God of Concurency..i have bene flowing you since my 2014 when you used to write blog post only...by going through your post i attend interview like a LION when they ask mutithreading qns...Thanks a Lot form 10+ yrs exp guy from BLR,India.
Recently I am taking the Parallel and Concurrent Programming course. Your explanation is very clear and thank you for making the video to let me have the basic idea before the course.
The Java concurrency playlist is helping me get through college homework (thank you for that). They're well made and easy to understand. I do also appreciate that an experienced programmer like you decided to do this work basically for free.
@@JakobJenkov Hello, Jakob) Last section, you talking about "parallel, not concurrent" case. Question: if one big task is broken into tasks (in differents threads of course), but there is only one CPU, will it be concurrency and not parallelism?😁
First of all Thanks a ton for all these excellent elaborated contents, it is truly admirable. On the context of "Parallel not concurrent" I think we can say that multiple instances of an application could run in parallel however in each application there is only one thread.
I tried to understand this topic many times till now. This time I have confidence that I can explain others well, because I understood the concept. Thanks Jakob.
Great! :-) ... I have tried myself to have problems understanding a topic - until I found the "right" explanation - one that I understand, that is! I am happy to be able to help others get to that point too :-)
FINALLY! Every tutorial and course I've taken could not explain this the way you just did. Rule: "If you can't teach it to me like I'm 5, you don't know it either". You sir, clearly know your stuff. Thank you! Please tell me you have courses I can buy from you.
Thank you very much for your kind words! :-) At the moment I do not have any specific courses (or similar) that you can buy. In the future I will probably have some books etc. - but they will most likely be related to a very specific project I am working on. There will be lots of basic to advanced stuff to learn from those too, though. But - not yet :-) You will see when it comes out - because I will be making videos about this project too :-)
Hi Jakob. What do you exactly mean in CPU? Do you use CPU and CPU core interchangeably? Because as I know, computers usually come with one CPUs and multiple cores. So let's take Intel Core i7, is i7 a CPU or it has bunch of CPUs? Thanks.
In terms of multithreading just think multiple cores. I don't think it makes a huge difference concurrency-wise whether you have one CPU with 8 cores / 16 threads, or 8 CPUs. You have to deal with the same issues.
Jakob, you have explained well the difference between concurrent and parallel execution in terms of CPU processes. But is there any difference in terms of sharing data? Is it possible to create 2 processes running on the different CPUs sharing the same data like counter variable, and the process is to read the value and inc it. I guess in previous videos when you're talking about 2 Threads sharing data, these 2 processes were executing on the same CPU concurrently. Is it possible that those 2 processes could have been executing on the 2 different CPUs in parallel? How synchronized block manages threads(processes) sharing data but running on the different CPUs?
Yes, it is possible to share data between threads that run on different CPUs. And yes, volatile and synchronized keywords work the same in that situation. If you access a variable from within a synchronized block, only one thread can access it - regardless of how many CPUs your computer has running threads inside your application. Also, you are guaranteed that updates made by one thread to a variable from within a synchronized block is also visible to other threads - even if the threads run on different CPUs. The same is true about volatile variables.
Very well explained. In most real-world applications with multicore CPUs, it's a mixed concurrency and parallelism. So "parallelism" can be interpreted as a situation where you have a high percentage of parallel executions over concurrent executions. Vice versa. I think the concept of simultaneity of executions flow is important here.
Thank you. Yes, there is a lot of parallel execution going on in multicore CPUs. Some threads may run on the same CPU, other threads may run on different CPUs etc. Additionally, you have the serially parallel instruction pipelines of the CPUs too, that also provide some level of parallelism - although within a thread - not between threads.
I am not an expert in Node.js - but I think fork() runs a JavaScript in a new process - which might be able to run in parallel with the process spawning it. But - please double check the Node.js docs !!
would you say that parallelism is the same as parallel concurrent execution, except its concerned with a single task being broken up rather than different tasks?
Hi, Jacob! If i have 1 core(CPU )CPU and run 2 instances of the same application with only 1 thread (main) in each instance, does it mean parallel execution? Or Parallel execution of any application, process might be only on 2+ cpu? Sorry for possible obvious question.
Thanks a ton or the excellent video! Could you please help me understand in case of Parallelism splitting of a task(thread) done in the program(By Programmer, is so how?) or it is done by the compiler while generating the machine code or CPU while executing the instructions?
Splitting a job up into multiple smaller tasks that cen be processed in parallel must be done by the programmer. The compiler does not know enough about the tasks your code is performing to know how to do that, in most cases. In some cases, if you give some hints, perhaps it can be done more automatically - and sometimes at runtime by e.g. TornadoVM or other tools. But the developer is at this point in time still required to at least hint at how to break up the task.
@@JakobJenkov What are the ways in programming through which a task (Thread) can be split into multiple subtasks? If you can give any hint in any language.
@@JakobJenkov Yes, last time we talked on LinkedIn as you don't like social networking much facebook or whatsapp :-) I think study pattern is changing now.People are more interested in video content as video maker can express more as compared to blog and Listener can also play with any speed he/she likes.Earlier I used to keep 1.x but now I am comfortable with 1.5 or sometimes 2.0 and this is great experience. Hope you will make more content whenever you have time specially on distributed architecture and whatever you have done in last couple of years.
Thank you! I may get around to making videos about the classes in java.util.concurrent soon. Until then, I have textural tutorials about them here: tutorials.jenkov.com/java-util-concurrent/index.html
Hi jakob, can you please create any course on how to do real time project development for experienced Java developer for production ready project. It should have source code for real time Java project. Thanks for this video.
Hi Rajan, I will eventually make some videos about Realtime Architectures - but it's a big topic with lots of details, nuances and tradeoffs, so I don't think I can just make a "ready-to-go" template you can just start from. But I can cover various topics that can help you design your own realtime architecture, or at least take steps towards a realtime architecture.
Does anyone know if any of these three backend languages have both concurrency and parallelism? Node/Express Django Go I want to become a multi-thread API dev. I don't care about what is 'popular' now. I just want what has the best foundation for concurrency and parallelism.
I don't know myself, actually. I vaguely remember something about Node having support for multiple threads that can run on multiple CPU cores, but I don't know how it works internally.
Not exactly. Async in Java still runs within a single thread, as far as I know. Concurrency in Java could be achieved using multiple threads - yet still running on the same CPU (or different CPUs).
@@JakobJenkov First of all, please excuse my lack of knowledge as I'm still learning. I believe it matters because it sounds odd to say: Imagine we have a computer with 2/multiple "CPUs" 2:10 3:19 instead of the commonly known "CPU Cores" with 2 Threads each. I first thought this might be possible with dual CPU servers, which you didn't mention, until you mentioned "8 CPUs" 6:37 . I'm only pointing this out because I got confused and wanted to get more clarification.
Thanks for the nice explanation . I have got a question here . If we have two threads and two CPUs. CPU-1 is executing thread-1 and CPU-2 is executing thread-2. Hence we have parallelism here. Now if at some point of time one of the threads takes the monitor of a lock and halts the execution of other thread until the monitor is released and acquired by other thread, in that case which one is true : 1. Thread-1 and thread-2 are running concurrently 2. Thread-1 and thread-2 were running in parallel , but at that point of time when monitor lock was acquired they ran concurrently . 3. Thread-1 and thread-2 are running in parallel
If Thread 1 takes a monitor object, and Thread 2 does NOT try to take that monitor at the same time, Thread 1 and Thread 2 are still running in parallel. If Thread 2 DOES try to take the monitor object while Thread 1 holds it, Thread 2 will be blocked until Thread 1 releases the monitor object - in which case they are not running in parallel while Thread 2 is waiting.
You tried so hard to be clear but you failed and even in the comments you are sticking to it. People are asking, do you mean CPU or CPU core. It matters!!! Stop saying it doesn't. The average person does NOT have a multi CPU machine. So it matters. By not CLEARLY defining what you mean, it causes more confusion, especially to those trying to learn the VERY difficult concept. Say what you mean, say CPU Core. It means something.
Allright - I will say CPU core in the future... I don't personally think it makes a big difference, except a bit about the internal design with caches etc. But I can refer to the cores as "cores" ... We are also getting a GPU API in Java in the future, and I guess we also just call it "one GPU with many cores".
@@JakobJenkov I appreciate your response. I really liked your video but I did have to look at a couple more to fully understand that you meant CPU core. Thank you for taking the time and being willing to adapt. As a consumer of content, I can only appreciate those who are willing to accept feedback and adapt. Thank you for your response.
Taking off your time to tech other and make it totally free, is a virtue to admire to.
Thank you for your awesome tutorials and your time. :)
You are welcome! :-)
God of Concurency..i have bene flowing you since my 2014 when you used to write blog post only...by going through your post i attend interview like a LION when they ask mutithreading qns...Thanks a Lot form 10+ yrs exp guy from BLR,India.
Thanks :-)
I really liked the graphics on this, as a visual learner it makes it easier to follow.
Great !! 😊
Recently I am taking the Parallel and Concurrent Programming course. Your explanation is very clear and thank you for making the video to let me have the basic idea before the course.
You are welcome! I am glad my video helped you! :-) ... I have a whole playlist of 21 videos in this topic - in case you are interested ;-)
The Java concurrency playlist is helping me get through college homework (thank you for that). They're well made and easy to understand. I do also appreciate that an experienced programmer like you decided to do this work basically for free.
I am glad you like them! :-) ... that's why I made them :-)
@@JakobJenkov
Hello, Jakob)
Last section, you talking about "parallel, not concurrent" case. Question: if one big task is broken into tasks (in differents threads of course), but there is only one CPU, will it be concurrency and not parallelism?😁
First of all Thanks a ton for all these excellent elaborated contents, it is truly admirable. On the context of "Parallel not concurrent" I think we can say that multiple instances of an application could run in parallel however in each application there is only one thread.
True. Multiple processes could run in parallel - even if each process only has one thread running internally.
I will come back and watch this once I am done with office work. Thanks for all the good work. Cheers
You are welcome! :-) ... I'll hold a warm copy of this video for you for when you return ;-)
@@JakobJenkov Thanks. The world has turned cold and drab in most places. Little warmth is always appreciated :)
I tried to understand this topic many times till now. This time I have confidence that I can explain others well, because I understood the concept. Thanks Jakob.
Great! :-) ... I have tried myself to have problems understanding a topic - until I found the "right" explanation - one that I understand, that is! I am happy to be able to help others get to that point too :-)
FINALLY! Every tutorial and course I've taken could not explain this the way you just did. Rule: "If you can't teach it to me like I'm 5, you don't know it either". You sir, clearly know your stuff. Thank you! Please tell me you have courses I can buy from you.
Thank you very much for your kind words! :-) At the moment I do not have any specific courses (or similar) that you can buy. In the future I will probably have some books etc. - but they will most likely be related to a very specific project I am working on. There will be lots of basic to advanced stuff to learn from those too, though. But - not yet :-) You will see when it comes out - because I will be making videos about this project too :-)
It's so satisfying to watch the tutorials in 4k
Glad you think so ;-)
Jakob, Thanks for your time, effort and a great way of teaching other's!
You are welcome !! 😊😊
Hi Jakob.
What do you exactly mean in CPU? Do you use CPU and CPU core interchangeably? Because as I know, computers usually come with one CPUs and multiple cores.
So let's take Intel Core i7, is i7 a CPU or it has bunch of CPUs?
Thanks.
In terms of multithreading just think multiple cores. I don't think it makes a huge difference concurrency-wise whether you have one CPU with 8 cores / 16 threads, or 8 CPUs. You have to deal with the same issues.
Best concurrency tutorial online!
Thank you! :-)
Mr. Jakob, your explanation is very good and I understand much better now for my Golang learning. Thank you
Great ! :-)
@Jakob, do you have git repository for all these tutorials of threads ? the examples which you are explaining in the videos.
I have some examples in my Java Examples GitHub repo (link below) - but not yet so many:
github.com/jjenkov/java-examples
Well explained Jakob. Thanks for explaining the complex concept in a simple way...
You are most welcome ! 😊
Literally 0 dislikes speaks for itself... You're a god!
Thanks :-D
Jakob, you have explained well the difference between concurrent and parallel execution in terms of CPU processes.
But is there any difference in terms of sharing data?
Is it possible to create 2 processes running on the different CPUs sharing the same data like counter variable, and the process is to read the value and inc it.
I guess in previous videos when you're talking about 2 Threads sharing data, these 2 processes were executing on the same CPU concurrently. Is it possible that those 2 processes could have been executing on the 2 different CPUs in parallel?
How synchronized block manages threads(processes) sharing data but running on the different CPUs?
Yes, it is possible to share data between threads that run on different CPUs. And yes, volatile and synchronized keywords work the same in that situation. If you access a variable from within a synchronized block, only one thread can access it - regardless of how many CPUs your computer has running threads inside your application. Also, you are guaranteed that updates made by one thread to a variable from within a synchronized block is also visible to other threads - even if the threads run on different CPUs. The same is true about volatile variables.
@@JakobJenkov Thanks! It's clear now)
Very well explained.
In most real-world applications with multicore CPUs, it's a mixed concurrency and parallelism.
So "parallelism" can be interpreted as a situation where you have a high percentage of parallel executions over concurrent executions. Vice versa.
I think the concept of simultaneity of executions flow is important here.
Thank you. Yes, there is a lot of parallel execution going on in multicore CPUs. Some threads may run on the same CPU, other threads may run on different CPUs etc. Additionally, you have the serially parallel instruction pipelines of the CPUs too, that also provide some level of parallelism - although within a thread - not between threads.
Has Parallelism to do anything with fork() specifically for Node.js ?
I am not an expert in Node.js - but I think fork() runs a JavaScript in a new process - which might be able to run in parallel with the process spawning it. But - please double check the Node.js docs !!
Great work, thanks for helping on Java.
Thanks - and you are welcome! :-)
would you say that parallelism is the same as parallel concurrent execution, except its concerned with a single task being broken up rather than different tasks?
That is my understanding of the official distinction, yes. Approximately.
Hi, Jacob! If i have 1 core(CPU )CPU and run 2 instances of the same application with only 1 thread (main) in each instance, does it mean parallel execution? Or Parallel execution of any application, process might be only on 2+ cpu? Sorry for possible obvious question.
Parallel execution would require more than one CPU. Otherwise it is just concurrent execution
Or at least more than one CPU core...
Marvelous explanation, thank you!
You are welcome :-)
"daddy, can i get a bitcoin pleasee?" idk why, I just thought that when I saw your name lolll
Nice explanation! Please make more videos about latest concurrency api introduced in Java.
I will try :-)
Thank you for this video. It was very well explained.
Thanks :-)
Thanks a ton or the excellent video!
Could you please help me understand in case of Parallelism splitting of a task(thread) done in the program(By Programmer, is so how?) or it is done by the compiler while generating the machine code or CPU while executing the instructions?
Splitting a job up into multiple smaller tasks that cen be processed in parallel must be done by the programmer. The compiler does not know enough about the tasks your code is performing to know how to do that, in most cases. In some cases, if you give some hints, perhaps it can be done more automatically - and sometimes at runtime by e.g. TornadoVM or other tools. But the developer is at this point in time still required to at least hint at how to break up the task.
@@JakobJenkov What are the ways in programming through which a task (Thread) can be split into multiple subtasks? If you can give any hint in any language.
great explanation for all. thank you.
Thank you - and you are welcome :-)
nicely explained.
simple & neat
Thank you ! :-)
Thanks for making video on this topic...enjoyed more than the blog :-)
Thanks! Glad you like this one Manoj! Long time since we talked the last time, right? Usually on LinkedIn... :-)
@@JakobJenkov Yes, last time we talked on LinkedIn as you don't like social networking much facebook or whatsapp :-) I think study pattern is changing now.People are more interested in video content as video maker can express more as compared to blog and Listener can also play with any speed he/she likes.Earlier I used to keep 1.x but now I am comfortable with 1.5 or sometimes 2.0 and this is great experience. Hope you will make more content whenever you have time specially on distributed architecture and whatever you have done in last couple of years.
clear and concise explanation. Thankyou. Please do make a video on demonstration of java concurrent collections.
Thank you! I may get around to making videos about the classes in java.util.concurrent soon. Until then, I have textural tutorials about them here:
tutorials.jenkov.com/java-util-concurrent/index.html
@@JakobJenkov Thankyou
Awesome explanation
Thanks ! :-)
Thank you for making it very clear. Great job!
Thanks - and you are welcome! :-)
thank you , i undersood this concept now.
You are welcome :-)
Hi jakob, can you please create any course on how to do real time project development for experienced Java developer for production ready project. It should have source code for real time Java project.
Thanks for this video.
Hi Rajan, I will eventually make some videos about Realtime Architectures - but it's a big topic with lots of details, nuances and tradeoffs, so I don't think I can just make a "ready-to-go" template you can just start from. But I can cover various topics that can help you design your own realtime architecture, or at least take steps towards a realtime architecture.
Thanks, awesome explanation. A video Idea - Concurreny vs Parallelism vs Asynchrony.
Java does not have async await features - if that is what you are referring to.
fantabulous, thanks for the great share
You are welcome! :-)
Thats realy good explanation
Thank you! :-)
Thank you for the clear explanation !!!!
You are welcome!! :-)
beautiful graphics. ty so much for the explanation!
You are welcome ! ... Hope it was easy to understand! :-)
That's awsome Jakob!
Thank you !! 😊
Thanks SIR !! for the Great effort !!!
You are most welcome! :-)
nice explanation
Thanks :-)
Awesome explanation! Thanks
You are welcome! :-) Glad to hear its useful!
Excellent! So clear. Thank you.
You are welcome! Glad it was easy to understand :-)
Good video! Only thing I think is that "CPU" could say "Core" in the example images.
True... but I decided to keep it simple :-)
I am having so much fun right now
Great! :-D
Nice Explanation
Thanks :-)
Well explained! Thank you!
Thanks, and you are welcome 😊
this video is awesome
Thank you very much !! :-)
Thank you very much Jakob.
You are welcome! :-)
Does anyone know if any of these three backend languages have both concurrency and parallelism?
Node/Express
Django
Go
I want to become a multi-thread API dev. I don't care about what is 'popular' now. I just want what has the best foundation for concurrency and parallelism.
I don't know myself, actually. I vaguely remember something about Node having support for multiple threads that can run on multiple CPU cores, but I don't know how it works internally.
thank you! Love this video! Take care.
You are welcome, and thank you too !! 😊
Thanks for clearing this up for me 😀
You are welcome! :-)
Concurrency looks like async JavaScript. Did I miss some key difference here?
Not exactly. Async in Java still runs within a single thread, as far as I know. Concurrency in Java could be achieved using multiple threads - yet still running on the same CPU (or different CPUs).
@@JakobJenkov I see. Thank you.
Good video. Very clear content. :)
Thank you very much!
Damn, nice video!
Thank you 😊
Thanks a million
You are welcome! :-)
omg, this is so clear
Thanks! Glad it was understandable :-)
When you say CPUs do you mean CPU cores?
Does it matter?
@@JakobJenkov
First of all, please excuse my lack of knowledge as I'm still learning.
I believe it matters because it sounds odd to say: Imagine we have a computer with 2/multiple "CPUs" 2:10 3:19 instead of the commonly known "CPU Cores" with 2 Threads each.
I first thought this might be possible with dual CPU servers, which you didn't mention, until you mentioned "8 CPUs" 6:37 .
I'm only pointing this out because I got confused and wanted to get more clarification.
That was great....
Thanks! :-)
Thanks for the nice explanation . I have got a question here .
If we have two threads and two CPUs. CPU-1 is executing thread-1 and CPU-2 is executing thread-2. Hence we have parallelism here. Now if at some point of time one of the threads takes the monitor of a lock and halts the execution of other thread until the monitor is released and acquired by other thread, in that case which one is true :
1. Thread-1 and thread-2 are running concurrently
2. Thread-1 and thread-2 were running in parallel , but at that point of time when monitor lock was acquired they ran concurrently .
3. Thread-1 and thread-2 are running in parallel
If Thread 1 takes a monitor object, and Thread 2 does NOT try to take that monitor at the same time, Thread 1 and Thread 2 are still running in parallel.
If Thread 2 DOES try to take the monitor object while Thread 1 holds it, Thread 2 will be blocked until Thread 1 releases the monitor object - in which case they are not running in parallel while Thread 2 is waiting.
@@JakobJenkov Thank you very much Jacob for your reply . Very clear and concise .
You tried so hard to be clear but you failed and even in the comments you are sticking to it. People are asking, do you mean CPU or CPU core. It matters!!! Stop saying it doesn't. The average person does NOT have a multi CPU machine. So it matters. By not CLEARLY defining what you mean, it causes more confusion, especially to those trying to learn the VERY difficult concept. Say what you mean, say CPU Core. It means something.
Allright - I will say CPU core in the future... I don't personally think it makes a big difference, except a bit about the internal design with caches etc. But I can refer to the cores as "cores" ...
We are also getting a GPU API in Java in the future, and I guess we also just call it "one GPU with many cores".
@@JakobJenkov I appreciate your response. I really liked your video but I did have to look at a couple more to fully understand that you meant CPU core. Thank you for taking the time and being willing to adapt. As a consumer of content, I can only appreciate those who are willing to accept feedback and adapt. Thank you for your response.