Describe the role of GPU

Features of GPU

• Highly parallel structure, ideal for complex graphic calculations.
• Can be integrated in CPU or discrete (a separate card).
• Communicates with software via APIs like DirectX or OpenGL.
• Increasingly used for machine learning and other computationally intensive workloads — not just graphics.

GPU architecture

• Distinct architecture sets GPUs apart from conventional CPUs.
• Processes large blocks of data concurrently (同时的) → more efficient for certain task types.

GPU Features	Characteristic	Example
Parallel Processing	Thousands of small cores designed for parallel computing	Processes multiple pixels at the same time, reducing image filter time
High Throughput (吞吐量)	Optimized to process large amounts of data simultaneously — useful for rendering/complex calculations	Renders complex 3D scenes in real time; calculates color/position/texture of thousands of pixels concurrently
Memory	High-speed memory (VRAM) handles large datasets and textures for high-res video and scientific calculations	VRAM stores/manages textures; high speed enables real-time high-res rendering without buffering

Real World Applications of GPU

• Video game graphics
  • Renders high-res textures, realistic lighting, smooth frame effects.
  • Higher frame rates; offloads rendering from CPU.
• AI and machine learning
  • Faster processing of large datasets.
  • ML training involves extensive matrix multiplications that parallelize well on GPU.
• Scientific computing and simulations
  • Quicker physics simulations and climate modeling.
• Graphic design and video editing
  • Real-time 3D modeling/visualization.
  • Software like Blender uses GPUs to render complex scenes (lighting, shadows, textures) in real-time (实时的).