Is this a "pre-announcment"? or a reference to someone else's work (i'e' Pemasu)?TEASER
Anyway, looking forward .....
______________________________________
@Iguleder shares these great Benefits. LightHouse64 has demonstrated these.Iguleder wrote:My advice to you, playdayz: make a x86_64 flavor of this kernel.
This has five benefits:
- Access to 4 GB (or more) RAM, without the PAE issues (e.g instability under certain circumstances).
- The ability to run x86_64 binaries (e.g compile x86_64 stuff from a chroot environment).
- The performance boost of 64-bit computing - e.g some heavy computations (such as compression) implemented in kernel mode are now faster.
- Same userland - no need to recompile anything or create a separate package repository, as with i486/i686 switching.
- Dual-kernel ISOs with two identical kernels - one for 64-bit machines and another for 32-bit ones - it's quite cool.
There is one issue that I hope my statement clarifies. PAE doesn't have any more of an instability issue than "normal" or the "4GB only" 32bit kernels present. It, like the others 32bit mentioned, is stable.
The PROBLEM with PAE is that there are a very few CPUs manufactured by Intel and others which (for their cost considerations) did NOT include the PAE hardware. And, some BIOSes were also shipped with a settings that could also have impacted PAE use. But, for the most part, this is a ratity given the 15 years PAE has been built-into CPUs.
Thus, if you can, run 64bit! If you cannot, use 32bit PAE as it affords use of ALL of the RAM you can throw at it (assuming no CPU issues - which you will find out almost immediately), Lastly 32bit non-PAE runs just as fast (reviewing performance measurements already done by both vendors (intel and AMD) as well as independent test reports. PAE merely shines in Puppy which run RAM based because the filesystem is ALL of your RAM.
Hope this helps