While "fast allocate" method takes more time to allocate files (than with "sparse"), that files are still sparse, and are very fragmented after downloading. I completely removed torrent with files, set file allocation to full prewrite, then add it back - it takes much longer to allocate but files created still have sparse flag. Maybe it is an old bug, because I found this topic ( https://forum.tixati.com/support/3056  ) with similar problem.

what OS and what version of Tixati?

it all seems to work correctly on linux, version 2.64.

I should have done it in first message.
Windows 10 version 1909, 64-bit
Tixati 2.64 64-bit
on NTFS
Also I tried rebooting and using another HDD volume but it didn’t help.

I compared Tixati allocation with µTorrent one.
Same torrent (~24 GB nearly all in one file), same NTFS volume (44% full), pre-allocation enabled in µTorrent, "fast allocate" in tixati.
Right after allocation: Tixati - 2957 fragments (sparse flag present), µTorrent - NO fragmentation (and no sparse flag). Fragment count observed via Defraggler.

Windows: "Fast allocate" didn't seem to work for me too, I switched to "Full prewrite" since. The name fast allocate suggests the usage of fallocate that's a unix tool, I don't know whether it's actually working at all on Windows as it is now.
There's certainly a need for Windows-custom code to make it work, not sure it's actually what Tixati is doing:
- https://stackoverflow.com/questions/53334343/windows-refs-ntfs-file-preallocation-hint
- https://stackoverflow.com/questions/7970333/how-do-you-pre-allocate-space-for-a-file-in-c-c-on-windows

I recently moved all files from one drive to another using Tixati (worked like a charm actually! much better than uTorrent did at only fraction of files/size) and this topic made me check fragmentation. I expected all moved files to NOT be fragmented at all, but some of them are. Mostly multi-file torrents were copied in a fragmented state, single-file torrents appear to be alright. This here could be an OS problem (Windows), not Tixati.