0
\$\begingroup\$

I´m working with floating point numbers in System Generator. I need to perform this arithmetic operation y = x*(-1) . I think it could be done by using the mult block, but I don´t like this way because is very expensive in terms of FPGA demanded resources. So I was wondering if there is any block to perform some kind of data conversión. The only thing is needed to do is to change one bit. Does anyone know how to change the sign a floating point number using a reasonable way? Thank you so much.

\$\endgroup\$
3
  • \$\begingroup\$ Interesting that this question came within hours of electronics.stackexchange.com/questions/9941/…. Homework, perhaps? \$\endgroup\$ – Brian Carlton Feb 10 '11 at 15:49
  • \$\begingroup\$ More or less, is part the reflexions I have doing a undergraduate thesis project. \$\endgroup\$ – Peterstone Feb 11 '11 at 6:17
  • \$\begingroup\$ Really? I was starting to get the impression we were doing your undergraduate thesis project ;) \$\endgroup\$ – drxzcl Feb 14 '11 at 0:23
1
\$\begingroup\$

Negating a floating point number involves inverting the sign bit.

(For fixed-point numbers there's a Negate block in the sysgen library which should avoid using a multiplier)

Be aware though, that sysgen is not designed for use with floating-point - you might find yourself jumping through all sorts of hoops!

\$\endgroup\$
5
  • 2
    \$\begingroup\$ No, negating a floating point number involves swapping the sign bit. The mantissa is always unsigned. As Peterstone said, the only thing that needs to be done is change one bit. \$\endgroup\$ – Kevin Vermeer Feb 10 '11 at 14:26
  • \$\begingroup\$ Doh! You're quite right - not sure what I was thinking then! All that needs inverting is the sign bit. But not the whole bit pattern! \$\endgroup\$ – Martin Thompson Feb 10 '11 at 21:30
  • \$\begingroup\$ Yeah, "not the whole bit pattern" you are right...Sorry I´ve done that and it works more or less, even the reason it´s not clear...I take it as a provisional solution.If anybody could tell me how can acess to just one bit, the sign bit... \$\endgroup\$ – Peterstone Feb 11 '11 at 6:21
  • \$\begingroup\$ I corrected your answer as we discussed in the comments - You can use the "edit" link if you can add some more information or make a correction to your answer. \$\endgroup\$ – Kevin Vermeer Feb 13 '11 at 21:47
  • \$\begingroup\$ @reemrevnivek: Many thanks - I've added a small tweak \$\endgroup\$ – Martin Thompson Feb 15 '11 at 12:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.