Shop OBEX P1 Docs P2 Docs Learn Events
Good news about Floating Point support for ICC — Parallax Forums

Good news about Floating Point support for ICC

ImageCraftImageCraft Posts: 348
edited 2008-08-28 22:44 in Propeller 1
I just did some quick check and it appears that we can use the code in the Float32Lib object for our library. We have already implemented a "CLIB" interface for launching library functions in another Cog (with access control for use by multiple Cogs), so it should be relatively easy to modify them to conform to our calling convention. And while not strictly necessary based on the MIT license, we also got official blessing from the author for its use in this manner. This will significantly lessen the amount of the work we have to do.

I think an end of October, but may be even mid-Oct release is possible.

***

And of course C supports native floating point, so at the source level, you write things like

float a, b, c;
int i, j, k;

a = i * sin(b / c) + atan(k);

etc. without explicitly calling the low level FP operations.

// richard

Comments

  • Mike GreenMike Green Posts: 23,101
    edited 2008-08-28 05:08
    My goodness, you-all have been busy. Very nice.
  • Mike HuseltonMike Huselton Posts: 746
    edited 2008-08-28 07:15
    I think Richard has been inspired by the Expo, among other developments.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    JMH - Electronics: Engineer - Programming: Professional
  • RaymanRayman Posts: 14,357
    edited 2008-08-28 13:51
    I don't know... I was kindof expecting a 64 or 80 bit intermediate result for floating point routines... I haven't looked, but I guess that Float32Lib is all 32-bit right?
  • jazzedjazzed Posts: 11,803
    edited 2008-08-28 14:03
    Wouldn't a 64 bit float be a double?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    --Steve
  • simonlsimonl Posts: 866
    edited 2008-08-28 20:13
    @richard: ImageCraft's starting to sound a little bit like Parallax - you're clearly listening to your customers. I salute you!

    OK; you get it released before last week in October and I'll buy ICC7 with my burfday money! Of course; I'll have to save up for a good introductory C book - any suggestions?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Cheers,

    Simon
    www.norfolkhelicopterclub.co.uk
    You'll always have as many take-offs as landings, the trick is to be sure you can take-off again ;-)
    BTW: I type as I'm thinking, so please don't take any offense at my writing style smile.gif
  • ImageCraftImageCraft Posts: 348
    edited 2008-08-28 22:44
    re: 64 bits double
    95% of the users don't need floating point. 95% of those only needs 32 bits. 95% of the remaining will complain that 64 bits FP is too slow. Hmm... 5%x5%x5% is still too high. Lets change all those to 1% smile.gif

    Anyway, so far after many years of nagging, we released 64 bits FP support for the AVR compiler, but it remains mostly as a check mark item for all but a small handful of users: everybody thinks they want it, but once they find out how slow it is, they use other solutions. It matters not that we told them all along that's the case...

    @simon
    OK, it's a promise. 30,000 or however many it is Propeller forum'ers see your message smile.gif

    I have not looked at a C book in years, but I always like the classic K&R, V1 and V2.
Sign In or Register to comment.