This post was last modified: 2020-07-03, 04:40 by martinheck.

With the recent LR 9.3 update the LRT-Exporter is bottlenecked.
CPU usage jumps between 50% and 100% in a ~10s interval. This leads to about a 50% reduction in render speed.
I did extensive testing to rule out any SSD performance and Temp-Folder/Windows Explorer/Scheduler issues.
I thought that the 32/64 threads from the LR-Renderer might overload the temp-folder/LRT-Exporter process but it works perfectly fine with LR 9.2 and the same hardware. In fact the R/W intensity of the whole export process is max 60MB/s read 200MB/s write for me. Easily to handle by any SSD.

Input: 5K RAW (Canon 6D)
Output: 5K TIFF 16bit, uncompressed

LR9.3 stock export speed: ~2fps
LR9.3 LRTExporter: ~1fps
LR9.2 stock export speed: ~2fps
LR9.2 LRTExporter: ~2fps

Workstation-Specs:
AMD TR 3970X with HT enabled
System and Software on C: (Gen4 nvme ssd) 2TB
Win-Temp-Folder, LR Cache, Input and Output folder on D: (Raid0 2xGen3 nvme enterprise ssd) 16TB
256GB RAM, Titan RTX

Hi Martin, thanks for reporting this - I've done some tests with LR Classic 9.3 and currently I cannot reproduce your findings.
I'm working on a 16/32 Threadripper, NVM SSD etc. and can easily max out my processors by exporting with the latest LRTExport on the latest LR Classic. See my attached screenshot.
Can you maybe uninstall LRC and reinstall just to make sure nothing got stuck on your side?

FYI: LRTExport doesn't export uncompressed tiffs as you wrote, it uses ZIP compression. This might be relevant for comparisons.

I re-installed LRC9.3 and it seems to have fixed the issue. You never try the easiest solution on your own. Classic

Looks decent now. :-)