Motorola's MPC7457 Soon?

[Catfish_Man]

Re: Re: Re: Re: Re: Re: Testing was biased!

Originally posted by eric_n_dfw

I'm no SIMD expert, but I think your're fooling yourself if you believe that "the whole OS is vector optimized."

Only certain things can be optimized for SIMD processing and I'd be surprised if much of the Darwin core has any AltiVec calls.

Quartz probably makes use of it for some of the eye candy - I'd venture a guess that a lot of the alpha-chanel stuff and the genie effect and things like that use AltiVec a lot. (I know dropping a G4/400 into my B&W G3/400 sped up Aqua a bit for me - and was like putting NOS in FCP!!)

Unfortunatelly, even with AltiVec doing the heavy lifting in applications like QuickTime, FCP, iMovie, iTunes and iDVD I still beleive it is ham-strung by the slow memory access speed. <soapbox>We need REAL DDR, like yesterday Steve! </soapbox>

All true. Assuming 1 vector floating point multiply-add per cycle (for example), that's (if I understand it correctly) 96 bits of data per cycle (3 source, 1 destination). On a 1GHz G4+ that's 96,000,000,000 bits per second, or 12GB/sec (There's gotta be something wrong with my math, that seems too high). A 167MHz bus can theoretically get 1.3GB/sec.

<edit>

Well, I was wrong. It's worse than I thought. On a 400MHz G4 (not G4+) using Altivec to add two streams of numbers together can eat up 12.8GB/sec. Using multiply-adds and writing the results back to memory can quadruple that (it turns out that multiply adds don't work like a = b * c + d, they work like abcd = efgh * ijkl + mnop, all in one cycle).

</edit>

<edit2>
This means that certain operations could go over 100 times faster if we had a fast enough bus (100GB+/sec).
</edit2>

 

[shadowfax]

Re: Re: Re: Re: P3 Vs. p4

Originally posted by Telomar


I'm really not sure how you got onto that tangent I said nothing about time to market of any technology. I was only pointing out the flaw in his calculations. People seem to fail to understand judging system performance is considerably more tricky than a number.

you were talking about how the "whole" computer had to be addressed, as opposed to simply the processor. i was agreeing with that, and considering it. i realize that you were pointing out a flaw in the argument, and i am pointing out the potential problem in yours. components, such as video and sound cards, make up a significant base of the "whole" computer. you can have a pretty crappy processor and still run some awesome games and intensive video apps if you have a nice enough video card. it speeds your system. and the PC world gets the fastest ones first; case in point, the 9700 radeon. i fail to see how you fail to see the relevancy of my comment.

 

[Telomar]

Re: Re: Re: Re: Re: P3 Vs. p4

Originally posted by Shadowfax


i fail to see how you fail to see the relevancy of my comment.

I'd say the reason I find it confusing is I never made comparisons between Apple and PC products.

 

[MacKid]

Re: Re: Re: Re: Re: Re: P3 Vs. p4

Originally posted by Telomar
I'd say the reason I find it confusing is I never made comparisons between Apple and PC products.

How are you supposed to get all your friends to switch then???

 

[shadowfax]

Re: Re: Re: Re: Re: Re: P3 Vs. p4

Originally posted by Telomar
I'd say the reason I find it confusing is I never made comparisons between Apple and PC products.

let's hav a look at what you were responding to:

Originally posted by Maxkraft
The P4 is at 2.8 ghz only encodes video as 4 time as fast as a p3 at 500 mhz. The p4 is 40% slower per Mhz. This is not a myth, but it does mean apple needs to ship a 2.2 ghz g4.

hmm. ok. i can see that technically you aren't making an apple-PC comparison. however, your argument as a response to this post seems to deal with such a comparison.

you said:

Originally posted by Telomar


Actually that suggests the processor is significantly worse than 40% slower given how much other aspects of the computer have improved since then.

The CPU doesn't make up the computer on its own. People do realise this right?

your argument is touching the speed of intel's processors, compared to themselves, of course, but it seems to me that you are saying this to refute his statement that we need a G4 2.2 GHz.

i inferred that you meant that apple has a competitive edge on the "rest of the computer," and responded that i don't think it does. but perhaps i was way off. sorry.