Re: [eigen] Eigen code gets optimized away by avr-g++ - what to do?

[Christoph Hertzberg <chtz@xxxxxxxxxxxxxxxxxxxxxxxx>]

On 30.06.2012 00:00, Christoph Hertzberg wrote:
> On 29.06.2012 22:21, Ryan Pavlik wrote:
> > However, I am now seeing quite strange behavior. It looks like Eigen code
> > (and possibly the rest of the function in which it is used?) gets
> > optimized
> > out of the compiled binary.  (It's so efficient, it doesn't have any
> > code!
> > :D )  When built by the IDE, it uses -Os optimization (which is -O2
> > without
> > the ones deemed to potentially enlarge the code) which triggers the
> > issue.
> > -O1 optimization doesn't seem to trigger it (by looking at the
> > disassembly,
> > I actually see a call to sqrtf when my test program calls .norm() on a
> > vector) though I haven't actually run that code on a device.
>
> The code looks simple enough that gcc can calculate the result of
> v1.norm() at compile-time. You should pass an unknown vector to the
> function, if you want see whether it actually does anything.


Looking a bit closer into the generated assembly, for O1 the sqrt 
implementation recursively calls itself in line 1970:
     f10:	0e 94 88 07 	call	0xf10	; 0xf10 <_ZSt4sqrtf>

In Os this seems to get optimized to an infinite loop (line 173):
 116:	ff cf       	rjmp	.-2      	; 0x116 <loop+0xc>

So I guess your sqrt implementation does not work ...

Christoph


--
----------------------------------------------
Dipl.-Inf. Christoph Hertzberg
Cartesium 0.049
Universität Bremen
Enrique-Schmidt-Straße 5
28359 Bremen

Tel: +49 (421) 218-64252
----------------------------------------------