Ok, so I understand that there’s a general requirement for a score of 2000 per 1080p transcode but i then see reference to the use of HA to offload this work but ne mention of the capabilities of QS.
Is there an Intel variant of this sort of table (showing nvidia gfx transcoding capabilities) in relation to the CPU with QS HA?
Intel’s iGPU is subject to several external factors including driver, OS, shared memory, cpu model, 3 different iGPU packages. With Intel about to release discrete GPU’s the memory will at least be separated from the OS.
With 1080p h.264 material, Intel is on par with the Nvidia. Some OS’s limit the shared ram to 1536 MB and that will be what caps the streams with h.264 and more so with h.265(HEVC). I have a low power chip with a 4000 and without QSYNC can only handle two 1080 streams, I was able to handle 8+ with HW.
Because Audio and Subtitles are still processed by the CPU, don’t overlook the Single Thread Rating, with the lossless audio I would not drop below 1800.
You’ll also need somewhat decent cpu power when “burning in” subtitles.
If you use subtitles (even if they’re just ‘forced’ ones), and you don’t convert them into SRT but leave them as PGS or VOBSUB format, you can expect these needing “burning in” via CPU on almost every client type (with nVidia Shield and the dedicated Plex desktop player apps being the exceptions).
