[Ti] A real world comparison

[~flipper]

Chris Olson wrote:

>On Jun 13, 2005, at 6:16 AM, ~flipper wrote:
>
>>Is there any way to sort out which apps are more vector intensive?
>>
>>I'm doing a little research (on spec) for a couple guys regarding 
>>stock market stuff.
>>
>>I'm curious about Final Cut Pro (and the HD version, and bundled 
>>apps), as well as the bigtime audio engineering stuff.
>
>All very dependent on AltiVec on PowerPC.  Of course, FCP is Apple's 
>problem.  They'll either make it perform or it will quickly be 
>labeled a "dog" and the userbase will mass migrate to something else.
>
>Apple's version of GCC is optimized for vector code on PowerPC.  At 
>WWDC they like to paint a rosy picture, and the picture isn't so 
>rosy until we either get a different compiler or use Intel's 
>compiler in place of GCC.  Otherwise porting PowerPC vector code to 
>Intel SSE is easy enough to do, and the performance seems comparable 
>on a couple instances I ran last night.  But Apple better eat some 
>byte soup and crap a different compiler like real quick if there's 
>going to be optimized universal binaries for apps that make heavy 
>use of vector code.
>--
>Chris

Thanks for the reply, Chris. I sort of figured that, but I'm not in a 
'sort of' situation. I'm curious id=f there's a searchable item, 
withing 'packages' to establish whether or not an application is 
using a lot of vector calls. (Is that the correct term? probably not, 
but you probably know what i think i mean).

I have access to just about everything, and some identifiable 
'target' (possibly 'unique', to cut down on multiple 'hits' within 
the same app, or false 'positives'?) would be very helpful, too.

I'm wondering if loss of the 'creative' community would be chalked-up 
to 'attrition' in the new mactel Way. (Not that i expect guesses from 
everybody, or some outright 'statement' from Apple), i can research 
the impact and extrapolate on my own, for my narrow purposes here.

brian s