[<<][davinci][>>][..]
Sat Jul 25 12:06:50 CEST 2009

ARM926EJ-S and floating point

Looking at this[1] I wonder what this is about:

  Linux does not currently support floating point hardware so you need
  to say Y here even if your machine has an FPA or floating point
  co-processor podule.

I needed to enable CONFIG_FPE_NWFPE to make the debian binaries work.

< CONFIG_FPE_NWFPE=y
> # CONFIG_FPE_NWFPE is not set

I wonder why this wasn't necessary for the neuros userland..  Maybe
because it's uclibc?

So, does the davinci Arm have float support?  It doesn't look like it
(no `fpu' in /proc/cpuinfo).  From here[4] no mention of FPU either,
only fixed point DSP extensions (single cycle mac and 16x32 bit
multiplier).  It uses the ARMv5TEJ instruction set[5].  Makes sense,
since the DSP is also fixed point.

Are there any Arm chips with float support?

[1] http://cateee.net/lkddb/web-lkddb/FPE_NWFPE.html
[2] http://www.arm.linux.org.uk/mailinglists/faq.php#f3
[3] http://focus.ti.com/docs/prod/folders/print/tms320dm365.html
[4] http://www.arm.com/products/CPUs/ARM926EJ-S.html
[5] http://www.arm.com/products/CPUs/architecture.html




[Reply][About]
[<<][davinci][>>][..]