need explanation of powermacs..

Processor combinations

I do believe that I've seen this in at least one thread here, but I don't recall when or where it was. My understanding is that my original dual processor G5, for instance, has a combined maximum processor output of 4 GHz (2 GHz per processor), but really operates far closer to 2.4 GHz or so, with each processor saving some energy and running at, say, 1.2 GHz. If a processor-intensive task comes up—a 3D game, advanced mathematical computation (although I advise everyone who can to do Folding, as it doesn't use very many resources at all), DVD encoding, and the like—the processors will ramp up somewhat and, if enough power is needed, one may hear some of the internal fans start to spin much more quickly, loudly, and annoyingly.

In short, if every processor were operating at full capacity all of the time, the computer would be both noisier and hotter. As the actual frequency at which the processors typically operate tends to vary a bit, it's hard to pinpoint another acceptable measure of the computer's operating speed.

I consider it somewhat analogous to a car's engine. One may hear, for instance, '700 horsepower,' but that's simply a maximum output that the engine can produce when operating at peak efficiency in the perfect ambient conditions (temperature, altitude) and on the purest gasoline.

 

i dont believe that all applications are duel processor aware, thus not every program benefits from having the two processors

i could be wrong though

 

dual 2.5 GHz doesn't make 5 GHz. to claim that dual 2.5 GHz have the same computing power as a single 5 GHz would be a lie.

if a computation can be divided into two, then dual will finish the job in the same amount of time as the single with twice the power. but not all computations can be divided into two. many calculations depend on the results of previous calculations and in such cases, the computation cannot be split into two.

and many apps are not dual processor aware, further eroding the processing power of dual machines.

 

The only thing I might add to this is that even if all computations could be split amongst the two processors.. there would still be the added computations of splitting the original computations and then, after the fact, combining their solutions... it just doesn't add up to 5GHz with the overhead and exceptions.

I hope that's a clear explanation.

 

that is a good question with good answers. my old boss uses maya 5 on his DP 2ghz. and he would open up a small window that would show how much of each processor was being used during a render. its pretty cool cause he would still do other tasks such as email and web browsing at the same time.

i would like to add a question to this, as why did apple choose to show the clock speed of the G5 when AMD Athlon chips don't? whenever i see an Athlon benchmark comparison it just has the name of the chip maybe with the L2 cache and FSB but it says something like AMD Athlon 3200+ vs. Pentium 4 3.0ghz. i have no idea what the Athlon clocks out as, though it probably doesn't matter. am i just looking at the wrong place?

 

The AMD chip clocks somewhere around 2.4GHz I believe.

Its very hard to fight the Intel marketing. Common users will hear that and think that the 3.06GHz p4 will be much faster.. so AMD combats that by naming the chip based on comparitive benchmark results.

Really, I hate it. Its so confussing.

 

The good news from a usability stand point is that the OS is pretty good at splitting up the load among the two processors. So, you can have something use one whole processor and still be able to check email, etc. with little to no slow down (as long as there isn't anything too disk intensive going on).

 

cr2sh has the right idea. not all problems are fit for parallelization; for most problems, the compute power of added processors is not linear w/ the number of added processors.

note that some problems, such as how stanford is distributing its folding, do work well w/ massive parallelization. but for most general computing, you quickly hit a point of diminishing returns.

check out the pdf found here:
http://docs.info.apple.com/article.html?artnum=86429

it's got some performance comparisons that include the single 1.6 GHz g5 and the dual 1.8. the dual outperforms it by only 10% or so across the tests. i.e. it comes nowhere near close to doubling performance.