Recent Intel Clover Trail+ benchmark results obtained via the AnTuTu benchmark showed Intel's new mobile chips offer a big performance advantage versus ARM-based alternatives, but now people are questioning the value of this widely used benchmark because it offered Intel chips an unfair advantage.
Enthusiasts discovered that since version 2.9.4 of the AnTuTu benchmark, the x86 version is compiled with Intel's C++ compiler, which implements various optimizations like auto-vectorization, while the ARM version is compiled with GCC and lacks support for NEON, a widely used ARM instruction set.
Since this controversy has hit the web, the creators of AnTuTu have released version 3.3.2 of the benchmark. The changelog of this version show that changes were made to "enhance the stability of scores", and benchmark results show there's a massive difference between the performance of Intel chips in AnTuTu 3.3.1 and AnTuTu 3.3.2. In the news version, Intel's processors score 20 percent slower in the overall and CPU test results, and a whopping 50 percent lower in the RAM tests. Curiously, the results of ARM-based devices like the Samsung Galaxy S4 remain relatively unchanged.
His testing showed that on his Lenovo K900 smartphone and the Intel Atom Z2580 processor that his AnTuTu CPU and overall scores dropped by approximately 20 percent, while the AnTuTu RAM score plummeted by approximately 50 percent. It shouldn't come as a shock that Samsung S4 Oct and Qualcomm Snapdragon 600, both used in the Samsung Galaxy S4, remain relatively unchanged. Ouch!! Was Intel trying to put one over on ARM? The release notes of the latest build of AnTuTu Benchmark show that changes were made that 'Enhances the stability of scores'.
To make matters worse, we ran across this post on Berkeley Design Technology's (BDTI) Web site found that "the ARM-based [Samsung] Exynos processor performs all the operations specified in the benchmark source code, while the Intel Z2580 processor skips some steps." Guess what benchmark this issue was found on? The AnTuTu Benchmark once again. Double Ouch!