| 1 |
There flight tests proved the current system inefficient for all of the |
| 2 |
planned computation and communication with the groundserver. They |
| 3 |
claimed to be using a 416 Mhz xscale, which seems fairly beefy to me, |
| 4 |
and I was under the impression that ramming floating point instructions |
| 5 |
through a fixed-point register was grossly inefficient. |
| 6 |
|
| 7 |
However, I am very new to embedded systems, and your question |
| 8 |
makes me think that the control algorithm itself is the culprit. |
| 9 |
|
| 10 |
On 7/14/07, M. Edward (Ed) Borasky < znmeb@×××××××.net> wrote: |
| 11 |
> |
| 12 |
> Andre wrote: |
| 13 |
> > Check out the open source paparazzi project. |
| 14 |
> > http://www.recherche.enac.fr/paparazzi/wiki/index.php/Main_Page |
| 15 |
> > |
| 16 |
> > and gumstix boards. |
| 17 |
> > |
| 18 |
> > _____ |
| 19 |
> > |
| 20 |
> > From: Jack Poulson [mailto:poulson.jack@×××××.com] |
| 21 |
> > Sent: Thursday, July 12, 2007 11:17 AM |
| 22 |
> > To: gentoo-embedded@l.g.o |
| 23 |
> > Subject: [gentoo-embedded] ARMs w/ Floating Point |
| 24 |
> > |
| 25 |
> > |
| 26 |
> > I apologize for the off-topic question, but I would like to become |
| 27 |
> involved |
| 28 |
> > in the group and need to order an ARM board which actually contains a |
| 29 |
> > floating point register first. It will serve as the flight control |
| 30 |
> system on |
| 31 |
> > a |
| 32 |
> > Unmanned Aerial Vehicle project for my college. Their previous attempt |
| 33 |
> > proved too slow, and I would like to set up a new system running a |
| 34 |
> stripped |
| 35 |
> > down version of Gentoo. |
| 36 |
> > |
| 37 |
> > Does anyone have any suggestions for a chipset? Any responses are |
| 38 |
> greatly |
| 39 |
> > appreciated. |
| 40 |
> > |
| 41 |
> > Jack Poulson |
| 42 |
> > |
| 43 |
> > |
| 44 |
> > |
| 45 |
> > |
| 46 |
> |
| 47 |
> Just out of curiosity, why do you think floating point arithmetic is |
| 48 |
> necessary for a flight control application? People have been doing |
| 49 |
> fixed-point flight control and other real-time applications for at least |
| 50 |
> fifty years, and the hardware requirements for fixed-point arithmetic |
| 51 |
> are *always* going to be less than for floating point arithmetic. |
| 52 |
> |
| 53 |
> Now floating point hardware may be inexpensive enough now that it's |
| 54 |
> *feasible* to use it for this application, but I really don't think it's |
| 55 |
> *necessary*. |
| 56 |
> -- |
| 57 |
> gentoo-embedded@g.o mailing list |
| 58 |
> |
| 59 |
> |
Archives