For some reason I’m always a bit wary of the benchmark results I find here and there on Internet, so I thought I’d make sure I store mine in an easy to find place. I.e., as usual, here ^^
Usual reminder: benchmarks are a way to very quickly compare various hardware in a standardized fashion, however they are not as relevant as real-life applications. Just to give an example, between the 3rd and 4th generation, Intel CPUs got a new instruction set (AVX2), which brings large improvement speed to some algorithms, such as x265 encoding. This isn’t reflected in this benchmark.
Intel i7-6700K (quad core, 4.0 GHz) + GeForce GTX 1060 (mobile). Factory voltage and clock for both.
- CPU (multithread): 873
- CPU (single core): 177
- OpenGL: 145.19 FPS
Intel i5-6500 (quad core, 3.2 GHz). Factory clock, no discrete GPU. 130 mV undervolt, but it shouldn’t impact performance.
- CPU (multithread): 542
- CPU (single core): 140
- OpenGL: 36.49 FPS
Intel i7-3632QM (quad core, 2.2 GHz) + AMD Radeon HD 7970M. Factory voltage and clock for both, except that I ran the OpenGL benchmark both in “performance” mode (= normal clock) and energy-saving mode (= the clock is about half slower). The computer is a Schenker XMG P702.
- CPU (multithread): 508
- CPU (single core): 102
- OpenGL: 57.94 FPS
- OpenGL (“maximize battery life” setting in Catalyst): ~43 FPS
On a side note, the reference benchmarks (bundled in Cinebench) for CPU single core indicate 101 for the i7-3840QM (which is 2.8 GHz) and 122 for the i7-3720QM (which is 2.6 GHz). Doesn’t sound very logical and supports my mistrust in published benchmarks.
AMD A6-3500 (tri core, 2.1 GHz). Factory voltage and clock, no discrete GPU.
- CPU (multithread): 166
- CPU (single core): 61
- OpenGL: 19.65 FPS
AMD Phenom II N930 (quad core, 2.0 GHz) + AMD Mobility Radeon HD 5650. Factory voltage and clock.
- CPU (multithread): 188
- CPU (single core): 49
- OpenGL: 14.15 FPS