How did we get here?
Having completed most of the work on CoolBlue, the new editing rig, it came time to start testing. Theoretically it should be better, but that still needed proving. This would primarily be a test of the CPU, because the version of DaVinci Resolve I’m using doesn’t have support using the NVENC encoder used by NVidia. I’ve been using it like this for a couple of years, but there is a wrinkle later.
I devised the rather unscientific procedure of three test runs each on identical projects. I was able to do this because the files are now kept on the server, and available to every computer on my network. It would be three passes with each the AMD Ryzen 7 5800x and the Intel i7 12700KF. I would render a twelve minute video, alternating between the two systems allowing a cooldown between each run.
Both systems are similar with anything handling the media being nearly identical. Both systems are using an intel NVMe drive, with other assets stored on WD Black 7200 RPM. Each saves to a drive on the home network to the server, I recently built. The GPUs aren’t a factor (yet), and the DDR4 memory is 32 GB at 3600 MHz. The only difference is the platform.
The testing
The first run was almost what I expected, except the AMD was twenty seconds faster. I started to question my idea of swapping rigs at this point, but there was more to test. The 5800X clocked in at 4 minutes and 34 seconds, and the Intel rendered took twenty seconds longer. The second run saw them reversed, with the 12700 turning in the faster time. The average of both runs were almost identical, I could go with either system at this point.
That changed with the third run. The AMD rig turned in a time similar to the other runs, but the Intel system rendered the video in three minutes, nineteen seconds. It shaved a full minute off anything the AMD offered. Both of these systems are the previous generation with the same generation of memory, and the same storage solution. It would be a no brainer to use the new build, but there was more.
I mentioned that the version of Resolve I am using doesn’t take advantage of the RTX3060Ti that is in the current editing rig, but it does allow for the AV1 encoder on the Arc A750. The same A750 that is in the new build. Might this be the test I need that convinces me to switch?
AV1 in the new editing rig
One advantage of the Arc graphics cards is the addition of the new AV1 encoding process. An encoder that is also now included on the new 7000 series AMD Graphics cards. After a small misstep, I was able to configure Resolve properly, and I ran the render test, which blistered the previous efforts.
The new run took one minute and fifty two seconds. My fastest run was now cut almost in half! Half! I was convinced. I used to start rendering, then leave the room to go get something to drink or a bio break. Now, I will barely have time to find which image will be my thumbnail for the video. This will save hours over a year’s time. It will pay huge dividends as I move forward.
For good measure and piece of mind, I rendered the support video for this story using the same method and the eleven minute, seven second video, finished in two minutes even. It’s a huge improvement and it makes me excited using this editing rig going forward. It does raise a question of whether I should completely swap PCs or have two systems for two different functions, though, but that’s a question for later.
The video for the testing and results is here.
Back to the blog.