Blender 2.77a Cycles rendering time using the BMW benchmark scene. Tested rendering on 1 thread on 1 core only (thermal constraints).
NVidia Tegra (Denver, Nexus 9)
304 MHz - Time: 02:47:54, 33854 samples/s/GHz
1.06 GHz - Time: 25:28.61, 63987 samples/s/GHz
2.3 GHz - Time: 11:19.43, 66389 samples/s/GHz
Cortex A57 (MSM8992, Nextbit Robin)
634 MHz - Time: 01:20:44.57, 33754 samples/s/GHz
960 MHz - Time: 00:57:40.78, 31206 samples/s/GHz
1.82 GHz - Time: 00:28:19.71, 33515 samples/s/GHz
Cortex A53 (MSM8992, Nextbit Robin)
384 MHz - Time: 03:05:50.19, 24214 samples/s/GHz
960 MHz - Time: 01:22:19.72, 21863 samples/s/GHz
1.44 GHz - Time: 00:58:23.91, 20548 samples/s/GHz
Exynos 5250 (Cortex A15)
200 MHz - Time: 08:20:11.91, 17273 samples/s/Ghz
1.7 GHz - Time: 01:00:14.09, 16875 samples/s/GHz
Krait 400 (MSM8974AB, HTC One M8)
300 MHz - Time: 10:23:04.33, 9244 samples/s/GHz
1.03 GHz - Time: 03:00:3.34, 9317 samples/s/GHz
Krait 300 (APQ8064, Nexus 4)
384 MHz - Time: 09:32:53.12, 7855 samples/s/GHz
1.03 GHz - Time: 04:01:39.48, 6942 samples/s/GHz
Some x86 comparisons
Skylake (Core i5 6200U)
400 MHz - Time: 52:56.05, 81610 samples/s/GHz
2.8 GHz - Time: 09:56.78, 62047 samples/s/GHz
Sandy Bridge (Core i5 2400S)
1.6 GHz - Time: 21:42.19, 49762 samples/s/GHz
3.3 GHz - Time: 10:46.81, 48574 samples/s/GHz
The one I can't quite figure out is the Tegra Denver numbers. I guess it has something to do with the ARM translation software it runs.
NVidia Tegra (Denver, Nexus 9)
304 MHz - Time: 02:47:54, 33854 samples/s/GHz
1.06 GHz - Time: 25:28.61, 63987 samples/s/GHz
2.3 GHz - Time: 11:19.43, 66389 samples/s/GHz
Cortex A57 (MSM8992, Nextbit Robin)
634 MHz - Time: 01:20:44.57, 33754 samples/s/GHz
960 MHz - Time: 00:57:40.78, 31206 samples/s/GHz
1.82 GHz - Time: 00:28:19.71, 33515 samples/s/GHz
Cortex A53 (MSM8992, Nextbit Robin)
384 MHz - Time: 03:05:50.19, 24214 samples/s/GHz
960 MHz - Time: 01:22:19.72, 21863 samples/s/GHz
1.44 GHz - Time: 00:58:23.91, 20548 samples/s/GHz
Exynos 5250 (Cortex A15)
200 MHz - Time: 08:20:11.91, 17273 samples/s/Ghz
1.7 GHz - Time: 01:00:14.09, 16875 samples/s/GHz
Krait 400 (MSM8974AB, HTC One M8)
300 MHz - Time: 10:23:04.33, 9244 samples/s/GHz
1.03 GHz - Time: 03:00:3.34, 9317 samples/s/GHz
Krait 300 (APQ8064, Nexus 4)
384 MHz - Time: 09:32:53.12, 7855 samples/s/GHz
1.03 GHz - Time: 04:01:39.48, 6942 samples/s/GHz
Some x86 comparisons
Skylake (Core i5 6200U)
400 MHz - Time: 52:56.05, 81610 samples/s/GHz
2.8 GHz - Time: 09:56.78, 62047 samples/s/GHz
Sandy Bridge (Core i5 2400S)
1.6 GHz - Time: 21:42.19, 49762 samples/s/GHz
3.3 GHz - Time: 10:46.81, 48574 samples/s/GHz
The one I can't quite figure out is the Tegra Denver numbers. I guess it has something to do with the ARM translation software it runs.
Last edited:
Facebook
Twitter