Skip to content
RED SAGSkills
hardGame Developer interview question

Explain how a modern real-time rendering pipeline draws a frame, and where the common performance costs are.

What the interviewer wants to hear

  • CPU prepares visible objects: culling, sorting, draw calls
  • GPU vertex stage transforms vertices to screen space
  • Rasterisation and pixel/fragment shading
  • Forward vs deferred shading for lighting
  • Post-processing, shadows and transparency
  • Costs: draw calls, overdraw, shader complexity, bandwidth

Model answer

On the CPU side, the engine first decides what is visible using frustum and occlusion culling, picks LODs, sorts objects by material and transparency, and submits draw calls to the GPU. On the GPU, the vertex shader transforms each vertex from object space to screen space. The rasteriser turns triangles into pixels, and the pixel or fragment shader computes their colour using textures and lighting. In a deferred renderer, the scene first writes surface data to a G-buffer and lighting is computed later per pixel, which handles many lights well; forward rendering lights each object directly and handles transparency and MSAA more easily. Shadows are extra passes, and post-processing like bloom and tone mapping runs at the end. Common costs are too many draw calls, overdraw from transparency, complex shaders, high resolution and memory bandwidth.

Can you answer it under time pressure?

Take a timed hard round and get scored on every answer.

Start mock interview

More hard Game Developer questions