[H] Core 2 Gaming Benchmarks
- Thread starter Skrying
- Start date
$1,300 and you don't even get a cooler with it. 
Wheee. [H] is getting more aggressive in their language on pushing their benchmarking methodology as the One True Faith. "Lie" is getting tossed around there pretty liberally about others.
Wouldn't you want to see multi-gpu results before you came to such conclusions as they come to tho?
The AMD fanboi brigade has their
Kyle t-shirts on this morning, no doubt.
Wouldn't you want to see multi-gpu results before you came to such conclusions as they come to tho?
The AMD fanboi brigade has their
Kyle t-shirts on this morning, no doubt.geo said:Wheee. [H] is getting more aggressive in their language on pushing their benchmarking methodology as the One True Faith. "Lie" is getting tossed around there pretty liberally about others.
Wouldn't you want to see multi-gpu results before you came to such conclusions as they come to tho?
The AMD fanboi brigade has their
I agree with this post as well.
One thing Kyle has not taken into consideration is G80 and possible quad Sli. 4 x G80 will not be as frame rate limited ( you have to suppose ) and we know dual Sli takes a fair bit of cpu power .. does quad SLi take 2 x that. If so then I think I would be happier with the most powerful cpu I could get my hands on ..especially if it saved on power as well so my mighty 4 G80's could have just a bit more.
I'm not sure I buy this argument. I mean sure, a faster CPU will last longer, but it will also cost more, and as you get toward the high-end cost increases more rapidly than performance. I don't see how the economics can work out.Chalnoth said:
There's a minimum performance delta which is worthwhile longevity-wise. A CPU with 10% extra performance gets you what in extra longevity? Pretty much zero I'd say, in practical terms. Both are going to be hung out to dry by whatever the next game is going to be; that extra 10% isn't going to make the unplayable playable.
So how far do you have to go? 25%? 50%? 100%? Want a processor twice as fast? You're going to pay a lot more than twice the money, if you want it now. Or you wait 18 months, upgrade and get twice the performance for the same money you just paid, less if you sell your old kit on eBay, less still if AMD and Intel are having one of their silly little price wars!
While the HOCP approach is flawed, - running a game at GPU limiting settings, they do one thing correct: They don't used canned demos. Demo playback is little more than feeding geometry to the GPU (and a bit of skinning and shadow volume extrusion in Q4 games, but alas). Canned demos do not exercise the most important parts of a modern game engine: AI and physics.
But even then Core 2 still wins by a big margin, as an example see Tech report's Oblivion tests.
Cheers
But even then Core 2 still wins by a big margin, as an example see Tech report's Oblivion tests.
Cheers
Talk about cherry picking testing situations to arrive at your desired results... again.
Beyond the use of situations where you are GPU limited, the benchmarks are even more misleading due to the lack of other processors. Lets toss in some Pentium Ds and mid range Athlon64s. Based on benchmarks I have seen elsewhere in GPU limited scenarios your lower end CPUs are going to be very close to the high end.
Does this mean a 30% faster CPU is irrelevant? Of course not.
Which leads to questions like: If this reasoning brings us to a conclusion that these high end CPUs are worthless to gaming, why are we comparing $1000 CPUs to $500 one?
Reading comments like this about the pricing disparity in Intel SKUs raising an interesting point:
Yes even when he admits the cheaper CPU from Intel beats the Athlon64 FX-62 he does not harp on the same price disparity. Instead we read:
Nevermind that Conroe is showing to be a much better overclocker. So why does the AMD chip, at 2x the cost, get the benefit of the doubt, "But you could just overclock it" and yet the same arguement is not put ahead for the cheaper and better overclocking CPU?
At least they do link to the CPU scaling article at the end, but of course the lack of user friendly bard graphs to use side by side the entire spectrum (which Firingsquad, Xbitlabs, etc use) make it much harder to follow the obvious conclusion that in most games you are GPU limited unless you have the absolutely newest and great graphics hardware, and not even then at times.
Kind of sad that average enthusiests like those of us on these boards can identify procedural and logical issues with benchmarks from people who basically get paid to do this and take such dogmatic stances. The points he chooses to accent when and where in his article and the contrasts he chooses to draw out are the most telling.
Beyond the use of situations where you are GPU limited, the benchmarks are even more misleading due to the lack of other processors. Lets toss in some Pentium Ds and mid range Athlon64s. Based on benchmarks I have seen elsewhere in GPU limited scenarios your lower end CPUs are going to be very close to the high end.
Does this mean a 30% faster CPU is irrelevant? Of course not.
Which leads to questions like: If this reasoning brings us to a conclusion that these high end CPUs are worthless to gaming, why are we comparing $1000 CPUs to $500 one?
Reading comments like this about the pricing disparity in Intel SKUs raising an interesting point:
Yes even when he admits the cheaper CPU from Intel beats the Athlon64 FX-62 he does not harp on the same price disparity. Instead we read:
Nevermind that Conroe is showing to be a much better overclocker. So why does the AMD chip, at 2x the cost, get the benefit of the doubt, "But you could just overclock it" and yet the same arguement is not put ahead for the cheaper and better overclocking CPU?
At least they do link to the CPU scaling article at the end, but of course the lack of user friendly bard graphs to use side by side the entire spectrum (which Firingsquad, Xbitlabs, etc use) make it much harder to follow the obvious conclusion that in most games you are GPU limited unless you have the absolutely newest and great graphics hardware, and not even then at times.
Kind of sad that average enthusiests like those of us on these boards can identify procedural and logical issues with benchmarks from people who basically get paid to do this and take such dogmatic stances. The points he chooses to accent when and where in his article and the contrasts he chooses to draw out are the most telling.
dizietsma said:
Not to mention that most people don't buy a cpu for 2 months, or 4 months, or even 6 months. They buy it for a year, or two years, or three years. There are exceptions, of course, but most people I hang out with hold onto their cpu's longer than their viddy cards.
It's not common for two CPU's at roughly the same performance level from each manufacturer to have the same cost to the consumer, or for two CPU's at the same cost to have the same performance (particularly when you factor in such user-specific things when upgrading like the motherboard and memory).nutball said:
But then it's not deterministic, is it? More importantly, if you reduce resolution from 1600x1200 to 800x600 (so you're reducing the GPU load by a factor of four) and running a demo without a framerate limit, which is just feeding geometry, wouldn't you increasing the CPU load significantly (up to a factor of four)?Gubbi said:
So, my point is--does it matter for comparison purposes if it's not stressing AI or physics bits of an engine? Is a specific type of load more important than the amount?
That's a bit of a problem I have with the article too, as well as why his sudden switch in stance on the enthusiast crowd.Acert93 said:
When SLI was all the rage and anyone knocked it for being too expensive it was all about how some people like to game at 16x12 with lots of eye-candy and could use it with their FX cpus and such...even though that is the exact same crowd that will be eyeing the Conroes in just as lustful an e-peni fashion.
I think Kyle's got a valid point this time, he just didn't go about making it in the best way. His article wasn't a review of the Conroe, it was more like a disproof of it.
Conroe is without a doubt gonna be the best for a while, but I still might pick up a 3800x2 personally.
nutball said:
umm
i disagree. as a computer person (haha we all are, but i did it professionally a good bit, you know, just sayin) i'd say that the ppl i've met (aside from online e-penis dimwits) who bought highend CPUs have all done it (against my advice) for one reason -- they want to slap it in (or me to do that, really) and not look at it again for 2 or 3 years at least.
as opposed to me -- i buy a cpu every year or so, but i buy them all dirt cheap and save a bundle and get the same performance -- but then, i know how (not that it's hard, but that's ppl for ya!).
I'd rather spend $100-150 a year on a new CPU than spend $500-750 every 5 myself too.
Errr... yeah I know people like this too. That doesn't make it a sensible or economically justifiable choice though. I mean presumably you advised them against it because you know it's stupid?poopypoo said:
Chalnoth said:
Yes I also swap CPUs about once every 18 months. Generally speaking I expect to get twice the power for the same price, and generally speaking that's exactly what I get.
But whatever you think of what pricepoint people should go to, the HardOCP article just does a horrible job as a CPU review. It's halfway-decent as an article showcasing that a fast CPU doesn't do much for gaming, but to really say that it needs to make use of lower-end CPU's, so that people can get an idea of where the cutoff point is for current games where performance starts to drop.
As it stands, I don't think anybody wanting to buy a new CPU would find the HardOCP article informative.
As it stands, I don't think anybody wanting to buy a new CPU would find the HardOCP article informative.
And so the world turns. First it was GPU limited reviews showing that P4 is not-so-bad compared to Athlon XP/64. Now the situation is reversed and those same GPU limited benchmarks are being used for damage control. It's irrelevant though since even at performance parity in GPU limited games, Intel's new baby has so much going for it that it's going to win over many enthusiasts anyway. Hopefully this will result in some ridiculously low 939 X2 prices for those who stick with AMD.
We'll see if that remains true once AMD announces their price cuts at the end of the month. It is, of course, unlikely that AMD's FX62 will be able to compete on price with Conroe, but that was never the processor's purpose in the first place. More interesting will be the more mainstream designs.
Similar threads
