Jump to content


altera floating point library


  • You cannot reply to this topic
1 reply to this topic

#1 Lyle

    Member

  • Members
  • PipPip
  • 3 posts

Posted 03 March 2010 - 12:56 PM

Hi,

I am trying to simulate my design which has floating point operations in Modelsim. All generated floating point related files like fp_div_VHD are encrypted. Also there are no altera library for these operations. I am using Modelsim-ASE 6.4a. Anyone know how to deal with it? Thanks!

Wenqian

#2 Gus

    Newbie

  • Members
  • Pip
  • 2 posts

Posted 16 June 2010 - 01:33 PM

QUOTE (Lyle @ Mar 3 2010, 12:56 PM) <{POST_SNAPBACK}>
Hi,

I am trying to simulate my design which has floating point operations in Modelsim. All generated floating point related files like fp_div_VHD are encrypted. Also there are no altera library for these operations. I am using Modelsim-ASE 6.4a. Anyone know how to deal with it? Thanks!

Wenqian


Hello Weniqian,

We have had the same problem. We leave the Impulse libraries (11 cycles latency for mult and add) and the Altera cores (7 cycles for addition, 5 for multiplication) because the enormous latency that makes deep pipelines unfeasible in Altera devices.
We rewrote the FP operation with less latency (1 cycle for mult, 2 for add/sub, 6 for div, 5 for sqroot).
If you steel want to test the VHDL code you can use the altera core and modify the PSP acordingly.

Regards, G






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users