-
Research and watch basic intro videos on how GPUs and how they work
IBM Intro Video
What is a GPU
- Graphics Processing Unit
- CPU →only a few cores
- Serial computation
- App code to the CPU
- GPUs →hundreds of cores
- Parallel computation →all at once, very intense computations
- Providers →AMD and NVIDIA
- VDI →Virtual Desktop Infarstructure
- High intensive graphics
- 3D viewing
- Rendering
- Gaming
- AI →Artificial Intelligence
- Deep Learning
- Machine Learning
- Creating specifice applications
- Create and train neural networks
- Something a CPU cannot due on its own
- HPC →High performance computing
- Company's ability to spread out its computing power between nodes
- Cloud computing, often depends on GPUs
- Why use GPUs on cloud?
Android Authority
- Very good at certain forms of math
- 3D graphics → everything is triangles
- Might need to be transformed
- Using a vector
- Or multiply by a scaling matrix
- Needs to be good at floating point, vector multiplaciion, matrix multiplication
- Program shader
- Vertex shader
- Runs once per frame
- 3D end
- Pixel shader or fragment shader
- Color texture lighting and shadows that are thrown
- Controlls
- Runs once per pixel
- CPU vs GPU
- CPU is serial
- GPU is Parallel
- Multiple execution engines →execution cores
- GPU commuting
- Task that has to be done highly in computing
- Certain types of computer vision, types of of machine learning, etc
- 3D graphics on desktop vs phone
-
Research GPU to CPU pipeline
-
Add research and resources to appropriate Notion pages
-
Follow research path based on interests and appropriate findings (eg PCIE, DMA, specific areas that could be written about in a Zine)
-
Think about what the final deliverable might look like