you just explained nearly all of the content of a test I will take tomorrow in 15 minutes. Thank you!!! Awesome explanations, awesome teaching, you are VALUED.
One thing worth mentioning. The page table itself is stored in RAM, which means it takes way more than a single cycle to find out whether the page is on disk or not.
Thanks a lot David for the videos. I have signed up on your website and used the promo code that you provided to the comment below. Virtual memory makes sense to now. I am currently enrolled in a operating systems class, do you offer any courses that teach similar subject matter?
1.40. If the page we're evicting to the disk is dirty, we write it on the disk first. But aren't we already evicting it to the disk? Does that mean that the dirty page will not move, instead we take the changed version of the data from VA and write it to disk first, move it to RAM again, and finally evict it? I'm a bit confused about that.
Hi David, your videos are amazing, I love the way you explain the content. My profs could not compare. I am wondering if your course is still available? The link seems to be broken.
About the answer to the opening question: You said "we cannot have no entry in the page table." But the entry could point to an invalid location, i.e. not in RAM, not in the disk. (Say, RAM + disk < 4GB) As I understand, a page fault always implies that the requested page is on the disk. So "page fault" is a misnomer, then. It is not a "fault" that some VA map to the disk, the system was designed that way.
at about 4:10, why does it take 1000 cycles to change the mapping of just 1 page? Shouldn't it be just a simple write? Great lectures by the way! Much better than my CS professor!
questions: the size of virtual memory that we set in RHEl OS is the size of page table?? Is the size of page table fixed by OS architecture?? Can we modify the amount of Virtual memory set in the system, what would be the impact on PT?
really sir your lecture help me so much, sir i have one question i had read in so many places that "cpu generate logical address " how is it so can you explain it any any of your video if yes then please share link
no nvme has only bared halved sata ssd access times 7 years later. Here's a rough guide. Hard drive:~10 milliseconds good sata ssd:~100 microseconds good nvme ssd:~50 microseconds The only nvme ssds that are significantly better for this are optane ssds at ~10 microseconds and they also don't double in access time like other ssds the moment ur not doing purely a read or write. In other words in the real world, the ssds would have 200, 100 and 10 microsecond access times respectively.
Today this is the case for most users. Back in the day, around 1992-1995, you only had 4-8 MB of RAM and Windows 3.1 was crawling along. Windows NT, on which Windows 10 is based, didn't even run with so little RAM. The more RAM the computers got, the rarer the moments when the RAM became too scarce became. From around 8 GB of RAM onwards, this was practically no longer a big issue for the normal user. Of course, there is also software today that requires more than 8 GB of RAM; it is in the nature of things that the need continues to increase as the installed base of computers has more and more RAM. The gaming industry these days is oriented towards consoles. The current consoles like the PS5 have 16 GiB RAM, so 16 GiB RAM will be enough on the PC for a while. You can also leave a few other programs running in the background, so it doesn't hurt if you always have a little more RAM installed on your PC than the RAM on the consoles. You will only use your 32 GB of RAM with games from the next generation of consoles, unless you have your own areas of application where you need that much RAM. I have a few virtual machines running at the same time, so it helps to have a lot of RAM.
@@OpenGL4ever My first PC had 8MB of RAM and could run Windows 95. I have no idea how Windows could even run on 8MB of RAM, considering now even Notepad alone could use more than that.
@@typingcat Windows 95 is basically an improved 32/16 bit version of Windows 3.x. It's a 16 bit and 32 bit hybrid kernel. The version of Notepad in Windows 11 is a new version of Notepad. If I'm not mistaken, this was written in C# and .NET instead of the old version which was written in assembly language (Notepad for Windows NT was written in C). This alone increases the RAM requirement significantly. Since they are completely different versions, you cannot compare them. Additionally, the first Notepad could only read files no larger than 45 KiB. The limitation comes from the 16 bit real mode world it was once programmed for. A data segment can only be 64 KiB in size, if you want to use more, you have to do a lot of tricks in Real Mode. Notepad was written for Windows 1.0 which run in Real Mode only. Windows NT 3.1 required 4 MiB for the page tables alone. In addition, there was the memory requirement for the rest of the actual system, so it ran reasonably well from 12 MiB.
@@OpenGL4ever I was not talking about the new UWP Notepad, but those old one in Windows 10. Anyway, I remembered by reading your comments that in the Windows, if I tried to open a relatively large text file with Notepad, it said something like, "This file is too big, would you open it in WordPad instead?"
Random Access time of a typical hard drive where about 11 to 16 ms. Server hard drives that run at a speed of over 7200 rpm where a little bit faster and much louder. And to exchange two data blocks you will at least have to access the HD two times. And usually you had to exchange more than just 4 kiB of Data. As soon as the computer has to transfer data to hard drives and you want to access this data later, things get really slow. It therefore made more sense to close programs that you don't urgently need.
you just explained nearly all of the content of a test I will take tomorrow in 15 minutes. Thank you!!! Awesome explanations, awesome teaching, you are VALUED.
The only tutorial set on TH-cam that makes clear sense with clear explanations. Thanks a bunch 😊
YOu are just better than my OS professor at lecturing this!!! Thanks :))
That is an understatement.
These videos are golden. Thank you for getting me through college.
Explaining such complex concepts so simply is what every Prof should learn from you.
mind your own buissness
thank you so much, I've been looking for a content that explains VM and yours is just quintessential, thank you a million times
These videos are amazingly informative.The concepts are beautifully explained with the help of drawings.Thanks David Black for sharing the knowledge
Clear and concise explanation, I got more in these 6 minutes than I did from ploughing through the text. Thank you very much :).
One thing worth mentioning. The page table itself is stored in RAM, which means it takes way more than a single cycle to find out whether the page is on disk or not.
Finally , after years of confusions, I understood all this Virtual memory and paging stuff basics..................
Thank you. Very informative. A life saver. I was searching for something like this the whole day. Thanks Again. Cheers.
sanjith edward You're most welcome. If you're looking for more check out the full course at test.scalable-learning.com, enrollment key YRLRX-25436.
this is very clear. im having this lessons for my A\L examination. thanks a lot
Do us a favor and become a professor. You are incredibly good at breaking down things and explaining them :)
These really are an excellent resource. Many thanks!
Fast paced and super clear
Clear and precise explanation. Thanks for such videos.
I am learning a lot from your videos :D
Thanks a lot David for the videos. I have signed up on your website and used the promo code that you provided to the comment below. Virtual memory makes sense to now. I am currently enrolled in a operating systems class, do you offer any courses that teach similar subject matter?
+a101activist I'm afraid I only teach architecture and parallel programming at the moment.
Thanks David for your videos! It helps me a lot.
From 1:06, is it describing "swap" page between RAM and disk ?
Thanks for these very nice slides and presentation very valuable 🎉
1.40. If the page we're evicting to the disk is dirty, we write it on the disk first. But aren't we already evicting it to the disk? Does that mean that the dirty page will not move, instead we take the changed version of the data from VA and write it to disk first, move it to RAM again, and finally evict it? I'm a bit confused about that.
Hi David, your videos are amazing, I love the way you explain the content. My profs could not compare. I am wondering if your course is still available? The link seems to be broken.
It’s soooo good! Because it’s very very clear ! Thank you 🙏🏻 🙏🏻🙏🏻
Great lecture and very photographic learning , hope that to obtain more sophisticated resources from you
About the answer to the opening question: You said "we cannot have no entry in the page table." But the entry could point to an invalid location, i.e. not in RAM, not in the disk. (Say, RAM + disk < 4GB)
As I understand, a page fault always implies that the requested page is on the disk. So "page fault" is a misnomer, then. It is not a "fault" that some VA map to the disk, the system was designed that way.
at about 4:10, why does it take 1000 cycles to change the mapping of just 1 page? Shouldn't it be just a simple write?
Great lectures by the way! Much better than my CS professor!
cz OS is running on top of the processor. Bare processors will take only less than 100 cycles to write a 4kB page.
I like your video so much , clear english accent
you literally saved me. Thanks a lot!
Thank you so much for your explaination!
At which location of disk is the evicted page, how does the os knows from which location of the disk to fetch the os?
How do we know where to find the memory on disk?
Is there like a second translation table written somewhere...?
what software do you use ,to make the presentation
Looks like a regular presentation program. i.e Libreoffice Impress, Apple Keynote or Microsoft PowerPoint.
Fantastic lecture
good for sharing this exellent material.
Does page fault happen even if there is infinite amount of ram ?
great content still, thanks.
Oh my gosh. Thank you, thank you.
excellent explanation, thank you
Can processor use DMA and do other stuff while rewriting pages?
questions: the size of virtual memory that we set in RHEl OS is the size of page table?? Is the size of page table fixed by OS architecture?? Can we modify the amount of Virtual memory set in the system, what would be the impact on PT?
Do we have dedicated memory for Page tables or do they acquire some portion of RAM?
In Linux's Virtual Memory system, there is dedicated space allocated for the Kernel in the RAM. Page tables are stored in this dedicated space.
1:37 confuses me
really sir your lecture help me so much, sir i have one question i had read in so many places that "cpu generate logical address " how is it so can you explain it any any of your video if yes then please share link
Can you share these slides
really helpful .. Thank you
Brilliant
With the speed of NVME drives these days, page faults aren't that bad anymore
no nvme has only bared halved sata ssd access times 7 years later. Here's a rough guide.
Hard drive:~10 milliseconds
good sata ssd:~100 microseconds
good nvme ssd:~50 microseconds
The only nvme ssds that are significantly better for this are optane ssds at ~10 microseconds and they also don't double in access time like other ssds the moment ur not doing purely a read or write. In other words in the real world, the ssds would have 200, 100 and 10 microsecond access times respectively.
Ha! My PC has 32GB of RAM, and it probably is "too much RAM". I have never been able to use anywhere close to that amount.
Today this is the case for most users. Back in the day, around 1992-1995, you only had 4-8 MB of RAM and Windows 3.1 was crawling along. Windows NT, on which Windows 10 is based, didn't even run with so little RAM.
The more RAM the computers got, the rarer the moments when the RAM became too scarce became. From around 8 GB of RAM onwards, this was practically no longer a big issue for the normal user. Of course, there is also software today that requires more than 8 GB of RAM; it is in the nature of things that the need continues to increase as the installed base of computers has more and more RAM.
The gaming industry these days is oriented towards consoles. The current consoles like the PS5 have 16 GiB RAM, so 16 GiB RAM will be enough on the PC for a while. You can also leave a few other programs running in the background, so it doesn't hurt if you always have a little more RAM installed on your PC than the RAM on the consoles.
You will only use your 32 GB of RAM with games from the next generation of consoles, unless you have your own areas of application where you need that much RAM.
I have a few virtual machines running at the same time, so it helps to have a lot of RAM.
@@OpenGL4ever My first PC had 8MB of RAM and could run Windows 95. I have no idea how Windows could even run on 8MB of RAM, considering now even Notepad alone could use more than that.
@@typingcat Windows 95 is basically an improved 32/16 bit version of Windows 3.x.
It's a 16 bit and 32 bit hybrid kernel.
The version of Notepad in Windows 11 is a new version of Notepad. If I'm not mistaken, this was written in C# and .NET instead of the old version which was written in assembly language (Notepad for Windows NT was written in C). This alone increases the RAM requirement significantly.
Since they are completely different versions, you cannot compare them.
Additionally, the first Notepad could only read files no larger than 45 KiB. The limitation comes from the 16 bit real mode world it was once programmed for. A data segment can only be 64 KiB in size, if you want to use more, you have to do a lot of tricks in Real Mode. Notepad was written for Windows 1.0 which run in Real Mode only.
Windows NT 3.1 required 4 MiB for the page tables alone. In addition, there was the memory requirement for the rest of the actual system, so it ran reasonably well from 12 MiB.
@@OpenGL4ever I was not talking about the new UWP Notepad, but those old one in Windows 10. Anyway, I remembered by reading your comments that in the Windows, if I tried to open a relatively large text file with Notepad, it said something like, "This file is too big, would you open it in WordPad instead?"
A 3GHz CPU needs 0,0267 sec for 80 million cycles. A painful, long time...
Random Access time of a typical hard drive where about 11 to 16 ms. Server hard drives that run at a speed of over 7200 rpm where a little bit faster and much louder. And to exchange two data blocks you will at least have to access the HD two times. And usually you had to exchange more than just 4 kiB of Data. As soon as the computer has to transfer data to hard drives and you want to access this data later, things get really slow. It therefore made more sense to close programs that you don't urgently need.
Now I understand why the Apple products are fast!
i love you
it only slower than interaction with humans ....
iOS kills your programs X(
Ive seen 1tb of ram...