AMD’s New Video Shows SYSmark Benchmark Lacking Objectivity
AMD has had problems with BAPCo’s SYSmark benchmark tool ever since 2011, however, now the company goes public to shame the popular premier performance metric. In a video posted Thursday on Youtube, John Hampton, director of AMD’s client computing products, accuses SYSmark of being biased as it’s only optimized for Intel.
To experimentally verify the unreliability of the results of the SYSmark 2014 benchmark, Hampton and AMD engineering manager Tony Salinas ran two “similar” laptops, one based on the AMD FX-8800P with integrated graphics Radeon R7 (SP 512) and the other on the Intel Core i5-5200U with HD graphics 5500.
With SYSmark 2014 running, the AMD FX laptop scored 659, while the Core i5 laptop scored about 987, showing the latter configuration to be approximately 50% more efficient.
When compared the performance of same laptops in Futuremark’s PCMark 8 Work Accelerated workload, the performance delta was little over 7%.
Finally they ran an unidentified benchmark to check the speed of execution of the script on the basis of several popular office applications. The Core i5 finished in 61 seconds, while the AMD FX chip finished in 64 seconds.
Hampton concludes that Sysmark does not use realistic every day workloads to measure and compare PC performance, as it proudly claims to be, and the results are heavily dependent on the performance of the processor.
“There is an excessive amount of high CPU tasking being done (in SYSMark). That is, the benchmark is really only evaluating the CPU side of the system,” Hampton said.
He also pointed towards the FTC’s fine print, which dictated Intel to post a disclaimer when using data from this benchmark, stating: “Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors.”