AMD says SYSmark benchmarks are Intel-biased and don't reflect real-world usage

Chip maker AMD has complained that the popular benchmarking tool SYSmark is biased toward rival Intel. The company says that the software focuses far too much on CPU activity and that this is not reflective of real-world computer use.

In a stern video, John Hampton and Tony Salinas from AMD compare SYSmark benchmark with results from Futuremark's PCMark 8 as well as scripts devised in-house. Comparing an Intel Core i5 with a 'comparable' AMD FX chip, SYSmark was found to report a much greater performance difference -- in Intel's favor. AMD describes this as "quite astonishing and not realistic".

While SYSmark showed a performance score difference of around 50 percent, when the same chips were compared in PCMark 8 the 'delta' (as AMD describes it) drops to just 7 percent. Similarly, when comparing the performance of AMD and Intel chips using scripts that run Microsoft Office-based tasks, the difference was between 6 and 7 percent.

AMD not only suggests that SYSmark's results are unrealistic, but says tests are geared in favor of Intel chips. The company recommends people read an FTC statement which says SYSmark "may have been optimized for performance only on Intel microprocessors".

Check out the video, entitled Truth or Myth?: Is SYSmark a Reliable Benchmark?, to see what you think: