Model of Threading
Duration: 3 min
This video lesson is available to enrolled students.
AI Summary
An AI-generated summary of this video lecture.
The lecture introduces the concept of multithreading models, specifically distinguishing between user threads and kernel threads. It details the Many-to-One model, highlighting its limitations regarding blocking calls and CPU utilization. The presentation then transitions to the One-to-One model, explaining its advantages in concurrency and overhead trade-offs. Finally, it covers the Many-to-Many model as a hybrid approach combining the best features of the previous two.
Chapters
0:00 – 2:00 00:00-02:00
The lecture begins with a slide titled "Multithreading Models," defining user threads as application-level and kernel threads as OS-supported. The focus shifts to the "Many-To-One Model," where many user-level threads are mapped onto a single kernel thread. A diagram illustrates this with multiple wavy lines connecting to one 'k'. The instructor explains that if a blocking system call occurs, the entire process blocks, even if other threads could continue. He writes "T1L" and "T2L" on the diagram to label specific user threads and "P1" to denote the process. He notes that because a single kernel thread operates on only one CPU, this model cannot split processes across multiple CPUs. He mentions Green threads for Solaris as a historical example.
2:00 – 3:14 02:00-03:14
The presentation transitions to the "One-To-One Model," creating a separate kernel thread for each user thread. The diagram changes to show four distinct 'k' circles, each linked to a wavy line. The instructor highlights that this model overcomes blocking system call issues and allows splitting across multiple CPUs. However, he points out the significant overhead involved in managing these threads. He cites Linux and Windows 95 to XP as implementations. The final segment introduces the "Many-To-Many Model," which multiplexes user threads onto an equal or smaller number of kernel threads. The diagram shows multiple user threads mapping to three kernel threads. This approach combines the best features of previous models, allowing variable allocation of kernel threads based on CPU count.
The lecture provides a comprehensive overview of multithreading architectures. It starts by establishing the fundamental difference between user and kernel threads. The Many-to-One model is critiqued for its inability to handle parallelism and blocking calls effectively. The One-to-One model is presented as a robust alternative that enables true parallelism but suffers from high overhead. Finally, the Many-to-Many model is described as a hybrid solution that balances flexibility and performance by allowing the kernel to manage a pool of threads for user-level execution efficiently and effectively. This progression illustrates the evolution of threading strategies to optimize system performance and resource management.