I built a new computer with an Intel i7 8700 (six cores) because my old i7-3770 was just too slow. This new computer will mostly be used for transcoding video with Handbrake, my favorite transcoding program (which is also free). I cannot say enough good about this program and there is no shortage of online help.
In addition to my old desktop computer with an Intel i7 3770 CPU, I also used a laptop with an Intel i7 6700HQ for transcoding. Laptop CPUs are generally not as fast as desktop CPUs of the same model line, but this newer laptop CPU is a bit faster at transcoding than the older i7 3770.
I am not experienced with benchmarking so take this on face value. I transcoded large MKV videos to MP4 (H.264) over a gigabit network from one Synology NAS (network attached storage) to another Synology NAS. I am doing it this way because I have different types of storage in the computers and I wanted a consistent testing environment.
I transcoded the videos with the following custom Handbrake settings:
The test movie was in MKV format, had 1080p resolution, and was 1 hour 55 minutes:
- The desktop computer with an Intel i7 3770S took 1 hour 38 minutes.
- The laptop computer with an Intel 6700HQ took 1 hour 28 minutes.
- The desktop computer with an Intel i7 8700 took 54 minutes 58 seconds. This transcode test process at an average of 60 frames per second (FPS).
A friend with a ten core Intel Xeon E5-2650 v3 transcoded the same video over a gigabit network from a Synology NAS (same model as mine). His transcoding was also around 60 FPS, but fluctuated a lot more than mine. I think his systems had more going on so it was not an apples to apples test. Though he did not let the process run to completion, it still did not appear it was going to be faster. This CPU was also released in Q3 2014 and the i7 8700 is a four years newer.
I would eventually like to transcode large MKV files from Blu-ray rips with something like H.265, VP9, or the long awaited AV1 to save storage space, but maintain original quality (if possible). Though I am not quite ready to go with replacing my direct rips with transcoded files, I decided to test how the i7 8700 handled H.265. I selected an MKV that was 33.4GB with six audio tracks. These audio tracks were likely quite large since they were 7.1 audio, DTS 5.1 audio, AC3 5.1 audio, and three more tracks. I set all audio tracks to Auto Passthru so the sizes were not reduced (the test file had spectacular audio). I used Handbrake with the Matroska H.265 MKV 1080p30 setting, which has the encoder preset at slow. The transcode process took 2 hr 15 min and the resulting MKV was 10GB. Handbrake averaged 21.8 FPS transcoding over my network (from one NAS to another NAS).
I believed the majority of the test file size was from audio so I decided to transcode again. I used the same settings as above, but I only selected one DTS 5.1 audio track that I set to Auto Passthru. This same track was also used to create a stereo track since I like having one in all video files. This transcode took 2 hr 16 minutes to complete and the resulting file size was 2.76GB. The resulting quality looked as good as the original when I sampled various scenes.
Either way I encode these files H.265 will still be a significant size savings while maintaining great quality. The only drawback for me is the slow transcoding speed compared to normal H.264 transcoding. H.265 does also playback compatibility with some devices, but I almost always use the H.264 versions for playback on Plex.
I decided to try the H.265 (Intel QSV) transcoding in Handbrake that uses the Intel GPU. I first used the “Encoder Preset: Balanced” and it only took around 45 minutes to transcode a two hour movie at an average of 81 frames per second (FPS). The original file was a direct rip from a Blu-ray and it was 29.5GB. The resulting H.265 transcode was 3.03GB. I decided to tweak the settings and changed the Encoder Preset: from Balanced to Quality. This process averaged 48 FPS the resulting file was 4.4GB. I still need to watch samples from each movie and see if there is an obvious difference. The interesting thing about the Intel QSV setting was that it barely affected the CPU since it was using the GPU.