Is QuarkXpress utilized by Altivec?

in Mac Software edited January 2014
If I go down to a G3 iBook, I'm a bit concerned with some programs I use on a regular basis running slower.

Now I don't plan to layout designs in Quark on the iBook but to occasionally open files.

So my question is what programs really make use of Altivec? Specifically these:

1) Quark Xpress

2) Illustrator

3) Filemaker

4) AppleWorks



  • Reply 1 of 6
    eugeneeugene Posts: 8,254member
    Nope, I don't think QuarkExpress helps AltiVec at all.
  • Reply 2 of 6
    satchmosatchmo Posts: 2,699member
    [quote]Originally posted by Eugene:

    <strong>Nope, I don't think QuarkExpress helps AltiVec at all. </strong><hr></blockquote>

    Har de har har. My error...but i think you know what i mean. :-)
  • Reply 3 of 6
    buonrottobuonrotto Posts: 6,368member
    I seriously doubt it, but I'm just a nay-sayer and have no knowledge one way or the other.
  • Reply 4 of 6
    jccbinjccbin Posts: 476member
    XPress does not use altivec.

    Most of the code is a decade old, even QXP 5. Xpress does take advantage of Classic's use of dual processors (Since Xpress is a sub-app of Classic and Classic can be more responsive under dual systems).

    If rumor has it correctly, Xpress 6 (due in early 2003), will be a complete re-write of Xpress and MAY take advantage of the Velocity Engine (VE) (Altivec/IBM's SIMD). If the re-write is Cocoa-based (I don't know) then Xpress may take more advantage of the VE due to tweaking of the Cocoa frameworks by Apple, over time.

    IF it's a carbon port, it will likely be the last or second-to-last version of XPress ever. Of course, InDesign will bury Xpress long before then. :-)
  • Reply 5 of 6
    dfilerdfiler Posts: 3,420member
    I can't think of any types of algorithms used by Quark which could be rewritten to utilize altivec. These units are designed for parallel, non-dependant calculations such as image and video filters. The simplest example of this would be adjusting the hue of a bitmap. Altivec can simultaneously compute new RGB values for multiple pixels. The bulk of quark's calculations are comprised of 2d page layout computation, something that is typically done in a sequential manner. While, it is technically possible to analyze the tree of objects on a page and perform parallel layout computations, I doubt it would offer a perceptable speed improvement.

    Now I'm curious, has this been attempted in comercial products? I assume that academia has been working on parallized layout algorithms involving predictive branching... (?)
Sign In or Register to comment.