[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Release status of i386 for Bullseye and long term support for 3 years?

Hello.  32-bit Pentium 4 user reading this out of personal interest throwing in a comment.

If the choice is to primarily support pae or non pae for 32-bit moving forward, then I suggest non-pae for the reason that everyone can use it.  If you have more than 4GB of memory you probably have a 64-bit cpu anyway and really should switch to a 64-bit kernel, and there is no performance gain by using pae with less than 4GB.  I feel the only people who should use a 32-bit kernel with bullseye should be those without 64-bit support at all, and comments in this thread seem to agree.  The popcorn results show most people using pae, but I suspect that is just because it is default, I am too, but only have 512MB so certainly don't need it.

As to the kernel version and spectre/meltdown mitigation I think 4.19 in buster already resolves that.   A possible problem with using a newer kernel for old 32-bit machines is they may not support the drivers we need, at least in my specific case, nvidia-304 proprietary.  From nvidia directly it wont even build for kernel 4.15; with community patches I can build for 4.19 and is also in buster repos.   But it won't build on 5.4 for me (as-is or with patches that work for 4.19).  So it makes me concerned if it will be available to me in bullseye because whoever packages it would have to do extra work which may or may not be trivial.  If upgrading to bullseye means no more nvidia driver, it makes it less appealing for me.  Although I have read 5.4 improves floppy disk speed, something 32-bit users are more likely to still have, but I doubt use very often.

Thanks for continuing to support 32-bit in any form, excited to see what happens.

Reply to: