Best Of

@NormanPCN wow. Some good info in that writeup. While you've obviously only tested your own machines I still think you've tested more thoroughly than anyone else. For example I've never really noticed the render client being a transient process.

Thanks Mike.

1. I do not have a webcam, so can't do that.
2. I am a bit loath to tell others what is so, unless I have tested it. At least a quickie test. Anecdotal is fine with enough trusted sources in rough agreement. I have only tested on my machines. Prev a i7 4770K (4C/8T) at a flat 4Ghz (non turbo) , 16GB ram (1600Mhz), GTX 1080, 7200rpm 1TB spindle drive (256GB SSD boot drive). Currently an i9 9900K at a flat 4.8Ghz, 32GB ram (3200Mhz), same GTX 1080, 2TB PCIe NVMe TLC SSD.
3. I just have not done enough with Hitfilm for general recommendation. Beyond a forum post getting me to work up a test project and thus gain some knowledge.

That said.

• CPU: Clock rate, clock rate, clock rate. Given all the Hitfilm front end, and single thread bottlenecks, then a faster CPU core is better. I would want 4Ghz min. For me I did notice a descent bump in perf of the particle simulator moving from my 4Ghz 4770K to the 4.8Ghz 9900k.
• Cores: For HD then 4C should work well. For 4K work then 8C. FWIW Vegas has always recommended 8C for 4K work and in an old post of me testing UHDp30 up to 4 stream compositing (AVC, Cineform, Prores, DNxHR), I could see why. This was on the 4770K machine. Hardware decode of AVC 8-bit 4:2:0 media takes away some of the 8C need for 4K. The encoder will still be happier. Certainly the AVC(MP4) encoder and maybe Cineform. Most other encode codecs are a single thread or a few threads (so clock, clock, clock). Laptops often quote a sexy turbo/boost frequency. That cannot be sustained due to heat and power. Laptops just cannot dissipate enough heat. Editing playback and certainly encode/export will have the CPU fall down to its spec'd, probably quite low, clock rate. The lack of Hitfilm performance here can help(!) the CPU boost clock rate, especially if HW AVC decode is being used. In this context I refer to the inability to fully utilize HW resources (utilization).
• RAM: I just do not have good projects to tell this properly. One can get a good benchmark just looking at the Window task manager of Hitfilm mem usage with a suitable project open. Take the sum of all processes under the Hitfilm task manager hierarchy.
  • RAM preview: This is a duh. Easy to compute. The timeline duration one a RAM preview is dependent on the frame size and frame rate. AKA The number of frames buffered. Now Hitfilm is kinda stupid here. RAM preview is always using full frame size and paused quality. So you do not get any benefit here if you set your working resolution to 1/2. Half cuts RAM needs to 1/4 of full. So the Hitfilm way is need more ram, and sit a twiddle your thumbs waiting longer for a ram preview to finish so you can see your work.
• Hitfilm has a render client. A background render executable that is used for pre-renders and exporting. If running this you will need/want have some RAM headroom for that to activate. It is transient. Only running while one of the mentioned tasks is running. The render client does not want as much ram and Hitfilm "proper" but I should be proportionate. This is where my lack of test projects hurts my knowledge on this specific detail. For Export we are not likely trying to edit/work while that is running since it will dog out edit performance so it is probably not a concern here.
• GPU: I have had the GTX 1080 for a long time. I had a GTX 980 briefly and before that an AMD 7950. I cannot remember, nor did I likely test, the amount of benefit here. I would say, 4K, all the GPU you can get for anything beyond simple. Nuff said. Of course, I did not really say anything there.
• Storage: SSD all there way. Especially with 4K and Cineform/Prores/EXR/PNG media. My old machine UHD showed this. I was on a 7200rpm spindle. I can't remember but at 3 or 4 simultaneous streams I was technically beyond the speed for the spindle drive. I forcibly cached the media into the system disk cache to work around that. I also temp moved the media to the SSD drive and saw the same perf as cached.
• I would try to stick with TLC SSD storage for common use. QLC SSDs, which cheaper for GB, do not have the write performance in a heavy work environment. For mostly read media, then perfecto. Of course for heavy work/write use then MLC SSDs are best but they are pricey. MLC = 2 bits per flash cell, TLC = 3-bits, QLC = 4 bits. For the price conscious QLC flash can still be worthwhile if you don't have heavy contig writes. They have MLC caches to hide the perf hit. In Hitfilm terms, exporting an AVC(MP4) file, at common/upload bitrates, is not really a heavy write condition. the MB/s is not that much. Again, without testing I have a bit of a hand waving argument here. FWIW when building my current PC I was going to go with/try an Intel 660 QLC SSD PCIe NVMe since at 2TB it was a good price at the time. The perf was way more than I cared about. Less than TLC but in real terms, still crazy good. Then along came a Sabrent 2TB TLC SSD PCIe NVMe on sale at AMAZON and I jumped on that.

For most users out there I would say get over UHD/4K. Unless you have a power machine. The edit experience will just be smoother. There are just a ton of pixels to process at 4K. That and delivery/viewing formats/bitrates typically do not do 4K true justice. There is the whole Youtube better encoder setting(s) for 1440+ rez but that is a whole other topic.

--

I think I am just going to stop. Any questions, hit me up.

Thanks gents,

then this it was what I got wrong.

Now I will have to twiddle with the camera settings I guess. It's great to have all those different views, but it still is confusing at times to get the right preview of what is happening.