Speed of Rendering: What affects it?

I have tried Davinci Resolve Studio on two different systems. One with an old Nvidea discrete graphics card, and one with AMD integrated graphics. Benchmarks show these systems to be on a par with each other.

I took a simple video and rendered it using the settings below.

The render time between the two systems is shockingly different. But in both cases, Resolve was not using all of the CPU or GPU resources available.

What is going on?

AMD Ryzen 5 3600X Base clock 3.8 GHz, turbo 4.15 GHz
AM4, 1 CPU, 6 cores, 12 threads
Nvidia GTX 1050-Ti
Nvidia(10DE 1C82) ≥ 4GB
Render time 5’12”

AMD Ryzen 7 5700G Base clock 3.8 GHz, turbo 4.4 GHz
AM4, 1 CPU, 8 cores, 16 threads
AMD RX Vega 8 5000 (iGPU)
AMD(1002 1636) 512MB
Render time 26”

Settings

2021-10-21_cr.png (17.31 KiB) Viewed 973 times

Performance

2021-10-21 (2)_cr.png (9.64 KiB) Viewed 973 times

You can't use AMD hardware encoding with 3600x because it doesn't have an IGPU, you would use the Nvidia hardware encoder. I suspect you are hardware encoding with 5700G, and software encoding with 3600X, the huge difference you're seeing because you chose to encode h.265

You can't use AMD hardware encoding with 3600x because it doesn't have an IGPU, you would use the Nvidia hardware encoder. I suspect you are hardware encoding with 5700G, and software encoding with 3600X, the huge difference you're seeing because you chose to encode h.265

I don't expect to use the AMD encoder on the 3600X, I expect to use the NVIDIA encoder in the GTX 1050-Ti.

For h.265 the only choice is NVIDIA (takes 5"12")

For h.264 the choice is NATIVE (takes 5'12") or NVIDIA (takes 5'13").

Odd they are all so similar.

The fastest encodes happen if your target storage is different from the one where the sources reside.

The fastest encodes happen if your target storage is different from the one where the sources reside.

That may be true, but it doesn't explain what is going on here.

Maybe I'm missing the question but if you think about the process

Computer reads the file and decodes it.

It then hands it to the encoder that writes it out.

Best speed means a fast enough decode. Fast enough I/O. Fast enough just means no slower than the next step of the chain.


I assume on both systems you're using the same GPU to decode and encode the file. If you're using the same drive that's even more bandwidth sharing.

If you're asking why the AMD is so much slower than Nvidia the other thing to consider are the settings. The encoder if left to the defaults doesn't always have the same settings. I don't have an AMD system to check but comparing Nvidia to Intel the settings are different. Rate control etc.

If you really wanted to compare the two systems you would need to make sure the settings are matching to the closest possible level.