The 4 Core CPU, Overkill?

[[R-DEV]eggman]

@mdterp .. hehe yeah I want a freakin LCD like wallpapered dome I climb into

[Ghostrider]

Originally Posted by Downtown_two

Actually, right now top end PCs are plauged with bottlenecks, usually the CPU.

While the bottlenecks in computers is true, the part that says that the CPU is the bottleneck is wrong.

Bottlenecks in computers are the bus, the hdd, and RAM memory...mainly.

The HDD because (leaving the human out) that's the slowest component in the PC. Then follows the RAM, which has to be continually refreshed in order to not loose the bit-state (voltage indicating a 1 or a 0), among other factors. also, the system bus, which has an significantly lower operating freq. than the CPU (which I'm sure you guys must already know).

You can see this in the CPU multiplier, which indicates how many times faster than the system the CPU is working...the bottleneck of a system has never been the CPU...because the entire system is always slower than the CPU...now you end up with a system, that causes a bottleneck to the CPU...all by itself.

Maybe someone might argue that "If the CPU is not the bottleneck, why do I overclock the CPU?", to which the answer is quite simple. If you increase the system bus, you're overclocking the system, not the CPU...and if you increase the CPU multiplier, you're increasing the freq. of the CPU...but the system bottleneck stays the same....maybe some memory-related operations or CPU intensive tasks (w/ cache memory hits) will operate significantly faster...but that's at the CPU level...the system gets no benefit from this..


I think I wrote too much.. o.o

-Ghost

PS: I love this, you make a simple statement, and get a complicated answer.

[Ghostrider]

Originally Posted by [PIT]dizzy

If there isn't even that much 64 bit and dual core code out there right now so why would anyone expect there to be quad core code earlier than 4 - 5 years away. Why buy a more expensive CPU now that will give you nothing for at least 4 years???

Couple of reasons that I can mention are:

• hardware and software are very dependent on each other, so you need "someone" to come out first;
• making software to be compatible with hardware is easier (as complicated as it may be) than doing it the other way around;
• if the hardware doesn't come out, noone will write code for it..so you need to , sort of, "deliver"..
• since software is made to be compatible with hardware...hardware must come out first..

Those are only some small factors to consider. Someone always pushes the other side to make some sort of progress..

Those that benefit the most from this are gamers. Game developers are some of the most technical software devs in the entire industry, and the code that they develop tends to push hardware to their limits. If not, why would you need a new gfx card every now and then...or CPU to handle better A.I. routines and the such?

And that's leaving simulations (like Armed Assault or Falcon 4.0: Allied Force) out of the ecuation.

-Ghost

[Hellfire]

Originally Posted by Downtown_two

Actually, right now top end PCs are plauged with bottlenecks, usually the CPU. If any apps were to use all 4 cores it would certainly help. I really doubt 4 cores will become useful for a while though.

Point is, in alot of games, my X1900XTX is a waste of time because my AMD 4000+ cant keep up. All depends on the game though.

Are you serious?
I have an AMD 3800+, 1 gig of Corsair DDR2 800, and a Geforce 7900GT and the only bottleneck i've EVER experienced is BF2's memory whoring... =\
(Not to mention all the facts Ghost pointed out that I was too lazy to type out. :P)

[Downtown_two]

Originally Posted by [R-DEV]Ghostrider

While the bottlenecks in computers is true, the part that says that the CPU is the bottleneck is wrong.

Bottlenecks in computers are the bus, the hdd, and RAM memory...mainly.

The HDD because (leaving the human out) that's the slowest component in the PC. Then follows the RAM, which has to be continually refreshed in order to not loose the bit-state (voltage indicating a 1 or a 0), among other factors. also, the system bus, which has an significantly lower operating freq. than the CPU (which I'm sure you guys must already know).

You can see this in the CPU multiplier, which indicates how many times faster than the system the CPU is working...the bottleneck of a system has never been the CPU...because the entire system is always slower than the CPU...now you end up with a system, that causes a bottleneck to the CPU...all by itself.

Maybe someone might argue that "If the CPU is not the bottleneck, why do I overclock the CPU?", to which the answer is quite simple. If you increase the system bus, you're overclocking the system, not the CPU...and if you increase the CPU multiplier, you're increasing the freq. of the CPU...but the system bottleneck stays the same....maybe some memory-related operations or CPU intensive tasks (w/ cache memory hits) will operate significantly faster...but that's at the CPU level...the system gets no benefit from this..


I think I wrote too much.. o.o

-Ghost

PS: I love this, you make a simple statement, and get a complicated answer.

Touché ghost, you obviously know what your talking about. Your 100% correct in the context of the memory hierarchy, you have your internal registers right up there as the fastest and your local storage device at the bottom.

In terms of gaming though, because of the advance in the GFX cards the bottleneck shifted to the CPU (obviously dependant on hardware). Collision detection, per-polygon collision detection, AI and every aspect of physics are calculated by the CPU and while CPUs are getting better games engines are advancing just as fast if not faster. If you have a game that uses alot of these thing (I know I missed alot) your CPU load will max out and BOOM, bottleneck.

A practical example is GRAW (with all is physics glory) I get the same FPS with everything low than with everything high with 6AA 16xAF.

I guess that was the thinking behind the AGEIA PhysX processor, taking the load off the CPU to allow CPU intensive games to utilize the other components.

Again though, I should say this is dependant on the game. Its not my best writting but I hope I got my point across, Im not arguing with you ghost because you ARE right but so am I!

I may have written too much aswell.

[dizzy]

Ghost you bottleneck post is exactly what i was going to write, i actually started writing it but then remember I was too lazy so I closed the window.

On you response to what I wrote is that I mean why would I spend the money if it will not help me now. Not that Im saying we should never move to quad core. Also, it makes me very sad to see that we are moving to dual and quad core b/c that means the CPU companies having giving up on pushing smaller and smaller transistors and pushing up the clock speed.

IMO a single core running super super fast will always be better than dual cores. But I guess when you hit a limit you hit a limit and now its software dev's turn to actually right good threaded code.

[[R-DEV]Masaq]

Trouble is Dizzy, the physical laws of the universe govern how fast a CPU core can operate... you reach a point where a single core simply cannot go any faster (short of suspending the whole ensemble in liquid nitrogen to deal with the heat. As for the other limitations, well, no way around 'em.)

Two/multiple cores *are* a more effective and efficient way of getting more processing power out of a CPU, so long as the software utilising them is written properly.

And as the original article posted said- Intel haven't abandoned pushing the limits yet; they're going to a 45nm process next year and 35 in 2009.

[dizzy]

Originally Posted by Masaq

Trouble is Dizzy, the physical laws of the universe govern how fast a CPU core can operate... you reach a point where a single core simply cannot go any faster (short of suspending the whole ensemble in liquid nitrogen to deal with the heat. As for the other limitations, well, no way around 'em.)

I dont believe it's a heat issue alone, its that they can not make the transistors any smaller (at least with the current technology) If you can't make them smaller you can push the speed up.

Originally Posted by Masaq

Two/multiple cores *are* a more effective and efficient way of getting more processing power out of a CPU, so long as the software utilising them is written properly.

I dont agree with this. There is nothing more efficent than a single "whatever". Example a raid stripe with two 7200 RPM drives will be slower than a single 15k RPM drive (by a good amount). Every time you add layers onto something you introduce overhead.

If CPUs have reached their physical limit then overhead must be introduced and eventually the overhead will be absored by better and better code and such.

Anyone that likes this nerdy crap should check out this article.

The PC Turns 25 Today We explain why we don't think it's going to live to see 50. - http://overclockers.com/tips01014/

[Downtown_two]

Interesting article but to me it sounds like alot of conjecture with not alot of reasearch. Reminds me of the 60s perception of what 1985 would look like with out hover cars and moon colonies.

[[R-DEV]Masaq]

I can't see much gold in the grit there either I gotta say. Microsoft dying? Not likely. They'll simply change what they do- a company that size doesn't just vanish over three decades.

Dizzy: Yup, it's not just the heat (but it's a very important issue. My P4EE currently produces enough warmth at idle to warm my small flat's living room adequately at night lol), there also comes a point when you can't make transistors any smaller. Eventually you get to the limits stemming from the way electrons behave, but that's a long way off. (It's interesting to look at the top clock speeds of processors over the past ten years- they've hardly budged for several years now).

As for redundancy, well, simply do the following. If you take a hyper-threaded processor, disable HT and run PCMark05 three times, take the average CPU result. Then re-enable HT and repeat, see which score's higher.