It hasn't been long since the previous release, so it's logical to consider this article as a kind of addition. It happened so that we obtained Intel Core 2 Duo E6600 processor after the previous article was published. It's not that interesting itself as it differs from Core 2 Duo E6700 only by one step lower multiplier (and 266MHz lower clock rate). Of course, it would be much more interesting to test E6300/6400 with its "halved" cache or even the series' most junior E4200 that also has bus pared down to 800MHz. Unfortunately, these CPUs haven't arrived to our lab yet. So, for now we offer you another article about Intel's new architecture performance in popular real tasks. We hope it will not be boring, since it's our second review of that. :)
Hardware and software
* - "2 x ..." means "... per core".
In this article (and seemingly in all that will follow) we marked 64-bit software by bold font to attract the attention of readers. Because it seems some of them have the "Why no 64-bit OS and apps again?" in their email templates :).
Preface to charts
Our form of results representation has two peculiarities. First, all results are equated to relative integer "marks" (the performance of tested processor relative to Pentium D 805, given the performance of the latter is 100 marks). Second, the detailed resuts are provided in the Microsoft Excel table and the article features only the summary charts by benchmark category.
The clock rate difference between Core 2 Duo E6700 and Core 2 Duo E6600 is 2.66/2.4~=11%. The mark difference marks is 9.5%. Of course, we'd prefer is these values were perfectly congruent, but taking into account ~1% measurements inaccuracy it can be considered satisfactory: performance of the new core scales good with clock rate increase.
Again, it's perfectly predictable: simple extrapolation based on E6700 performance values would give us values precise enough.
In case of compilation E6700 performance even exceeds the "ideal" (or E6600 lags behind). Though, again, since there are no zero-inaccuracy tests, the difference of 1-2% should be considered only if it repeats in every test.
CPU RightMark didn't surprise us as well.
Raster image (photo) processing
The lower performance scalability in the E6600/6700 pair might be related to the objective reasons. Adobe Photoshop is rather sensitive not only to CPU performance, but also to that of memory subsystem and processor bus. And these are identical in both cases.
So far it's the only case when the difference in results can't be explained by clock rate difference. We could blame either BIOS and drivers, or test that allows such spread, or just green Martians. But we'll refrain from commenting this for now, since our database of Intel Core 2 Duo test results is far from complete.
The relatively low scalability can be explained by archivers' traditional sensitivity to memory subsystem that is often a bottleneck.
A rather predictable, but slightly excessive result.
This one is close to perfection.
No need to comment the obvious...
Efficiency per gigahertz of clock rate
That what was bound to happen happened the way it should have :). So congratulations. Knowing the results of E6700, we haven't expected any miracles from E6600 and it performed accordingly. But we can now add another Intel Core processor test results to our database. We believe they will look more informative near the results of, say, AMD's dual-core CPU with similar clock rate. Most likely, we'll publish such a review in the near future.
Memory modules for testbeds are kindly provided by
Corsair Memory Russia
Stanislav Garmatyuk (firstname.lastname@example.org)
July 12, 2006
Write a comment below. No registration needed!