PCMark 8

Website: Futuremark

PCMark 8 Photo Editing V2

This workload involves making a series of adjustments to a set of photographs using ImageMagik – an open-source image processing library – to adjust brightness, contrast, saturation, and gamma. When a favourable balance is found, the changes are then applied to the rest of the images in the set. TIFF files up to 67MB in size are used.

Video Editing V2 Part 2

This workload uses FFmpeg to apply video enhancement filters to a high bitrate H.264 video and then encode it to a format suitable for distribution. The FFmpeg binary used is custom-built by Futuremark using a development version of the source available from the project's code repository. The test applies a deshaking filter to a source video at 3,840 x 2160 (4K UHD) before scaling down and outputting at 1,920 x 1,080 (1080p).

HandBrake Video Transcoding

We perform a simple encoding test using the freely available HandBrake software, which converts a 60-second, 400MB, 4K, MKV video sample to a high quality, 1080p, MP4 file using the available presets. Our results show strong scaling with increasing threads, which is indicative of high-performance video editing suites, albeit without any GPU acceleration.

PCMark 10 Photo Editing

Terragen 4

Website:Terragen 4

Planetside Software’s Terragen 4 is a highly realistic landscape generator used to create background images in films and games such as Star Trek: Nemesis, Stealth, and The Imaginarium of Dr. Parnassus. Our script renders a single frame of a snowy mountain scene at 640 x 480 on all the available CPU execution units.

Cinebench R15

Website: www.maxon.net

Cinebench uses Maxon's Cinema 4D engine to render a photo-realistic scene of some shiny balls and weird things (we miss the motorbike). The scene is highly complex, with reflections, ambient occlusion, and procedural shaders, so it gives a CPU a tough workout.

As Cinema 4D is a real-world application - used on films such as Spider-Man and Star Wars - Cinebench can be viewed as a real-world benchmark.