Originally posted by: MrK6
Except that in the reality, a 6GHz single core isn't feasible, while a 3GHz dual core is and, by extension, a 6GHz dual core isn't feasible,
I didn?t say it was (currently), only that I would prefer one if I could get one.
but a 3GHz quad core is. The simple fact is that as we reach the physical limits of of processing capabilities, we have to take alternate routes to achieve greater performance - enter multithreading. Quad cores exist for the same reason as SLI and CF - they don't scale perfectly, but 40/60/80% extra performance is still much better than nothing, and is far beyond the capabilities of any single part.
None of this is relevant to anything I?ve stated.
And you state you'll get more consistent performance, but how much more? 5%, maybe 10% higher on average? You'd rather have that than a potential 60-80% jump in performance?
What are you talking about? The vast majority of CPU limited situations do
not benefit from multi-core; therefore in those instances a 6 GHz single core will be twice as a fast as a 3 GHz dual-core, and a 3 GHz dual-core will be no better than a single core 3 GHz (assuming everything is the same except clock speeds).
Even the amount of titles that take advantage of two cores is but a drop in the bucket compared to the total number of Win32 games out there.
And I've already stated - the parts cost the same, this isn't a question of price, but what you get for your money.
And I?ve repeatedly stated that what you get overall is inferior because the overwhelming amount of applications do no benefit from quad-core. You?re simply trading clock speed (and cache sometimes) in exchange for more cores and getting a less robust system in the process.
But if you want to run more than one of those single or dual threaded processes?
You want to play more than one 3D game at a time? Why?
Even when it?s possible (the 3D API/driver often locks the device so it often isn?t), most games dramatically reduce their CPU usage when minimized in the background.
Do you even have the slightest clue about SLI/CF AA modes? Because based on your comments it appears not.
You can't argue that scaling with CPU multithreading is horrible and wasteful while SLI/CF is fine - it's quite the opposite in my experience. Also, I don't get any extra AA modes when I go into SLI - 16xSSAA is still the max I have.
So you?ve never used SLI AA modes, and know nothing about them?
But you didn't say GPU power, you said SLI/CF.
SLI/CF
is GPU power, and it can still benefit those old games by running higher AA modes with no scaling required.
But, again, you don't spend any extra money.
Yes, I did. I spent it to get the same clock speed and cache size as the dual-core.
My point is that it's shortsighted (to say the least) to recommend a dual core over a quad core for a modern gaming machine (or pretty much any PC that has a focus of performing well in applications).
My point is that it?s shortsighted to be sinking money into buying more cores when you can still be buying as faster GPU system, because that?s where the real performance bottleneck is in games.
Again, quad cores don't cost more than dual cores.
To get the same clock speed and cache they most certainly do.
And again, so do you, more so because your benchmarks completely don't even consider real performance of CPUs in games, but for some reason you don't understand this (I thought my drag car analogy was pretty good, I guess not).
Heh, by ?real performance? you mean 1024x768 with no AA, right? :roll:
And a higher clocked dual core is NOT always faster than a lower clocked quad core.
I didn?t say it was; I said it was
almost always faster. It?s right there in what you quoted.
Again, referring to the article I posted, a 3.6GHz clocked Q6600 bested a 3.85GHz E6850 in 5/5 gaming tests; why do I have to repeat all this?
Overclocking is neither guaranteed or fail-safe, and that makes your results questionable at best since there?s no guarantee anyone using those processor will get the same clock speeds.
OTOH the stock results are guaranteed by Intel because the processor would be faulty if they didn?t work.
Why do I have to repeat all this?
Why should I post more benchmarks when you can't even be bothered to read the ones I have?
Because the ones you posted don?t back your claims yet you continue to pretend they do. Furthermore you fail to answer even basic questions.
It pretty much is though:
http://www.newegg.com/Product/...x?Item=N82E16819115037
http://www.newegg.com/Product/...x?Item=N82E16819115057
I realize the Q8400 has less cache, and you can probably save about $30 going dual core vs. quad core. So for $30 (in the scope of a typical $800-1000 build, mind you), you're giving up an entire extra dual core CPU, and all of the benefits it gives beyond just gaming.
It has more than less cache; the clock speed goes down from 3 GHz to 2.666 GHz.
That actually brings up another great point for my argument - physical limitations of the silicon. Generally, dual cores aren't going to overclock much further than quad cores of the same architecture. The architecture, whether in quad core or dual core form, has its limitations.
And yet on the flip-side quad-core has twice the chance of failure with overclocking given double the amount of cores have to be run outside of their rated spec compared to dual-core. In any case, my original point about running hardware outside of its rated spec still applies.
And, as I already stated and proved with benchmarks, the quad core doesn't need to run as fast as the dual core to outperform it, even in single-threaded applications (due to resource management).
And as I already stated, using settings no-one games at; these gains are nullified at any reasonable detail level.
They aren't invalid, they just aren't pertinent to what this enthusiast community is about. We wouldn't be having an involved discussion about the merits of CPU design if we were too scared/apathetic/dumb to not know about what goes on in that "metal box hooked up to the computer screen."
But overclocking isn?t guaranteed, regardless of whether you?re an enthusiast or not. Stock speeds are because the processor is faulty if it doesn?t hit them.
Then why do you try to draw inferences from those results and attempt to make them the norm? If you?re admitting your examples aren?t robust then by extension the arguments you base on them aren?t either.
I didn't say anything even close to that, how on Earth did you come up with that. Again, overclocking limitations generally arise from an architecture, and lower-binned chips have further to go to reach that limitation, so by default they are better overclockers when compared using % increase over stock speed.
But again that isn?t guaranteed or fail-safe by any stretch of the imagination. The fact that those
particular processors overclocked like that for that
particular reviewer means nothing given they might not overclock like that for others, and given different processors from the same family might overclock differently again.
Again, you're just grasping at straws, I never made any statements close to that, so how did you come up with that?
So you?re using them as examples to back your claims, but then turning around and admitting these examples aren?t concrete or robust? Interesting.
I showed that the Q6600, clock for clock, is faster than the E6850 even in games that don't support quad core.
Yeah, at settings no-one games at, at overclocked levels that aren?t guaranteed, not even by Intel.
Cool, then why do you recommend paying the same price for two less cores if the performance of the CPU is irrelevant (because quad core shows tremendous benefits outside the scope of gaming).
I don?t; I recommend buying the fastest clocked dual-core rather than trading cores for clockspeed because if you do happen to run into a CPU limited situation, in the majority of cases clock speed will win over quad-core.
Again, you need to learn more about computer hardware and how it pertains to the performance of applications (gaming or otherwise) if you want to stop making incorrect statements.
Again you fail to answer simple question and run around playing rhetorical games.
Again I?ll ask: do you game at 1024x768 with no AA?
Do you expect anyone with a quad-core to be gaming at such levels?
They?re very simple yes/no questions, so I fail to see why you have such difficulty answering them.
It wasn't shown to be an example of real world gaming,
Oh, so you admit that your results don?t correlate with reality then? Wonderful. By extension your arguments don?t either since they?re based on your examples.
but to prove the fact that the quad core is a faster and better gaming processor, something you insisted a dual core was. Obviously it doesn't make a huge difference in real world gaming, but it just negates another one of your points (the dual core being faster in most games).
Not dual-core per-se, but a higher clocked dual-core.
Your five games are but a drop in the bucket compared to the thousands of titles out there that react more favorably to higher clock speeds than four cores. But again these situations are largely irrelevant since in real-world situations the GPU will be the primary bottleneck, which brings me back to my original point of quad-core being largely worthless for gaming.
And if I do truly run into a CPU limited situation, I?d much rather take the guaranteed performance gain from a higher clock speed rather than hope I?m multi-tasking at the time and/or the process I?m running benefits from four cores.
Again, if the quad core is faster in games (and much faster in everything else), and costs the same price, why are you recommending to get a dual core?
It doesn?t cost the same because you?re not getting the same thing but with two extra cores. You have to trade clock speed and cache sizes to get those two cores, and since the vast majority of games show no benefit from quad core, that trade off isn?t worth it.
Again, go read the gaming benchmarks. Quad core is faster in 5/5 gaming benchmarks.
Nope. I see at least three games where dual (stock) outruns quad (stock).
Or are you again referring to the overclocked results where you yourself admitted they weren?t even robust?
Except that they aren't - canned benchmarks barely stress (if at all) the CPU.
What the hell are you talking about? There?s nothing ?canned? about them, especially if they?re manual game runs.
CPU stress comes from real world gaming, where dynamic environments, physics, AI, etc. must all be calculated in real time (instead of just read and processed)
And what do you suppose a manual fraps run is?
Yes they are. If people use the settings my tests are run at, they?re real-world. Yours aren?t because people that buy quad-core systems aren?t likely to be gaming at 1024x768 with no AA.
But they're much more relevant than yours.
No they aren?t because quad-core users don?t game under the scenarios you presented.
It?s like claiming a Ferrari is slower than a Jeep and providing examples of off-road racing. Those examples irrelevant because people don?t buy Ferraris to take them off-road.
My benchmarks at least remove the GPU boost, and show a portion of the work the CPU does (the read and process part). Again, you need to learn more about hardware and how it pertains to application performance.
But that portion is largely nullified in the real-world because the bottleneck is with the graphics system.
Again, you need learn more about how bottlenecking works, specifically with the slowest part playing the biggest influence if the rest of the system is waiting for it. Amdahl's law and all that.
Unless you test using real world gaming or benchmarks specifically designed for the CPU, it's laughable to generalize that canned GPU benchmarks have an significance regarding the performance of CPUs in real world gaming.
The only thing that is canned here is your artificial scenarios that quad-core users don?t use. Again, I?m talking about the real-world performance impact to gaming and how an actual gamer is impacted by quad-core. What are you talking about exactly? Some artificially contrived scenario that is meaningless in the real world, just to prove a point?
The reason to get a quad core is that, for a similar price, they give you the capability to receive massive performance boosts should a game support quad core (a situation that will only become more pertinent in the future).
But they don?t, not unless you game at settings quad-core users don?t use.
Again I?ve repeatedly asked to see these ?massive performance gains? but you?ve yet to show them.
Also games like Arma A 2 slideshow even with quad-core @ 4 GHz, so if future games follow that trend, current quad-cores will be obsolete for them.
Most were answered in the above replies,
Nope, you just continue to play rhetorical games.
but let me just reiterate so that I (hopefully) won't have to later - you need to learn more about computer hardware and how it affects game performance.
Aha-haha. And you need to learn how synthetic and artificial scenarios like yours mean nothing in the real world.
The significance of the tests were to highlight some of the performance difference between the Q6600 and E6850 when it comes to gaming.
But again those performance differences aren?t real world because quad-core gamers don?t game at those settings.
I'd imagine it'd be tough to if all you do is run canned GPU benchmarks, I can see why.
I?d imagine it?d be tough for to you to understand how your 1024x768 results are irrelevant. Oh wait?you don?t run those setting yourself, yet you continue to preach them.
I'm talking about GPU performance, I'm saying that in order to get the best gaming CPU performance, you need to get the fastest CPU architecture (intel i7). And this is very pertinent in overall gaming performance - look up some benchmarks on i7 unlocking the power of high-end SLI set-ups. Actually, I'll do it, since you seem to like to be spoon-fed information (and then interpret it incorrectly):
Again, I never argued quad-core isn?t theoretically faster, only that at real-world settings the GPU is by far the biggest bottleneck in practice.
But yes, if you have the most powerful GPU configuration available (quad CF/SLI) then by all means, buy a quad-core given you?ve already maxed out your graphics configuration.
I?m referring to the frequent situations of someone getting a quad-core system and then pairing it with something like a 4850. What a waste of time and money. Dropping back to a higher-clocked dual-core and getting a GTX285/4890 is far better for gaming.
That's the exception, not the norm. Programmers don't waste time writing in multi-CPU support for the hell of it.
Multi-threading something doesn?t imply you?re supporting multi-core. I would suggest you do some basic research about multi-threading because your comments are laughably inept on the subject you so furiously argue.
In fact, why don?t you show us a commercial Win32 game that
doesn?t use multi-threading? Heck, even X-Wing95 uses seven threads, and that game is about thirteen years old.
Again, simply threading something does
not mean you?re supporting multi-core.
I think John Carmack said it best:
?and it?s not ?oh just thread your application?. Anyone that says that is basically an idiot?