Nvidia Says Core i7 Isn't Worth It

Terrible article. How do some of you actually get jobs posting this garbage?

For once I actually agree with Nvidia. You buy gpu's to game, not cpu's. Put it this way, whatever lower priced chip you buy, you spend on a higher priced gpu. GPU's do games, cpu's do the rest.

Show me one gaming benchmark of an i7 920 with a gtx260 beating a Phenom II 940 with a gtx285 or drop the i7 fanboism altogether.

Is frame rate an accurate measure of a games goodness?
Once your GPU is good enough to run at your monitors refresh rate (at the resolution you want) what's the point in going higher?
Surely games get better because the logic behind the pretty pictures is better - the physics, the AI, etc.
It's not an answer to GPUvCPU arguments 'cos either can be used for this logic, but it is an argument for a better metric.

in this times of crisys, everybody fights for it's part of the bread (no cake anymore) and companies are attacking themselves more and more often and more aggressively. sometimes, like this one, it's even silly and childish

It was silly of Intel to make such a rash claim, they would of been shot down on this even if we were not in a crisis, The statement made by Intel was misleading.

There is an inherent flaw in using benchmarking software and claiming results from it, that is why sites like Tom's hardware and others use real applications and games when testing equipment.

Nobody doubts the i7 is a great CPU. but it will never increase frame rates to any major extent, that is the realm of the GPU. Physics are handled by the GPU now.

It is the marketing people who are at fault, they grab one little score and blow it out of proportion, I personally think Nvidia did the right thing exposing this false claim, even though we all knew it was false, maybe in future Intel will be a little more careful with rash claims, but I doubt it...

Ok I'll come in and protect Intel, ok an 80% increase is possible and has been seen, when playing at 640x480 where it is more cpu bound.

however what Nvidia has done with that claim is said 'Don't SLI GTX 260's, a single GTS 250 is actually good enough for most eyes'

Honestly, it all depends on the game. Take World of Warcraft, in raids or in towns where you see lots of characters on the screen at once, a CPU upgrade is three times as effective as a GPU upgrade.

I know that because I first upgraded my 7600GT to an 8800GT, and my framerates hardly went up in towns or raids at all. Then I upgraded my P4 3.2 to a C2D E4500 (and overclocked it to 3.2).
Perfect framerates all of a sudden.

That seems to be the case with GTA IV..AMD Phenom II's being on par with the faster (AMD much more expensive) Intel i7.

Source: http://www.tomshardware.co.uk/phen [...] 62-10.html

Right from day one THG was telling people about CPU and graphics card 'scaling', about the benefits of a fast CPU being largely redundant past a certain resolution; the graphics card's bandwidth unable to keep up despite the colossal power of it's multiple cores.

I do however disagree with NVIDIA's stance on Quad vs. Dual-core CPU's, specifically their view that a dual-core is just as good if not better.
Look at GTA IV again, and you'll see that at lower resolutions (i.e before the graphics card fill rate comes into play), there's a distinct advantage to having 4 cores over two. So again, a cheap AMD quad outclasses the stronger cores of Core 2 duo.

I doubt very much we'll be having this kind of debate soon, however. Look at the way more and more software is being ported to run on GPU's.

This is the reason I can - if possible - avoid LCD monitors. Older CRT tubes look better to me, at non-native resolutions. I know that it's in effect driving the technology faster, I just don't relish having to buy a very fast, power-hungry card just to get a good picture when that game can play faster and look just as detailed if I play it on an older screen technology. I am not a fan of LCD's (plasma seems better), and I think there are instances whereby they simply suck, e.g. webcams.

Look at the environmental considerations..there's nothing to touch the energy efficiency of a GPU. Who wants to consume ridiculous amounts of electricity just to get an i7 to 4GHz, when GPGPU acceleration such as CUDA leaves it for dust?

This is where I think Intel could really screw up big time..look at the experience and the performance of current nVidia and AMD/ATI solutions. Intel will bring something good out I reckon, but like the i7 it may be considered by many to be too little, too late. Of course, AMD isn't going to race to the finish line just yet..bring out an integrated x86+GPGPU solution and what do you do with your current x86 portfolio?

Speculating again, but I think this stuff will be the preserve of cloud computing, as well as government 'research' geared towards making the net a less neutral, accountable medium on which we communicate with each other. The carrot will be the blistering net speed + the ability to outsource your geometry and rendering load, as well as your transcoding tasks to a remote server (a super-computer based on 1000's of GPU's). What do you think?

The problem with NVidia's analysis is that they've carefully selected the games.

They should have shown the results for Flight Simulator X which is actually worse off when you have Crossfire or SLI.

This is a heavily CPU-bound game and i7 would be much better than 2 Duo, only requiring a single 4870 or 260.