Sign In
Register
Help
Hello all:
I use FCP Studio on a Quad G5 with 2GB RAM and the 7800GT graphics card.
My question concerns the performance of Universal Binary Apps on a PowerPC machine. After Apple introduced their cross-grade program for FCPS, I was thinking if I should future proof myself by taking up on the offer so that if/when I buy a new replacement machine in the future, I don't need to buy a new copy of FCPS at full price. However, this means that in the mean time, if I needed to reinstall FCP on my Quad, I'll be installing a UB app and I'm wondering if I would take a performance hit. Furthermore, would a UB app be not as optimized to run on Intel machines compared to strictly Intel only apps? If this is the case, I'd rather stick with what I have and buy a fully Intel optimized copy in the future.
Thanks.
Sam
Page 1 of 1
PowerPC vs Universal Binary Apps
MultiQuote
#2
d00d 
- Advanced Member
-
- Group: Mac User
- Posts: 12,149
- Joined: 24-April 01
Posted 10 February 2006 - 11:34 AM
However, this means that in the mean time, if I needed to reinstall FCP on my Quad, I'll be installing a UB app and I'm wondering if I would take a performance hit.
No.
Furthermore, would a UB app be not as optimized to run on Intel machines compared to strictly Intel only apps? If this is the case, I'd rather stick with what I have and buy a fully Intel optimized copy in the future.
Again, no.
Just for completeness, let me elaborate. As part of this move to Universal Binaries is the adoption of the Accelerate Framework. This API is used to abstract the vector manipulation functions built into both types of processors. This negates the need for processor specific code (for the most part, I'm sure there are rare instances of certain applications that will not be able to use this).
Now, because this is an abstraction, there is a slight possibility of lost cycles that could theoretically be gained by using processor level code. However, that would require a developer to write tighter code than is in the framework itself. That's highly unlikely. Even if it were to happen, it would be a matter of a few lost cycles, which isn't meaningful in actual operation of the application. That amount of optimization isn't going to impact you.
Once this framework is adopted by a developer, its use will be considered an investment. Once the decision to go Intel only is reached (which I don't think we'll see until the Intels simply outpace the PowerPCs by orders of magnitude and the software requirements overwhelm the processing power of all PowerPCs, which is highly unlikely for a number of years), the justification of rewriting code utilizing the Accelerate framework would need to be a significant performance increase. In order to achieve that, a significant amount of code would beed to change (to pick up enough spare cycles to add up to something meaningful) in which case, the development cost is even greater, which is unattractive, particularly when the savings will be meager at best.
I don't think your concern is really anything of substance.
No.
Furthermore, would a UB app be not as optimized to run on Intel machines compared to strictly Intel only apps? If this is the case, I'd rather stick with what I have and buy a fully Intel optimized copy in the future.
Again, no.
Just for completeness, let me elaborate. As part of this move to Universal Binaries is the adoption of the Accelerate Framework. This API is used to abstract the vector manipulation functions built into both types of processors. This negates the need for processor specific code (for the most part, I'm sure there are rare instances of certain applications that will not be able to use this).
Now, because this is an abstraction, there is a slight possibility of lost cycles that could theoretically be gained by using processor level code. However, that would require a developer to write tighter code than is in the framework itself. That's highly unlikely. Even if it were to happen, it would be a matter of a few lost cycles, which isn't meaningful in actual operation of the application. That amount of optimization isn't going to impact you.
Once this framework is adopted by a developer, its use will be considered an investment. Once the decision to go Intel only is reached (which I don't think we'll see until the Intels simply outpace the PowerPCs by orders of magnitude and the software requirements overwhelm the processing power of all PowerPCs, which is highly unlikely for a number of years), the justification of rewriting code utilizing the Accelerate framework would need to be a significant performance increase. In order to achieve that, a significant amount of code would beed to change (to pick up enough spare cycles to add up to something meaningful) in which case, the development cost is even greater, which is unattractive, particularly when the savings will be meager at best.
I don't think your concern is really anything of substance.
#4
Bobicus
- Newbie
-
- Group: Members
- Posts: 14
- Joined: 21-August 05
Posted 13 February 2006 - 10:19 AM
Or in basic terms, Universal Applications have code for both the PowerPC and Intel chips, so it runs with no performance hit or other annoying hiccups on either platform. Put your mind at ease and get your cross-grade without worry.
Page 1 of 1
