Announcement

Collapse
No announcement yet.

How does the GPU and CPU work together in the graphics pipeline?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    How does the GPU and CPU work together in the graphics pipeline?

    On most game engines, does the GPU take the brunt of the work, or does the CPU? Generally speaking, how do they work together in the graphics pipeline. I know this can be a very open ended question, but I'm curious as to whether or not buying a top of the line GPU to work in tandem with an older CPU (AMD Ryzen) will produce any bottle necks for a GPU like the 1080 Ti. This is specifically for gaming purposes.
    Programming today is a race between software engineers striving to build
    bigger and better idiot-proof programs, and the Universe trying to produce
    bigger and better idiots. So far, the Universe is winning. (Rich Cook)

    #2
    It depends on what the game is doing, for example things like crowds depend on the CPU more

    Comment


      #3
      A simple was of explaining it is that one thread on the CPU is telling the GPU what to do also known as draw-calls. As long as the CPU can handle the draw-calls in a timely manner, then the GPU can unleash its power to improve the fidelity of the image. The single core performance on Ryzen is not the greatest so it is very possible that the CPU will reach the upper frame-rate limit before the GPU.

      Comment


        #4
        Originally posted by darthviper107 View Post
        It depends on what the game is doing, for example things like crowds depend on the CPU more
        Why crowds?

        Comment


          #5
          Originally posted by Dudester01 View Post

          Why crowds?
          If I had to guess, it's because the CPU needs to do a lot more work in batching up and preparing the data for rendering versus rendering a small number of objects.

          Comment


            #6
            So by draw calls, does this include the CPU telling the GPU which textures to render?
            Programming today is a race between software engineers striving to build
            bigger and better idiot-proof programs, and the Universe trying to produce
            bigger and better idiots. So far, the Universe is winning. (Rich Cook)

            Comment


              #7
              Originally posted by junfanbl View Post
              So by draw calls, does this include the CPU telling the GPU which textures to render?
              It includes any command to the GPU really, so it would include things like:

              * Binding textures
              * Enabling/Disabling stencil buffer, depth test etc
              * Buffering vertices / indices
              * Compiling / loading shader programs
              * Switching render target

              Some are more expensive than others, but generally you want to be as efficient as possible with your draw calls, so try to avoid constantly binding and rebinding textures and so on.

              Comment


                #8
                Interesting, but very informative. Thank you. So, I'm looking for another opinion on the bottle neck issue. I was reading online that a GPU can draw graphics faster than a CPU can send the instructions. Does anybody else think that would be a problem with a Ryzen 2700X and a GTX1080 Ti? I don't want to get a high end card and not be able to maximize it. Any thoughts?
                Programming today is a race between software engineers striving to build
                bigger and better idiot-proof programs, and the Universe trying to produce
                bigger and better idiots. So far, the Universe is winning. (Rich Cook)

                Comment


                  #9
                  Intels i7-8700K would be more in balance with a GTX 1080 TI if you want to build a gaming PC. Ryzen is better at compiling and off-line rendering though. It all depends on the software it has to run though. Find some benchmarks to get a better idea of what to expect.
                  Last edited by GarnerP57; 06-28-2018, 03:28 PM.

                  Comment


                    #10
                    There is NO WAY a 2700X will bottleneck a 1080Ti (I notice no-one has asked what else is using PCI-E lanes on the mobo, which could well be a problem with an i7-8700K, or what the monitor resolution is, either).

                    If I were buying today, I'd get a 2700X and a 1080Ti. But only if my monitor was higher than 1080P.
                    I'm @londonisunreal, the organiser of the London Unreal Engine Meetup group.

                    Comment


                      #11
                      Originally posted by Jezcentral View Post
                      There is NO WAY a 2700X will bottleneck a 1080Ti...
                      Benchmarks tell a different story. The average FPS is the same but the peak fps is when the CPU starts to matter and the i7-8700 gets a little advantage because of its extra single core speed. The difference is negligible but it is there.
                      Last edited by GarnerP57; 06-29-2018, 01:41 PM.

                      Comment

                      Working...
                      X