How many kernel system calls do runtimes make?

[@hnasr]

Learn more about how the kernel (Windows and Linux) , get the Fundamentals of Operating Systems oscourse.win

[@shapelessed]

So, the video was published 5 hours ago but uploaded at least 4 days earlier… Interesting…

[@juangomez4133]

Hussein in a few years. "I want to discuss the electric flow in the motherboard when you do network request"

[@jskr456]

looking forward for it , so finally i can understand how network connections are made and data flows between them

[@thisisneeraj7133]

More looking forward for how silicon quality makes software slow

[@greyshopleskin2315]

@@jskr456there’s little people of the network card shouting through the network cable. And in the receiving computer there’s little people listening. Don’t let big network deceive you

[@Brahvim]

@@thisisneeraj7133 _Intel's 13th generation:_

[@ViplavKatiyar87]

😂😂

[@MarekKnapek]

The write is about writing to the screen. Under the hood it writes to standard output. Under the hood it writes to a file descriptor. Under the hood the file descriptor describes a terminal "device".

[@efkastner]

Came here to say just that :)

[@strange-m6u]

Go is around 185. Node v20.17 is around 695.

[@suvodipmondal1858]

I was wandering for this

[@matveyshishov]

The year was 2024. The Internet discovered system calls. Next season: containers are just processes; blinking an LED with TypeScript; .eh_frame section.

[@casperx102]

c do no bullshit, straight to the point

[@me_12-vw1vi]

why is it always an anime pfp? what’s with C and anime pfp’s??

[@ikhlasulkamal5245]

@@me_12-vw1vi It is the law!!!

[@undergrounder]

@@me_12-vw1vi Ying and Yang

[@shikanokonokokoshitantan]

​@@me_12-vw1vi Cultured programmers

[@shikanokonokokoshitantan]

Well even with assembly you need to use syscalls so youd need bare metal to actually get "straight to the point", if youre patient enough you can use buttons and switches to write to the display...

[@theairaccumulator7144]

Bruh ES6 modules are almost 10 years old at this point. They came out in 2015.

[@thydevdom]

Man I’ve been following you and your podcast for a few years. You are heaven sent and a great teacher.

[@KangJangkrik]

Since we have heated debate between C vs Rust in Linux community, please also try to test Rust with this. Thank you!

[@omarbarra3456]

That is what I was thinking about!!!

[@belkocik]

Indeed!

[@cybergaz]

rust makes twice as syscalls as C

[@Serizon_]

@@cybergaz so 47 * 2 = 94 , golang makes 135 according to some other commenter Really nice that golang and rust are so close. what about zig though . some people say that its even more bloat free than C

[@cybergaz]

@@Serizon_ yes it makes half syscalls as compared to C

[@azhar_waheed]

.mjs is MICHAEL JACKSON JAVASCRIPT FILE 😂😂😂😂

[@MrAtomUniverse]

Where is rust

[@ryanseipp6944]

14:37 The write syscall is also used to print stdout

[@Sonyim414]

Specifically, each line is its own write syscall. First is "File contents:", next is the buffer, next is the empty string after newline. This is really not an equal comparison, because in python he used an iterator to iterate over the lines...

[@erisboxxx]

Include Go in future comparisons if possible please!

[@avishkaarthik1542]

This is very nice ...probably the trade off between readability and performance and efficiency. Also a little in depth so on every system call there is an interrupt triggers and on every interrupt trigger there is a corresponding "trap handler" which is again maintained by os where there is a system call table which maps calls to its trap handler. It quiet fascinating cause in the end there is a stack frame created for these calls also just like every other function

[@btimbyindy]

To your point re: batching calls. Look at sendfile. It takes two file descriptors and moves data from one to another. When switching from userkernel land memory buffers are copied (userspace cannot access kernel space) and that gets very expensive when sending large amounts of data.

[@7heMech]

So happy you jumped on Bun as well.

[@7heMech]

For Bun maybe you should've used Bun.file API, also AI can't write Bun (except for the one on their website)

[@sharmojj]

ai can not do shit

[@theairaccumulator7144]

It would've been the same

[@MarekKnapek]

At the end you described iouring, but it cannot be used everywhere, because sometimes you need to know result of one syscall in order to decide whether or which syscall to perform next. If you bring this to the extreme, everything will be running in kennel space. Enter eBPF.

[@btimbyindy]

To address that you would need a simple embedded vm or language in the kernel, like how redis uses lua. But that could be a security nightmare.

[@bart2019]

This seems to be mainly just the overhead of the interpreter. I'm more interested in the difference for bigger program, not just this simple "hello world" level program.

[@saurav1096]

python and js code be like: can you do this task? no but i know guy who knows a guy and he knows a guy and .... knows a guy name kernel , he can

[@mykolamorozov2099]

true :D

[@dhillaz]

It would be interesting to do control runs executing empty source files to adjust for the Python/JS initialization, and deduce the call count of the read action specifically

[@anamulakash2201]

Awsome explanation. I was exploring ebpf and got notification from your chanel

[@CjqNslXUcM]

This is a dumb comparison. No one starts a seperate runtime for every file they open. You should open the runtime before measuring.

[@sashirkl]

These are like small chunks . How someone become good at it. We need a good mentor . And detailed videos

[@rschmidtzalles]

This was a fun watch. Thanks

[@wlockuz4467]

This is awesome. I wish you did more languages.

[@danishmehmood6110]

thanks for thinking out loud , and letting everyone know that dumb questions and the best ones and they really fill in the loop holes in your understanding , it takes a lot of courage to do that to answer the question you posited at the end of the video - I don't think that would be possible lets assume you are reading a 10 MB file and printing its contents to the screen - lets say you create a 1 mb read buffer , now you read the file and print , to read the whole file , you would need to make the 10 read syscalls (keeping things simple here for understanding ) - now if we employ your idea of making all the calls at once , that would defeat the whole purpose of having a memory buffer for read - if we make all the read calls at once where would OS write the data it has read , because the buffer is only 1 mb and the point of having a buffer is to constrain memory usage.

[@Winnetou17]

You raise a good point, but maybe there can be situations where it does make sense to batch and that can be known to be safe to do. Like, to give a trivial example - only batch different syscalls. Actually, no, just realized, your example is bad. You can't have multiple reads in a batch, since there's user code running between the reads. The batching should only be done by syscalls that happen immediately one after another.

[@alkolaqi83]

Whats the point of doing bun, should’ve done go/rust

[@engine_man]

Sorry… why would he do rust/go? Are you just regurgitating whatever is trending?

[@alkolaqi83]

@@engine_man exactly

[@ayoubay8842]

printing is writing to stdout, hence the write syscall

[@7heMech]

Awesome video, I've been thinking about this.

[@shpetimselaci5884]

Thank you for doing this video. Keep it up!

[@hitmusicworldwide]

Speaking of chat GPT, a lot of times when I'm out jogging in between listening to your KZbin videos and other videos I will have conversations with the chat GPT voice mode about these sorts of backend and front and integration issues and other technology ideas I have in the same or simular domains. It would be interesting for you to record a voice conversation with your musings about this with chat GPT or PI AI to see how the conversation goes， then follow up on it to find out if it's expressing something useful or not。 You would get immediate feedback from what might be an expert source if you prompt it correctly， and it would also give us the ability to see if it's able to give useful information the way you do when we're not able to have a direct conversation with you.

[@anon_y_mousse]

Huh. 430K subscribers, 12K views, yet not even 1K upvotes? Only thing I can figure is that maybe you need to tighten up the editing. If this were half the length I bet you'd get more upvotes. The comment right under your pinned comment explains the call to write(). I guess I'll say it too, that writing to the terminal window which is linked to the stdout of the process is where that comes from, because on Linux everything is a file, even the output to the terminal.

[@_Xnap]

For write calls you are all forgetting kernel needs to update access time in metadata of a file. You can check that via stat command. It will get updated as you read the file and that would count as write.

[@_Xnap]

Also due to this, many times to improve performance you would mount device with noatime and nodiratime, so accesing files does not result in writes.

[@andreffrosa]

That doesn't make much sense. Why isn't that part of the read call itself and instead a separate operation made by the process?

[@_Xnap]

en.m.wikipedia.org/wiki/Stat_(system_call)#Criticism_of_atime

[@_Xnap]

We are talking about kernel calls here that happen when you access file and kernel is the one updating access time. I think nowadays there is a bit difference here and not every read of file results in an update to access time...

[@ImadRoshan-p1d]

Hi Hussein Nasser, i hope you're doing great. Do you any proxies that can bridge MQTT and gRPC protocols?

[@energy-tunes]

Hes the type of prof that could lecture for 10 hours without break

[@siya.abc123]

I can sense it, another dev tool them all episode is around the corner

[@ryanseipp6944]

With synchronous syscalls, I don't think it's possible to "batch" them. The kernel wouldn't have the arguments. If you objdump the C code, you'll see before each `syscall` instruction, the code places values in specific registers. With io_uring, this batching is straightforward. You can submit 3 SQEs and wait for all 3 to complete. In that time, your program yields to the kernel and the kernel only resumes that thread when everything is ready. Then you can loop through CQEs and get the results.

[@dmitrykim3096]

Well it reads the script first parses it and them runs it, so ofc there will be an overhead

[@passionforsciencel5180]

Even working with memory need a syscall like malloc function , you supposed to mention that in the beginning, but thanks for the video anyway. Follow you from Algeria 🇩🇿

[@mobod6822]

malloc isn't a syscall, it's calling the mmap syscall

[@passionforsciencel5180]

@@mobod6822 Oh , yeh , thanks ❤️

[@rsroyalrahul5]

can you please do Go, Rust, Zig, Java please?

[@sashirkl]

Why not u publish some long project based videos

[@undrash]

Oh, S-Trace. I thought that was your zesty way of saying amazon S3 💀

[@محمد-م5ث1ش]

I hope you can include Go in your future comparison please

[@abhinavpandey1374]

shouldn't we also take into account the system calls that are done anyways without reading files? i feel a lot of those system calls are common for all node processes. also some of those reads are due to the script itself, as node and python also have to read the script passed to them. i feel comparing them to a compiled language like c is a little unfair. a more fair comparison would have been go vs c.

[@Serizon_]

go is 135 I think

[@algorithm1313]

When did you move to vim!???

[@bean_TM]

the javascript code could have been much better. especially with such an ancient version of node. i think it misrepresented how bad JS is. JS is bad. but not that bad (copium)

[@away7172]

only-interpreted languages can't be described as programming languages IMO

[@oll236]

Omg update node. Can someone get the stats?

[@shapelessed]

I think this “benchmark” is quite frankly flawed. Not because it doesn’t show anything, it’s because it shows things in the wrong light. What you have showcased here is not the objective speed of execution of these runtimes and the number of system calls they make, but the behaviour they show when starting the environment, reading configuration, writing startup logs and all the things they do right on start, which they then almost completely stop doing. The program you wrote in C? - It’s compiled, it has no need to set things up as it’s been done in advance by the compiler. So many people structure their benchmarks in a way that is objectively flawed simply because interpreted languages require some startup overhead put in place by the runtime that simply isn’t there with precompiled programs. If we were to play a fair game, we would need to benchmark the compiler doing its own things together with the compiled program against a script running inside of a runtime.

[@TheOnlyJura]

lol, no

[@shapelessed]

@@TheOnlyJura Sure, a guy comes with an actual argument to which some random on the internet says "lol no" and leaves. Either provide an argument or leave jerk off.

[@shapelessed]

@@TheOnlyJura lol, not an arument

[@Winnetou17]

Yeah, others mentioned this too, and I was curious how much of those extra syscalls were simply initialization stuff that only happens once. Because usually you don't simply run scripts like this. And very often.

[@yashnikam9189]

I think nodejs python bun require exec to run the file so it might have more syscall but c is compiled in exec and directly executed so it have less syscall.i might be wrong though just my assumption

[@bart2019]

This seems to be mainly just the overhead of the interpreter. I'm more interested in the difference for bigger program, not just this simple "hello world" level program.

[@RoelCagantas]

He sounds like Gru.

[@shubhambaranwal7613]

I think you are not giving node bun and pthon a fair chance. Sure the c program is fast but if you think carefully we also need to make trace of when we are making the .out executable filein c. We are executing that file but to execute that file we need to first make it and that will surely take more syscalls. Just a observation to point out

[@Monawwar]

Did anyone run for rust?

[@dasten123]

It's interesting stuff, but I this was a very unfair comparison because the programs did completely different things and I really don't get why you are using an ancient Node version

[@shapelessed]

Watch out mistaking a library for a framework, let alone runtimes in the JS community… You’re either gonna start a third world war or end up in a mysterious us accident, really dangerous thing to get wrong…

[@kaushik.aryan04]

I though bun was supposed to be faster than node

[@andreilucasgoncalves1416]

He is testing an older version of node, the current can be even slower. The reason can be many things like security updates or support to new features

[@azhar_waheed]

3:15 "runtime I am sorry" 🤣🤣🤣

[@developersharif]

😮

[@oleholgerson3416]

The real (and unsurprising) takeaway for me was how ugly javascript is… Can‘t even read from a file without wrapping everything in a lambda, just so that you can await something. JS really gives me nightmares

[@dasten123]

No you don't have to write it that way

[@btimbyindy]

There are sync versions of those calls.

[@derockx]

the lambda wasn’t even needed as node has had top-level await support for years. Furthermore, await isn’t even needed either, could’ve done readFileSync instead of async.

[@shafiq_ramli]

So is bun a scam?

[@andreilucasgoncalves1416]

No, there are two things to consider here, it is not a full application and the node version is too old

[@fourcoding3198]

​@@andreilucasgoncalves1416 Bun is standalone, i.e. it is independent from the installed node

[@btimbyindy]

Also, bun may have more startup overhead even if the actual script execution is faster.

[@vishnuvarthan4028]

Awesome content as usual

[@nitsanbh]

14:40 isn’t that write just the printf?

[@Winnetou17]

Yup, others said it as well. It's also potentially setting the atime (accessed time) on the read file.

[@danieledll]

Sorry but this is video has way too many problems ;) (1) The number of syscalls impacts the performances, a context switch between two threads of the same process or a switch to kernel space and back can take up to 5 microseconds, so if you are reading 1 byte at time (silly example eh? sadly not too much :( ) from memory and do absolutely nothing else at maximum you can read 200000 bytes (probably its less but let's pretend it's a perfect world and you don't pay the price of the memory copies and various checks). Even if you pin the thread, you would get down to about 2us and this would mean just 1000000 bytes per second ... New kernel components like io_uring implement VERY complex mechanisms to avoid the need of context switching forth and back from the kernel! (2) The version of nodejs in use is fairly old, beginning of 2022, with the 20.15 (which is the LTS not the latests so there might be even more performance improvements) it's down to below 700 (3) Not clear which is the version of python (4) This is minor but the C code does 1 single printf when instead the nodejs and the python version do 1 print per line, this is minor because the file contains just 2 lines NOTE: I am not a fan of node.js at all lol I like facts ;) EDIT: Also would have made sense to have a comparison with some code that was actually doing something because perhaps a lot of these syscalls are done during the initialization phase and the number of them might be much lower while the code runs. It would have been an important and useful comparison.

[@ghassenlabidi5171]

This is what I was thinking too! Plus Bun provides its own way to read files which is way more efficient than using node's "fs".

[@danieledll]

@@ghassenlabidi5171 to be honest o am not sure of the specifics, I am more highlighting the approach to this specific test / video. I understand that the original. Idea was to highlight all the extra work that these platform do for little reason but it's also very much true that they are built for backend applications and/or long running applications in mind and therefore they are little optimized when it comes to the startup

[@btimbyindy]

To exclude initialization, you could start and end with a specific syscall and use awk to select just the relevant ones between, then shell magic to summarize.

[@gauravkumarsharma5904]

insightful

[@SRG-Learn-Code]

What I get is that we need to embed html and js into syscalls. Lol, but seriously. I big part for me to using web technologies is making visual interfaces that works in most systems, in that sense, I need a runtime which might be to two runtimes if also I want to have logic in bun/node/deno or even python with a web interface. I don't want to start talking about electron because holy shit, but then are thing like tauri that uses the os webview whichever is it. Now... How would be a good approach to make logic with C for example, but having a lean web interface?

[@MCorange9]

Clearly this guy is just a sell out and barely knows how libc, kernel, or even just python works. Like 80% of those calls are just loading the python standart library. I dont know what you expected, since python is an interpreted language and needs to read every single file thats imported every time its ran.

[@000TheMatheus000]

i feel like there is a hierarchy for this. system calls are bad, but, allocating memory is even worse, and making network requests is even worse still. so probably in the whole scheme of things, system calls is the last thing people thing about. but is very interesting.

[@samirallahverdi4948]

Wow.. that's great exercise. Is there any way to see what these calls are actually?

[@nahuellescano]

Great video! Next time try to do the same in rust and zig xD

[@user-zo9ye6de5e]

you forgot java 😏😏

[@lucasoliveira-xs5yh]

I'm curious (and maybe I'll try it) about Go would behave

[@Winnetou17]

Somebody else here in the comments said that Go is around 185. And node v20.17 is around 695.

[@nathanpotter1334]

11:47 - 11:53 😆😆

[@williandamascenomusic]

WTF, your videos are so fucking interesting, well done!

[@AHMED-et7md]

what about using node last version using V8 maybe low time more then C.

[@vladimirkashitsyn1871]

Brilliant! I am going to check java program)))

[@mintoo2cool]

should have compared cat command too...

[@Aditya_Vyas]

Hussein is basically saying we need to avoid those kernel-user switching again & again, instead do something like an UPSERT or an Asynchronous call you make in JS. Hussein I thought a little bit, I had a question. Wouldn't an Asynchronous call take more time? You will surely have less system call but execution speed wouldn't be less? And won't it create a problem if the 2nd system call is dependent on the 1st one so it went, came back, bought some metadata(read write) and the 2nd call uses it with some of the users input? I'm probably mad or I should do some breathing exercises.

[@mage3690]

You could use command buffers and queues, the way Vulkan does. And IDK anything about how any of this works, but if Vulkan does it, I'm inclined to think it's fast.

[@Aditya_Vyas]

@@mage3690 interesting

[@seydullanarkulyyev2729]

🔥🔥🔥

[@amadzarak7746]

Hussein, your channel is one of the most valuable channels for any modern developer