Data Hazards Pipeline MC and Architecture
- 1. Data Hazards An overview of data hazards in computer architecture and their impact on execution. Aashiq Ilahi 230071601246 230071601234 23007160197 Syed Ashiq Ameer Thananjai
- 2. • In computer architecture, data hazards occur when there are dependencies between instructions that can lead to errors or delays in execution. Data Hazards
- 3. • Read After Write (RAW) hazard, also known as true dependency, occurs when an instruction depends on the result of a previous instruction. This situation arises in a pipeline when the data being read has not yet been written back by the preceding instructions. Proper handling of RAW hazards is essential to maintain data integrity and correct program execution. Read After Write (RAW) Types of Data Hazards
- 5. • Write After Read (WAR) hazard, also known as anti-dependency, arises when an instruction writes to a location before a previous instruction has read from that location. This can lead to incorrect program behavior, as the reading instruction may not access the original data. Techniques such as instruction reordering can help mitigate this hazard in pipelined processing. Write After Read (WAR) Types of Data Hazards
- 6. • Write After Write (WAW) hazard occurs when two instructions attempt to write to the same location, with the order of writes affecting the final outcome. This type of hazard can cause incorrect results if the writing instructions are executed out of order. Managing WAW hazards is vital for maintaining data consistency in pipelined architectures, often requiring techniques such as careful scheduling and resource management. Write After Write (WAW) Types of Data Hazards
- 7. • Data forwarding, also known as data bypassing, is a technique used to reduce the impact of data hazards by directly connecting outputs of one instruction to the inputs of another without writing to memory. This allows subsequent instructions to utilize the most recent data quickly, thereby reducing latency and enhancing overall execution speed in a pipelined environment. Data Forwarding Handling Data Hazards
- 9. • Pipeline stalls are a straightforward method to handle data hazards, where the pipeline is temporarily halted to allow time for prior instructions to complete their execution. Although this approach can effectively prevent hazards, it may lead to performance degradation due to increased cycle times and resource underutilization in scenarios with frequent stalls. Pipeline stalls Handling Data Hazards
- 11. Questions 1. What are Data Hazards? Mention it’s types. 2. Compare ‘Write after Read (WAR)’ data hazard and ‘Write after Write after Write (WAW)’ data hazard 3. What are some methods to handle Data Hazards?
- 12. Thank you!