We've been using Arriraw HDE files from Alexa LF cameras for a while now. Like, over 3 years (?).
I'm referring to Arriraw coming out an Alexa LF camera, which, with HDE enabled in the Codex device manager, present themselves as .arx file sequences.
When we used Resolve 17.x to convert to DNxHD36 or ProRes Proxy dailies, I kind of remember it used to go at real time or faster with the .arx media on an external Thunderbolt 3 NVMe SSD (so the drive speed is fine).
With Resolve 18.x it kind of, 'feels' slower on the same systems.
Like, it starts out at 30 even 40 fps, but quickly slows down to 25 fps and then settles around 17-18 fps. On long timelines, it crawls and 'dries up' to 3-4 fps. If one stops the render, marks in from the last clip and starts render again, the speed improves, but then again 'dries up' after about 30-40 min render.
System in both cases is the same MacBook Pro Intel 16" 16GB RAM, 6-core i7, Radeon 5300 4GB.
A MacBook Pro Intel 16" 16GB RAM, 6-core i9, Radeon 5500 4GB performs somewhat better, but that too 'dries up' on long timelines. Same test done on a MacBook Pro M1 Pro 14" with 16GB RAM, also runs at 14-15 fps. All with Resolve 18.x. Same systems were used with Arriraw HDE using Resolve 17.x some years ago, and they 'felt' faster.
With the exact same media, same project, on a new MBP M1 Max 16" with 32 GB RAM, 10-Core CPU, 32-core GPU, 32 GB RAM, to convert to DNxHD36 or ProRes Proxy dailies, Resolve 18.x absolutely flies, and the same Arriraw .arx processing works fine even on hours long timeline. Even on an Intel Mac mini w 32GB RAM, and an eGPU RX 5700XT 8GB, the same render works fine. On both, the speed is a consistent 30-40 fps.
As you can see from the numbers above, it doesn't matter Intel or M1 Pro or M1 Max. Its the RAM that seems to matter. So, I'm wondering, does Resolve 18.x need more RAM than Resolve 17.x did? With Resolve 17.x, 16GB RAM got more work done than the same 16GB RAM on Resolve 18.x.
I haven't done systematic side by side comparisons, but wondering if anyone has the same experience.
Arriraw HDE runs slower on Resolve 18?
-
- Posts: 51
- Joined: Mon Dec 14, 2015 9:40 pm
- Contact:
I have experienced terrible throttling on the 16" Intel i9 MacBooks, similar to what you experience, don't know if this is what you see. HDE uses a lot of CPU cores to decode.
You can install this small util to see if this is the case https://github.com/macmade/Hot. It will show you CPU temp and throttle level in percent in the menu bar. Makes it easy to see what is going on.
My 16" i9 throttled to 30% after a few secs into the renders. from 120 fps to 40 fps with ALEXA HDE material from Compact drive to External SSD. Stopping the render would cool the CPU down and it could render at full speed for a few secs.
14" M1 Pro doesn't have a fan, but the 16" M1 Pro does, so the 14" M1 Pro might be throttling as well as there is no active cooling.
You can install this small util to see if this is the case https://github.com/macmade/Hot. It will show you CPU temp and throttle level in percent in the menu bar. Makes it easy to see what is going on.
My 16" i9 throttled to 30% after a few secs into the renders. from 120 fps to 40 fps with ALEXA HDE material from Compact drive to External SSD. Stopping the render would cool the CPU down and it could render at full speed for a few secs.
14" M1 Pro doesn't have a fan, but the 16" M1 Pro does, so the 14" M1 Pro might be throttling as well as there is no active cooling.
Anders Holck DFF
Instagram/Facebook: andersholck
info@andersholck.com
+4526799726
![Image](https://forum.arri.com/download/file.php?id=338)
Instagram/Facebook: andersholck
info@andersholck.com
+4526799726