New
#11
Context info
I'm long suspecting about file caching failures in Windows 10 and 11 that produce bad compressed files, although it could also be RAM or PSU (Win10 has a 50€ Cooler Master MWE White V2 700W from which I caught good detailed reviews) and a lot less probably storage lacking quality.
I use WinRar frequently to compress the Thunderbird profile with my emails (a folder of about 550 MB that gets compressed to a 170 MB rar file with stock settings). I've got two Win7 desktops (one very stable, other not so much) that never ever fail with WinRar, and one Win10 desktop and one Win11 miniPC that fail a small percentage of times (the Win10 one was failing more months ago, with different memory modules and fewer patches applied; in Windows 11 I tried setting WinRar to produce zip files instead, that's a much lighter task, but it didn't avoid some failures).
No error messages or anything let alone BSODs, the fails are silent. I noticed it the first time when trying to use a defective file, and since then I always check produced rar files with the WinRar context menu option. In Windows 10 I've noticed a pattern: if a bad file is produced, which currently might be happening 5-10% from the attempts (it was more like 30-50% months ago), its likely that I need a row of more than 2 attempts (2nd and further attemps are a lot quicker, specially in Win10, I suppose that for file caching). Yesterday it was 4 or 5 with a Thunderbird manual "Compact folders" in the middle, exaggerated for even the "30-50%" failures age.
In such age I emailed WinRar's technical service for a possible bug in WinRar but they said that the programs's extensively tested and the most likely culprit was RAM w/o discarding my storage, a Kingston A400 240GB SATA3 SSD in Windows 10 and a soldered eMMC in the W11 miniPC. The latter has soldered RAM (8 GB), Win10 had 16GB of high-end DDR3-1866 (JEDEC) @1866 in its worst age, I added other 16GB or DDR3 (1600 JEDEC 1866 XMP) totalling 32GB @1600 what improved WinRar hit rate, and time afterwards I replaced the initial modules with others like the second ones and they keep on running @1600, I cannot tell if this move has improved WinRar or not, as the failure rate is similar.
I do not game with any of the computers. Win11 is slow (Celeron J4125) but it doesn't have any other concrete instabilities or bugs. Win10 has other one: Windows Media Player uses to crash every hour or so when playing music from a SATA HDD (suddenly it says something about an error when trying to play a new song and one or two GUI things get corrupted, it always happens when going from one song to another). Win10 also wasn't able to install KB5034441 after 30-50 attempts of all kinds, although it took the equivalent Dynamic Updates version in the first attempt. Sometimes I go to news sites, open 10 or 20 tabs and read them, this is stressful as these pages have a lot of advertising, although both computers cape with it w/o problems. Never ever a glitch when doing partition imaging with "high" compression or looking for Windows updates either. The Win10 processor can stay over 80コC while doing this and the Win11 throttles at about 70-75コC, but neither fail.
----------------------------------
I'm not asking for help here, these issues are light and the systems work very well despite them. But I think that there's a possibly overlooked problem in Windows 10 and 11 regarding, at least, file compression.