AMD hints that benchmark comparison with Intel is the same thing as Volkswagen's corruption

Comparing and examining machine performance when choosing a PC is very ordinary thing, and among them, "sensible speed" when used is one of the important factors. As an index for estimating this "sensible speed", "Benchmark software"Although AMD played a major role, AMD released a movie that tests the funny result of the benchmark software by comparing it with the rival Intel machine. Among them, for example, Volkwagen, which has been involved in malicious programs for escape of exhaust gas, is a major problem globally.

Truth or Myth ?: Is SYSmark a Reliable Benchmark? - YouTube


"Machine performance" is the most important judgment factor when purchasing a PC, "said John Hampton, AMD's computer product director. However, the PC industry said that there is a "wrong" benchmark software as an element that can make this judgment wrong.


"Recently, there was corruption to the performance test results of a large car company," Hampton said, talking about Volkswagen 's test camouflage software problem. The Intel benchmark result implies that it is a product of a similar camouflage.


After preaching the importance of the benchmark that can measure the correct performance, "Then, benchmark software"SYSmark 2014Is it reliable? I will cut into the main theme.


AMD laboratory will verify the data whether SYSmark 2014 is reliable or not. Baton touch on Mr. Tony Salinas here.


"To verify the reliability of SYSmark 2014, as a comparisonPCMark 8I will try using it, "Salinas said.


First, we present the benchmark results of SYSmark 2014. PC score using Intel CPU Core i 5 is "987"


On the other hand, the score of the PC using the FX series of AMD CPU is ... ...


"659". The Intel CPU is about 50% high score.


"But this score does not reflect the impression of actually using the PC at all," said Salinas.


"Let's change the benchmark software to PCMark 8"


The score of Intel Core i5 at PCMark 8 is "4199"


Meanwhile, the score of AMD FX is "3908". That difference shrank to about 7%.


"Why do such a large difference in results occur?"


The reason is that it is in the character of benchmark software. According to Mr. Salinas, SYSmark 2014 presents the performance score considering only the computing power of the CPU, while PCMark 8 considers the capabilities other than CPU, such as CPU, GPU, and GPU subcomponent system, From displaying a performance score. Except for special use scenes such as merely using the CPU only furiously, it is claimed that general performance reflects the speed of exercise exactly in general PC usage scenes.


When measuring machine performance with benchmark, it is necessary to compare after measuring the measurement result of the benchmark software, understanding what factors are taken into consideration.


"Let's compare machine performance using the workload time in the Office suite, as it is a scene where many users routinely use PCs."


For Intel Core i 5 it is 61 seconds ... ...


AMD FX is 64 seconds. The difference is about 6.7%. It was said that it was close to the score of PCMark 8.


It is clear from these verifications that it is clear that SYSmark 2014 is not as good as PCMark 8, reflecting the speed of experience in common software usage scenes.


In conclusion, SYSmark 2014 is ineligible for benchmarking software to examine the sensory speed in daily life.


AMD laboratories have four key requirements for benchmarking software: "being realistic", "having no bias", "having objectivity" and "being transparent".