We came across an interesting test report at http://www.passmark.com/ftp/antivirus_10-performance-testing-ed2.pdf. Symantec commissioned a comparative performance test from Passmark. That is, a test measuring performance in terms of speed and resource usage rather than looking at detection rates.

Not surprisingly, Symantec came out very well overall, and deserves congratulations for demonstrating how far it's gone in addressing its reputation for slow and bloated software. Given that ESET Smart Security also came out rather well, it may seem churlish to raise objections: however, we did wonder about one of the test results. In the "Memory Usage While Idle" table, ESET's RAM usage is quoted as 31.7Mb, which is well below average and  less than 1/3 of the memory used by the most voracious RAM-eating product out of all the products tested. But Norton Internet Security 2010 apparently used only an impressive 10.85Mb, measuring with Process Explorer and Perflog++.

However, when we tried an alternative approach measuring commit charge, which we consider a more accurate measurement of a product's impact on the system, we found that Norton Internet Security 2010 increased the total system commit charge by 93 MB, whereas ESET Smart Security increased the total system commit charge by just 48 MB. The difference between the two methods is that commit charge measures the total amount of memory used by the system and how it increases when an application is running. Viewing the individual process memory consumption in Process Explorer does not expose all memory used by the application.

Which kind of proves that in performance testing,  there’s more than one way to skin a cat. Which skinning method you choose might depend on how sharp your knife is. ;-)

Andrea Kokavcova
Senior Marketing Research Analyst