Core 2 Duo: Effects Of Memory Timings

File Compression & Encryption:

Our file compression and decompression tests were split into two halves to cover a broad spectrum of performance. The first test we ran was to compress and encrypt the MPEG-2 source file from our video encoding test with the highest quality compression ratio. Secondly, we compressed and encrypted the folder of 400 photographs used in our Photoshop Elements test with the same compression settings.

We saw some tangible performance improvements when using tight memory timings in WinRAR - this was especially the case when compressing and encrypting large files. Tightening the timings from 4.0-4-4-12 to 3.0-3-3-9 resulted in a performance improvement of nearly 6% in the large encoding test on the Core 2 Duo E6600. In the small file test, we didn't see quite the same performance improvements - the benefit of running tight 3.0-3-3-9 timings was down to 3% over modules with 4.0-4-4-12 timings.

Interestingly, the performance improvements weren't as large with the 2MB L2 cache-equipped E6400 - we saw a 5% performance improvement in the large file test. The small file compression benefitted more from tighter memory timings on the Core 2 Duo E6400 than the E6600 did - performance differences between 3.0-3-3-9 and 4.0-4-4-12 were close to 4% in favour of the tighter timings.

File Decompression & Decryption:

The two RAR archives created during the compression and encyption tests were then decompressed and decrypted.

With decompression and decryption, the Core 2 Duo E6600 showed no measurable performance differences in the large file decompression test. With the memory timings set to 3.0-3-3-9, there was a one second performance improvement in the small file decompression test, but there were no performance differences between 4.0-4-4-12 and 5.0-5-5-15.

The E6400 showed some performance benefits when moving from 5.0-5-5-15 to 4.0-4-4-12, but there were no further improvements when moving to 3.0-3-3-9 in the large file test. In the small file test, fortunes were reversed - there was a performance benefit as a result of using 3.0-3-3-9 timings, but there were no performance differences between 4.0-4-4-12 and 5.0-5-5-15.