[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: Features/ArchitectureSupport - changing what we build for
- From: Benny Amorsen <benny+usenet amorsen dk>
- To: Chris Adams <cmadams hiwaay net>
- Cc: Development discussions related to Fedora <fedora-devel-list redhat com>
- Subject: Re: Features/ArchitectureSupport - changing what we build for
- Date: Sat, 31 Jan 2009 16:12:09 +0100
Chris Adams <cmadams hiwaay net> writes:
> IIRC, I asked about this a few years ago and got told that there was a
> performance hit (not big but measurable) for running the 64 bit kernel
> and 32 bit userland. Maybe that was wrong or is now outdated.
It should be the other way around -- 64 bit kernel should be faster.
The only downside is that the kernel may take a little more memory,
but that isn't really a problem for non-embedded systems.
> What are the chances GRUB/syslinux could be taught to look at a few
> special CPU flags (can you fetch the important flags from real mode?)
> and only present "compatible" (as determined by config arguments, not
> GRUB/syslinux poking into kernels) boot options? That would help the
> LiveCD case as well. You could have both 32-PAE and 64 kernels on the
> CD and choose at boot time, although that would take up a large amount
> of space (so maybe not feasible for LiveCD).
That would be rather neat. Tricky, but neat.
/Benny
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]