Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
19 years ago

Access to the Floating Point Costum Instruction

I Have simply add a floating point custom Instruction (named fpoint) to my nios II processor core under the SOPC Builder with Quartus II 6.1.

The generation was successful and now a want access the part of hardware under NIOS II IDE.

I will use the following function call to use the hardware:

---------------------

float a = 3.14

float b = 5.66

float result

result = fpoint(2,a,b); // 2 is the prefix

----------------------

How can i reach this?

Have i made changes in the "system.h"-file?

Can you explain step by step the work under the NIOS II IDE to get access to the CI?

Thanx.

PS: I habe read all altera pdf about the issue...

2 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    19 years ago

    --- Quote Start ---

    originally posted by tux.2000@Feb 9 2007, 03:49 AM

    ps: i habe read all altera pdf about the issue...

    <div align='right'><{post_snapback}> (index.php?act=findpost&pid=21400)

    --- quote end ---

    --- Quote End ---

    Yes, but did you download the source code that goes along with the floating point tutorial? I just looked through it, and it&#39;s quite clear what you need to do.

    After looking through the sample code, you don&#39;t have to do anything to enable floating point. All floating point operations will automatically use the custom instruction. If you want to use SW floating point, you have to use# pragma&#39;s to disable the custom instruction.

    - slacker
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    19 years ago

    Okay, i understand. I didn´t know, that the custom instruction is already in use. I Think, i have to set up a define-statement or my own macro., but this was a mistake...

    Thanx.

    Now, i get the following result.

    Floatingpoint addition:

    with Custom instruction --> 5 times faster

    floatingpoint multiplikation:

    with custom instruction --> 17 times faster

    I think these are realistic values...