How to optimize RawTherapee?

Software RawTherapee
21

@chaimav

In my fork I have experimentally built a RawTherapee for windows using skylake-raptorlake as the target processors, should be tuned to run fast on your i7. Further speed up is possible with the use of link-time optimization but that feature is not currently working for the windows build. (it does work on Mac and Linux)

See if you can give the release version a try:

Workflow link:

about_this_build:

Version: nightly-github-actions-810-g2a8e549b7
Branch: fastwin
Commit: 2a8e549b7
Commit date: 2024-07-27
Compiler: cc 14.1.0
Processor: skylake-raptorlake
System: Windows
Bit depth: 64 bits
Gtkmm: 3.24.9
Lensfun: 0.3.4.0
libjxl: 0.10.3
Build type: release
Build flags:  -std=c++11 -ffp-contract=off -march=skylake -mtune=raptorlake -Werror=unused-label -Werror=delete-incomplete -fno-math-errno -Wno-attributes -Wall -Wuninitialized -Wcast-qual -Wno-deprecated-declarations -Wno-unused-result -Wunused-macros -fopenmp -Werror=unknown-pragmas -O3 -DNDEBUG -ftree-vectorize
Link flags:  -march=skylake -mtune=raptorlake
OpenMP support: ON
MMAP support: ON
Build OS: MINGW64_NT-10.0-20348 3.5.3-d8b21b8c.x86_64 x86_64
Build date: Sat, 27 Jul 2024 13:02:36 +0000 UTC
Build epoch: 1722085356
Build UUID: efe7a454-0778-4f9a-90bc-74f2f1a12109
22

Hello @HIRAM

Just tested myself you build (stand-alone) on Windows 11.
Processor: 12th Gen Intel(R) Coreā„¢ i7-12700H (20 CPUs), ~2.7GHz
Memory: 32768MB RAM
Card name: NVIDIA GeForce RTX 3070 Ti Laptop GPU

It looks super fast.
No crash whatsoever so far.
Thanks a lot indeed!

1 Like
23

@Silvio_Grosso appreciate your testing the product! Iā€™ve only access currently to a Dude, you got a Dell! skylake machine running windows 10 and it also seemed faster than generic. I also have a Ryzen 9 5k to test on when I get back home in a couple weeks.

24

I tested with just a previous edit and just panning side to side. Using a crude timing method (stopwatch on my smartphone) I did multiple pans.
With threads set to 0, the processing bar showed up for about 2.5 seconds, but with threads set to 16 it was there for about 1.4 seconds.

@Silvio_Grosso can you please turn on Wavelet Levels > Sharp-mask and Clarity and try panning the image while zoomed in. And then try it again with Threads set to a higher number.

25

Hello @chaimav

Just tried on an Acer laptop (Windows 11 - Kaspersky antivirus suite running in the background)

As usual:
Processor: 12th Gen Intel(R) Coreā„¢ i7-12700H (20 CPUs), ~2.7GHz
Memory: 32768MB RAM
Card name: NVIDIA GeForce RTX 3070 Ti Laptop GPU (RawTherapee does not take advantage of the GPUā€¦)
SSD: 1 terabyte - NVMe SAMSUNG MZVL21T0HCLR-00BT7

I tested some RAWs images from a Nikon D850 (NEF, around 45 Mb each)

As per your suggestion I activated: Wavelet Levels > Sharp-mask and Clarity

In all truth, I didnā€™t notice any difference, while panning (at 800% zoom), with the default settings (Performance > threads) versus, later on, increasing them to 20

26

OP is not testing the actual panning speed, but the time it takes for the background processing to complete when you release the mouse after panning.

1 Like
27

Hello @HIRAM

OK. My bad! :slight_smile:

I have repeated the test but this time I have released the mouse while panning in order to complete the OP every time.

The processing time was slower with 0 threads set in the performance. It was much faster with 20 threads with pretty much no delay (as expected, I supposeā€¦)

Here is the recording (OBS running in the background)

1 Like
28

Thanks. That is exactly what I am seeing (although yours is even faster than mine, probably due to the extra RAM).

So my conclusion is that leaving therads as automatic is not as fast as manually raising it. At least for modern high end Intel CPUs.

1 Like
29

Hello @chaimav

although yours is even faster than mine, probably due to the extra RAM

Yeah. I suppose soā€¦
32 Gb RAM was the minimun for me (in the past I run Windows computers with 16 Gb and I deemed them slow, especially for video editing stuffā€¦)

30

This Threads = Auto is not good enough for highly parallel cpus thing should be raised as a separate issue on github.

1 Like
31

Itā€™s on my long term plan to upgrade RAM, but I just dropped a chunk of change on this computer and not ready to spend more at this time. The most taxing thing I use it for is RawTherapee. Besides that, itā€™s basically web browsing.

I got it Prime Day for $680
ASUS ExpertCenter D5 Small Form Factor Desktop PC, Intel Core i7-13700, 16GB DDR4 RAM, 1TB PCIe SSD, Windows 11 Pro, Black, D500SE-AH704 Amazon.com: ASUS ExpertCenter D5 Small Form Factor Desktop PC, Intel Core i7-13700, 16GB DDR4 RAM, 1TB PCIe SSD, Windows 11 Pro, Black, D500SE-AH704 : Electronics

32
1 Like