Principles of Operating System - Lecture 5

This lecture covers chapter 4 on THREAD programming and thread mapping to processes. An introduction of IPC "Inter-process Communication" is also described.

[jao]

You're so good, love this. And i love how you randomly laugh at things, it makes you confused and then you try and listen ever harder to understand what it is you're cracking up about. That way you concentrate even harder on the lecture!

[KatrinaFishman]

Where have you been throughout my entire college career!? Thank you so much!! Very helpful!

[bjhecker]

The CPU runs one process at a time. You can have multiple threads running inside of a process but only one thread will run at a time as well. Each CPU can only execute one command at a time. It's a serial processor. This is why there is CPU Scheduling algorithms. These algorithms make queues of the instruction sets so they can all run to accomplish the task of the process. Each instruction runs one by one in a serial way depending upon how the CPU schedule decides how to run them.

[bjhecker]

You could do this and probably never run out of new concepts since every year new Operating Systems and advancements in computing are made.

[bjhecker]

With multiple CPUs you could run multiple threads in parallel. It's just the single CPU when scheduled only runs one process at a time. If you have multiple threads mapped to one process only one will run at a time.

[kamoliddinshukurov]

thank you very much! I like every of your Lectures

[TEJASCHOPRA]

Thanks for the detailed explanation!

[kevinparsley]

[mutual exclusion/ racing conditions] this seems very much simplified to the point it isn't really the subject at hand. seems to emphasize sequential turns, but ignores context switching in the midst of so that changes are taking place in unintended ways, which is what the producer consumer problem really is...

[mufcdanny]

At 6:30 you say only one thread can run at a time. Can we not have multiple threads simultaneously running inside a process?

[spacepod]

Could you explain why a many to one model won't allow multiple threads in parallel on multiprocessors but a one to one model does. I'm a bit confused by all this, thanks for your time

[penklislawnmowing]

I just want to go on learning about computers for the rest of my life and studying the History of numbers

[Thunterise]

Barbara, a previous comment you wrote I couldn't reply to so I'll do it like this:

"The CPU runs one process at a time. You can have multiple threads running inside of a process but only one thread will run at a time as well. Each CPU can only execute one command at a time. It's a serial processor. This is why there is CPU Scheduling algorithms. These algorithms make queues of the instruction sets so they can all run to accomplish the task of the process. Each instruction runs one by one in a serial way depending upon how the CPU schedule decides how to run them."

I feel like you are spreading misinformation here. You are completely ignoring multicore processors, which is really not something you can ignore.

Multiple threads can definitely run simultaneously - even from the same process - but only if you have multiple cores. The amount of cores you have in the processor is also the amount of threads that can run simultaneously.