• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

In the near future will graphics cards have the processing power of an i7?

marc512

marc512

marc512

Associate
Joined
15 Nov 2008
Posts
1,911
Location
Scotland / Blairgowrie
Stupid question but do you think graphics cards in the future ( 10-20 years) have the processing power of an i7?
You will think now "impossible" but look at how much graphics cards have changed. Before processors used to be 700mhz. Now graphics cards can now have 700mhz. And now there getting close to 1.4ghzs which is the same as a pentium 4 processor ( i think? ).
Remember its the future. Not todays technology :)
 
They're already considerably faster in certain applications, the speed they run at are largely irrelevent and definitely not comparable.
 
I think the fastest CPU at the moment is about 70GFlops (0.07TFlops) and the quickest GPU is about 5TFlops. For certain applications the GPU is already much faster, read up about PhysX on the GPU and it's speed compared to the CPU.
 
as mentioned above gpus are several hundred times faster already, but only for simple processes, not for everything unfortunately ;)
 
GPUs and CPUs are totally different. Comparing the two is like comparing a car and a train - they do basically the same thing but in totally different ways and there are inherit advantages and disadvantages of both.
 
nvidia tesla might use a linux alternative but thats a bit high end and considering all current home pc's have to use a motherboard with an arm or x86/x64 based cpu there isn't really a call for it. GPU's are an accessory for a pc not the basis of one, you can run a pc without a gpu but not the other way round.
 
GPUs are extremely specialised towards... er.. graphics calculations. They can perform many times more FLOPS than traditional processors but that's only a very small part of it. I believe Intel's Larrabee is supposed to narrow the gap between the two and allow far more programmability on the graphics card. Currently, GPGPU (General Purpose computation on the Graphics Card) is very difficult and still only allows a few applications.
 
ah ok that explains things D: i always thought there processors where ment for graphics therefor not needed to be so high, and when having something more powerful (like an i7 in a graphics card) that would make it even more powerfull to play games but seemingly there better. Never knew that!
 
Apart from the already mentioned point that GPU's are highly specialised, highly parallel number crunching processors, whereas CPU's are general all-round very versatile 'jack of all trades' processors, there's another consideration in scaling a GPU to I7 speeds: heat.

Consider the size & weight of the coolers we now need to put on fast CPU's, and imagine trying to design a graphics card with a similar sized cooler!

The thing is though, both Intel and AMD are moving towards integrating the CPU & GPU onto the same chip, negating the need for a graphics card altogether. So we will effectively have GPU's running at I7 speeds in a few years time as part of these new generation CPU's. I'm very curious about how this will work out actually, as the GPU will benefit from a faster clock speed, a direct link to the CPU and system RAM, but will the RAM be as fast as the GDDR stuff we get on graphics cards? I wonder if we will end up with motherboards which have regular RAM slots for main memory and additional faster slots for graphics memory?
 
Toastor said:
Consider the size & weight of the coolers we now need to put on fast CPU's, and imagine trying to design a graphics card with a similar sized cooler!
Click to expand...
Yes?

I can't see a graphics card moaning about having an intel stock 1366 heatsink on it. Designed to be sufficient for the i7s under load, some graphics card coolers have to get rid of even more heat than they do.
 
Take a look at the raw power available to both:

http://en.wikipedia.org/wiki/Transistor_count

The Fastest consumer CPU at the moment the i7 has a transistor count of 731,000,000 at a 45nm process, the fastest GPU the 5800 range is 2,154,000,000 at a 40nm process. GPUs are quite a lot more powerful and have been for a while now, they're not quite as general purpose they're good at some specific tasks, especially ones which allow code to be parallelized.

GPGPU is nice but I don't think it's quite there yet, once we can compile most of the popular programming languages to use the GPU then it will really take off.
 
marc512 said:
Stupid question but do you think graphics cards in the future ( 10-20 years) have the processing power of an i7?
You will think now "impossible" but look at how much graphics cards have changed. Before processors used to be 700mhz. Now graphics cards can now have 700mhz. And now there getting close to 1.4ghzs which is the same as a pentium 4 processor ( i think? ).
Remember its the future. Not todays technology :)
Click to expand...

It's not all about Clock speed you know. Intel P4's started around 2.4GHz (iirc), and their 3.4Ghz CPU's were matched (easily) in most applications and games by AMD's 2.2GHz models (hence why they branded them the Athlon 64 3400+)
 
PrincessFrosty said:
GPGPU is nice but I don't think it's quite there yet, once we can compile most of the popular programming languages to use the GPU then it will really take off.
Click to expand...


GPGPU isn't for general purpose tasks like a x86/64 processor is, and just directly comparing transistor counts isn't going to be much use either.

There are inherent limitations to GPGPU that you'll never be able to get around, therefore you'll never have a full implementation of of a general purpose high level programming language.
 
You must log in or register to reply here.
Back
Top Bottom