Bro, provide your script ?

This is a jlp kangaroo, with precalculated tame kangaroos.
You can make them yourself. By running a kangaroo in the required range with the -m 3 parameter in a loop with a false public key. In this way you will accumulate a lot of DPs.
After that, you will only need wild kangaroos.
This requires modification(s) to JLPs script.
i see you people solve much much faster with saved DPs. good trick, but how big data grows until merge is slower?
There is no slow down by using pre-compiled workfile(s). Well, I should say, there is no slow down unless you have a very low budget CPU. There should be no "merging" required for an 80 bit range. Pre-compile the tames (convert all wilds to tames), in a single work file, and use the -i to load and save to it; no merger required. What DP did you use?? I think, maybe people are using to low of a DP. I used DP 20 and it did now take any time to load the workfile. A low DP is not needed since you should have many more tames needed to solve, quicker. When I ran it, for the actual challenge key, it took less than 30 seconds using a single RTX 4090. I think the workfile size was right at 200 MB.