Sattva Descreen Plugin PC/AMD/Intel vs Apple Silicon M1/M2 Speed Benchmark

[Mr.V600]

This is to demonstrate the difference in speed between using Sattva Descreen Plugin on Windows PC's with traditional x86 processors vs Apple Silicon on MacOS.

This is not meant to be a scientific comparison between the 2 platforms, but only to validate the claim that Descreen on Apple Silicon is considerably faster.

In testing the M1 and M2 both showed considerable advantage from Windows PC's. Descreen plugin shows to primarily only use one Core when processing, regardless of platform. Although on the 3700X it looked like the thread scheduler was doing a round robin across multiple cores as the plugin ran, but never using at the same time. Also, my experience has been that results with descreen seem to be pixel size sensitive, not content sensitive. The amount of content needing descreen on the image doesn't affect the time taken....the size of the image itself does.

For the image testing, I used some sizes that are common for me in scanning album art. Resolutions are at 800DPI.
Just the front of the CD booklet
Front: 3761px x 3761px

A CD booklet spread open
CD Booklet: 7593px x 3817px

Most of a 12" vinyl record sleeve (still requires 2 passes even on A3)
Vinyl: 9240px x 10089px

Full A3 Scan (for a poster foldout)
A3: 9308px x 12611px

A physical comparison of the pixel dimensions

Results. Time in seconds. Lower is better.
The Mini M2 dominated. Even the slightly older and thermal limited Macbook Air M1 performs very well and easily beats the AMD Ryzen 7 3700X.

graph_front.jpg (66.84 KiB) Viewed 14 times

graph_booklet.jpg (67.75 KiB) Viewed 14 times

graph_vinyl.jpg (59.08 KiB) Viewed 14 times

graph_a3.jpg (66.12 KiB) Viewed 14 times

Comparison of times just for the Mac Mini M2 to see how the canvas size scaling looks

graph_mini.jpg (59.21 KiB) Viewed 14 times

And raw data

raw_data.jpg (45.12 KiB) Viewed 14 times