10 May 2011 22:08
Re: Proper CLFS setup for an ARM Cortex A8
Andrew Bradford <bradfa <at> gmail.com>
2011-05-10 20:08:52 GMT
2011-05-10 20:08:52 GMT
On Tue, May 10, 2011 at 12:03 PM, Hector Oron <hector.oron <at> gmail.com> wrote: > 2011/5/10 Angel Ivan Castell Rovira <al004140 <at> gmail.com>: >> - CLFS_FPU: There are a lot of features (vfp, vfpv3, neon). What do you >> think it is the best choice? > Depends on your hardware, your platform seems to have neon, that's the > one. Building vfpv3 would make your rootfs more compatible, i.e. able > to run on tegra devices which does not have neon FPU. Also note that VFPv3 is IEEE 754 compliant floating point. NEON is _NOT_. If that's important to you, definitely don't use NEON. Granted, if you want to do neat SIMD (single instruction, multiple data, see: http://en.wikipedia.org/wiki/SIMD) NEON is the way to go. Realistically, a multilib situation that can switch between NEON and VFPv3 is the best of both worlds but I'm unsure of how to implement that. >> As you can see, there is no Cortex A8. What is the proper Target Processor >> Type I should select here? Generic Arm is a good choice if you're using arm mode. -Andrew
RSS Feed