# Title: GPU. Lesson 7. GPU Core Architecture: Core Components and CPU-GPU Differences.

**Title:** GPU. Lesson 7. GPU Core Architecture: Core Components and CPU-GPU Differences.

1. **GPU Core** design optimizes parallel processing for high-throughput tasks.
    
2. **CPU Core** excels at sequential tasks with complex logic and branching.
    
3. **ALU (Arithmetic Logic Unit)** in GPUs performs simple operations in parallel.
    
4. **Control Unit** in GPUs minimizes complexity, focusing on large data processing.
    
5. **Register File** stores immediate data for fast, local access by each GPU core.
    
6. **Warp Scheduler** organizes threads in groups, enabling massive parallelism.
    
7. **Execution Units** in GPUs handle many simpler tasks simultaneously.
    
8. **SIMD (Single Instruction, Multiple Data)** units enable efficient data parallelism.
    
9. **L1 and L2 Caches** are optimized in GPUs for high-throughput, bulk data.
    
10. **Latency Tolerance** in GPUs enables handling of multiple threads with delays.
    

---

**Technical Examples:**

1. **SIMD in GPUs vs. CPUs**: SIMD units in GPUs are optimized for many simple calculations, while CPUs focus on fewer, complex tasks.
    
2. **Warp Scheduler and Latency**: GPUs use warp schedulers to keep data flowing despite delays, unlike CPUs that prioritize low-latency tasks.
    
3. **ALU and Execution Units**: GPU ALUs handle many parallel calculations, while CPU ALUs are optimized for single-threaded, complex operations.