Does on mean that /sys/kernel/mm/lr_gen/enabled returns 0x7?

Just set vm.page-cluster to 0, to disable the page read-ahead (AFAIK the default value was set when swap was still on rotating disks).

Arch wiki has some good options for tuning the VM

Debian does not enable it by default, cat /sys/kernel/mm/lru_gen/enabled will be a 0.

I am running Debian 12 on all of my devices with Debians vanilla kernel! :-) Just enable MGLRU on Debian like it is described in this blogpost.

One further tip for ZRAM: On my device the LZ4 algorithm was noticeable faster than ZSTD (didn’t try ZSTD with the enabled MGLRU, yet) and it was important to disable the RAM page read-ahead on my device.

deleted by creator

I don’t know either, but unless one uses zstd (lzo seems more like a thing for this hardware), I would hope that it is totally usable. (Running zstd memory compression on a Raspberry Pi 2, w/o any noticeable speed impact)

Show me some numbers! ;-) … Perhaps I miss something, but basically we have 32bit pointers vs. 64bit pointers, the rest of the data should be the same size. 64bit should be faster for tasks where the CPU is the bottleneck/computations, so IMHO it will be an interesting tradeoff with no clear winner for me.

The most important thing is not the distribution, but to enable ZRAM (or ZSWAP) and use a lightweight desktop. I am not sure how much difference a 32bit vs a 64bit distribution makes, but if possible you could take one for the team and run some trials and report your numbers (RAM usage) back here.

Of course I recommend Debian with a lightweight desktop of your choice, or Alpine.