what exactly does flops?

[WarWeeny]

i hear stuff about this thing can produce 3 tflops and is so much better then your x tflops!

 

but what exactly is the point of having a high g/t flop count?

 

example:

the ps4 has 1.84 tflops but if a videocard that has, lets say 1.5 tflops, would the ps4 performance be alot better then the 1.5 tflop card in terms of raw perfomance (gaming perfomance included) or is it just a gimmick that isnt really important at all when choosing videocards?

 

 

i never understood flops and its function in the gaming world, so i hope someone can give me an answer

[Andrewr05]

F.L.O.P.S

 

Floating-Point Operations Per Second

 

It's not video card related per se, but relevant to ALL processors that do floating-point calculations. Be it mega, giga, tera or teraflops.

 

 

 

With so many pixels on screen for gaming and physics calculations and advanced behind the scenes architecture GPU/CPUs need to be able to calculate more and more.

 

But I would never buy a GPU just because it says it calculates more xFLOPS than its competitor, there are more variables than just raw power.

[El_Capitan]

It's a measurement that Andrew pointed out. For gaming, GPGPU is utilized for special rendering tasks, if the game allows for it, like fluid, smoke, and particles with PhysX using CUDA on Nvidia cards. AMD is more efficient with OpenCL.

 

Most of that is done with Single Precision Floating Point, depending on how the programmer utilizes the API. Which is why the GTX 600 cards gimped their Double Precision Floating Point numbers. Am typing this on my phone, so not being so detailed.

 

Most complex compute tasks are done in Double Precision Point algorithms. This is why if you fold with the GTX 600 cards, they're not as fast as GTX 500 cards, even though the CUDA numbers are higher. And AMD's HD cards kick butt with compute tasks using OpenCL. OpenCL is getting more widely utilized, which is good for AMD.

[Guest_Jim_*]

F.L.O.P.S

 

Floating-Point Operations Per Second

 

It's not video card related per se, but relevant to ALL processors that do floating-point calculations. Be it mega, giga, tera or teraflops.

 

 

 

With so many pixels on screen for gaming and physics calculations and advanced behind the scenes architecture GPU/CPUs need to be able to calculate more and more.

 

But I would never buy a GPU just because it says it calculates more xFLOPS than its competitor, there are more variables than just raw power.

Continuing from what Andrew said, it is a statistic for all processors, but of varying degrees of importance. GPUs need to be able to do a lot of math all at once, but very often it is relatively simple math, while CPUs need to do more complicated math, but less of it at a single time. This is also why Folding at Home runs faster on GPUs (when properly optimized) than on CPUs.

Right now, Andrew's point about purchases is correct, that FLOPS should not be determining what GPU you buy (unless you are doing specialized work, in which case you wouldn't have asked your question). In the future though, that may change as we get the next Playstation and Xbox hardware, so multiplatform games are not limited by the old, current generation specs. How far in the future is hard to say though, but my guess is over a year (possibly multiple years).

[cjloki]

Good question war, I learned from these smart guys too !...they know their stuff at OCC

[WarWeeny]

Good question war, I learned from these smart guys too !...they know their stuff at OCC

 

that is why i posted it here ;P