Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
stuartluff said:
No I think you are confused. :) I meant "Is having more cores, lets say 8, more efficient than one big core equal in processing power to the 8 cores?"

Well next time say what you mean. It makes more sense. ;)
 
satty said:
But as some already pointed out, many applications can't use multiple cores, therefore you won't get any performance improvements with multi cores.

Im not talking about performance, more about energy usage. I thought maybe they are using more cores as it is more energy efficient than using less cores or one big one. But as someone has pointed out its more likely a case of not having to squeeze more transistor thingies on a chip, they may as well just add another chip. :)
 
tny said:
You realize there are probably only four people on this board who are old enough to get that joke, right?

I even had a client that was using one until about a month ago. The hard drive died. But, that Quadra just kept going and going and going.
 
zero2dash said:
...Quad Duo?
...Quadra Duo?
...the "holy hell this is faster than you'll ever need" Mac? :D

Ha, if only. ;) The difference between software developers "taking advantage" of new functionality and "bloatware" is a matter of semantics at times.
 
stuartluff said:
Is having more cores more energy efficient than having one big fat ass 24Ghz processor? Maybe thats a factor in the increasing core count.
Actually, this is well documented.

There are serious electrical and physical problems with jacking up clock speeds much further than they are now. Intel managed to push their chips to 3.4GHz, but the power consumed was tremendous.

When you can't ramp up the clock speed, your next best alternative is to go for as much parallelism as you can - increase the number of instructions you can execute in a single clock.

Chip makers achieve this in a wide variety of ways, including multiple CPU packages on a motherboard, multiple cores per CPU package, multiple threads per core, and multiple functional units per thread.

And yes, a single CPU at 3GHz can easily consume more power than two CPUs (or two cores) at 1.5GHz.

As for your theoretical 24GHz processor, such a thing is simply not possible with today's technology. (Well, there were some university experiments that hit insanely fast speeds, but don't expect commercial products any time soon.) Given the heat/power curves of today's chips, I wouldn't want to think about the cooling requirements of a 24GHz chip if you could somehow manage to build one.

Of course, breakthroughs do happen, and higher clock speeds might become practical in the future. But multi-core tech isn't going away - we'll simply end up with multiple cores at higher clock speeds.
 
zero2dash said:
...Quad Duo?
...Quadra Duo?
...the "holy hell this is faster than you'll ever need" Mac? :D

If it can't model WW2 D-Day in full holographic 3D, with AI ground troops, AI generals, real physics and weather effects I garuntee you there will be people asking for more power.

(Now, doesn't that sound like a sweet game? :) )
 
satty said:
But as some already pointed out, many applications can't use multiple cores, therefore you won't get any performance improvements with multi cores.
A single application, if not multithreaded, won't see any performance boost.

But if you're running multiple applications at once, your overall system performance will definitely improve.

Also note that many of Apple's system facilities (like Core Image) are internally multithreaded. So apps that use these system services will see performance boosts even if the application developer didn't write any multithreading code into the app.

I am also certain that we'll see more and more developers using multithreading, now that all but the cheapest systems sold will have at least two cores. Especially with those apps that are CPU-intensive, and could therefore gain the most from multiprocessing.

(Gee, it seems like it was only a few short years ago that we were having this same discussion about AltiVec :) )
 
Chundles said:
Well next time say what you mean. It makes more sense. ;)

I did but instead of saying core at the end I said processor :D Which is the same thing so i didnt think it would matter. :p
 
shamino said:
There are serious electrical and physical problems with jacking up clock speeds much further than they are now. Intel managed to push their chips to 3.8GHz, but the power consumed was tremendous.
Fixed
 
shamino said:
As for your theoretical 24GHz processor, such a thing is simply not possible with today's technology.

Just stating 'I knew that' I just used it as an example. Chundles gets confused easily so I have to make things simple. Hi Chundles :D
 
stuartluff said:
Just stating 'I knew that' I just used it as an example. Chundles gets confused easily so I have to make things simple. Hi Chundles :D

You'll keep...:p
 
tny said:
You realize there are probably only four people on this board who are old enough to get that joke, right?

My quadra still runs, I guess I'm the forth party to get it.

This feels almost like an onion article:

Home Computer Gives Birth to Octuple-Cores

<enter photoshopped picture of a Mac Pro craddling its new born octuplets>
 
stuartluff said:
No I think you are confused. :) I meant "Is having more cores, lets say 8, more efficient than one big core equal in processing power to the 8 cores?"
First of all, you assume that it is possible to make "one big core equal in processing power to the 8 cores". I don't think it is possible to do this (at least not with the x86 architecture using today's technology.)

But assuming such a chip exists, the answer depends on what kind of efficiency you're thinking of.

If you mean computational efficiency (meaning the most useful processing per clock-tick), then a single big core will do better. This is because single-threaded apps will be able to use the full power (whereas multiple threads are needed to take advantagte of multiple cores.) Also, the operating system can get rid of the overhead that is needed to keep software running on the multiple cores from stepping on each other.

If you mean energy efficiency (amount of processing per watt of electricity consumed), then it could go either way, depending on how the chips are made. But given today's manufacturing processes and the non-linear power curve that we see as clock speeds are increased, the multiple-core solution will almost definitely use less power.
 
miketcool said:
My quadra still runs, I guess I'm the forth party to get it.

This feels almost like an onion article:

Home Computer Gives Birth to Octuple-Cores

<enter photoshopped picture of a Mac Pro craddling its new born octuplets>

My first job as a graphic designer I used an enhanced SE/30 (with 20" external monitor). About a year later we upgraded to the Quadras, so I guess that makes me #5?
 
Something I'd like to see is the framerate on Doom 3 if I played it on an 8-core Mac Pro w/ 2 of the fastest videocards in SLI/crossfire mode and one of those physics coprocessors (saying Doom 3 was optimized for it).
 
shamino said:
First of all, you assume that it is possible to make "one big core equal in processing power to the 8 cores". I don't think it is possible to do this (at least not with the x86 architecture using today's technology.)

But assuming such a chip exists, the answer depends on what kind of efficiency you're thinking of.

If you mean computational efficiency (meaning the most useful processing per clock-tick), then a single big core will do better. This is because single-threaded apps will be able to use the full power (whereas multiple threads are needed to take advantagte of multiple cores.) Also, the operating system can get rid of the overhead that is needed to keep software running on the multiple cores from stepping on each other.

If you mean energy efficiency (amount of processing per watt of electricity consumed), then it could go either way, depending on how the chips are made. But given today's manufacturing processes and the non-linear power curve that we see as clock speeds are increased, the multiple-core solution will almost definitely use less power.
I remember hearing about how it is possible to make multiple cores act like one (Idon't remember where I heard this). Anyways, whether 8 cores acting separately or together like 1 big processor has an advantage depends on the program you use. If the program is multi-threaded, then the cores acting separately might have the advantage while single threaded apps will have an advantage if the cores are acting like one. However, many apps today won't see that much improvement either way (like a simple calculator, or solitare and word processing).
 
kev0476 said:
you need to do your math better, extra core = 1.5x - 1.8x speed increase. but still the same power usage as a normal core!

Wow Im having to do a lot of explaining of my posts, im beginning to think im a bit retarded. :(

24Ghz is (in my head) an 8 core machine running at 3Ghz i.e. 8x3=24

Its just an example to make things easier for people to answer my OP. :)

DISCLAIMER: All information contained within my posts sounded right in my head at the time of writing. However I may have been A) Quickly typing because I should actually be working B) Trying to be funny but failing miserabley or C) Drunk. It is not my intention to confuse people but I seem to do it quite easily and without knowledge. Please address any complaints to the complaints dept. at Microsoft Corporation.
 
reallynotnick said:
Anyone else think this is getting out of hand? Two cores, great improvement. Four cores, ehh it's faster but Joe can't tell. Eight cores, now thats just stupid.

No way. It would be stupid for a web surfing machine. But for people who need the power, they're going to absolutely notice when it does things TWICE as fast. I say bring it on (and I'm running a quad and see a *huge* difference).
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.