as many Athlon 64 X2 processors for Socket AM2. So Intel's low-end is few steps far from its high-end. Three steps to be exact: E6300 —> E6400, E6400 —> E6600, E6600 — E6700 (X6800 is already a Core 2 eXtreme, like Athlon 64 FX). AMD has clearer positioning due to a larger amount of products (Athlon 64 X2 / Socket AM2: 3800+, 4000+, 4200+, 4400+, 4600+, 4800+, 5000+, 5200+). We can also complain there's still no promised Core 2 Duo E4200, which was planned as series' junior: 1.6 GHz core clock rate, 800 MHz FSB, 2 MB of shared L2.
Judging by today's tests, Intel's "seemingly low-end" (dual-core) is rather noticeable on the background of AMD's "nearly high-end". But still we would like to see more diversity of products based on the excellent Core 2 Duo. We, end users, are capricious. We want to see perfection not only in technologies, but also in positioning :).
All right, let's finish the article on this tune moderately optimistic for Intel. Core 2 Duo E6xxx processors are all good, including the junior. But it's a pity there's just four of them.
Memory modules for testbeds were kindly provided by
Corsair Memory Russia
Stanislav Garmatyuk (firstname.lastname@example.org)
October 18, 2006
Write a comment below. No registration needed!