This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: LSU EE 7700-1 Homework 7 Solution Due: 15 April 2009 Read NVIDIA GeForce 8800 GPU Architecture Overview, linked to the course references page, and answer the questions below. Problem 1: The 8800 implements a unified shader model , meaning it uses the same processor for both vertex and fragment (pixel) processing. What is the advantage of doing so. Hint: This is really easy and can be answered directly from the white paper. The GPU area needs to be divided between vertex and fragment shaders (and other items). If there are too many vertex processors then some vertex processors will go unused, the GPU would have been faster if the chip area used for vertex processors was instead used for fragment processors. The situation is similar if there were too many fragment processors. The problem is that its the application that determines how many is too many, so for a given GPU there may be too many vertex processors running application A and too many fragment processors running application B. There is no easy way to choose. With a unified shader model the same processor can be used for both vertex and fragment shading, so theres no problem. Problem 2: Describe at least two possible disadvantages of the unified shader model. These disadvantages might have outweighed the advantages in previous generation GPUs, which had separate (not unified) programmable shaders. Hint: This question is harder and requires some understanding of computer architecture....
View Full Document
- Spring '08