• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

are current gpus more advanced than current cpus?

T

Tempered81

Diamond Member
I've been hearing about this gpgpu computing and how great it is for supercomputers to run applications and stuff on modern gpus


does this mean geforce 9 and radeon r700 are more advanced than phenom and yorkfield?

surely not, right?
 
like Fletcher here said, they are designed for different things.

but you could argue that GPUs have far mroe cores, shading cores that is, and are especially useful at multi-core calculations.

also memory on GPUs is far more faster (about x2.0 for an average system) than system RAM.

if anytime in the future system RAM catches up to VRAM, i would like to see Video Memory expansion slots. like a place for a stick of ram near the VGA card. that way you would only need to buy the VGA core and you would have the option to fit it with any amount of VRAM you want.

games start lagging? double that VRAM! still no good? buy a new GPU, keep that VRAM!
 
Originally posted by: Borealis7
also memory on GPUs is far more faster (about x2.0 for an average system) than system RAM.
Click to expand...

Yes, but GPUs are designed for random memory accesses, nor are they designed for heavy branching. The effective latency and performance cost of reads/writes on a GPU is far higher compared to a CPU. CPUs have highly optimized caches, prediction, and pre-fetching abilities.

The current amount of system memory bandwidth is more than adequate for today's applications. This is why upping the memory speed at this moment only results in minor increases in performance; the CPU is not bottlenecked by memory bandwidth.

GPU = expensive read/write, cheap compute
CPU = cheap read/write, expensive compute
 
Originally posted by: jaredpace
does this mean geforce 9 and radeon r700 are more advanced than phenom and yorkfield?
Click to expand...

Advanced is a vague term... GPUs do more theoretical "operations per second," but have a relatively limited set of operations they can perform. Only certain types of problems can be solved by the highly specialized hardware we call a GPU, whereas the CPU has the flexibility to do absolutely anything, a little bit slower.

GPUs excel at performing the same calculation over and over on a fixed (or predictable) set of data, i.e. 3D objects. CPUs handle, well, everything else you've ever wanted a computer to do. 🙂
 
I hear that GPUs are more parallel than CPUs, what with all those shaders and whatnot. GPUs are, like everyone's said, better at doing floating point operations [hence numbers like 1TFLOPS being thrown around] and CPUs are better for general-purpose stuff, with low FLOPS numbers. For people who really need GPGPU and have a use for it, there's CUDA and the Tesla.
 
GPU = expensive read/write, cheap compute
CPU = cheap read/write, expensive compute
Click to expand...

Cell= Somewhere in the middle

Overall due to the nature of typical desktop computing the x86 processors are very close in design and how they are capable of performing, but CPUs can be made to be much closer to GPUs then what Intel or AMD offer. Cell would get wtfpwnd trying to mine data in Excel against a desktop PC CPU, it utterly obliterates that same desktop CPU in something like folding.

I don't think it is accurate to say that one of them are more advanced or better- they are very different types of processors and they are all extremely good at doing what they are supposed to do.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top