 details for or at least these practicalities are quite clear if of course they're not clear for any of you you can write it on akmd and there are other channels that we provided to you so you there are zoom streams that some people are have joined right now so i guess we can just start with with what we mean with better scientific company do you want to say something yeah reach out first so what's the difference between scientific computer scientific computing and computer science how would you answer that that's a very nice question can i show already the link that i yeah we have there let's go for it yeah so anyway there's a so this first lesson is something that Enrico started about a year ago actually a little bit less than a year ago when we all went to remote work so in it he sort of tried to summarize all the different ways to use computers to do your work so there's a lot of different levels between using laptop and using national supercomputer so here we've tried to sort of go over this so this is sort of a bit focused for alto but there'll be a breakout room or the other meeting for the university of Oslo people but really almost any university you'd be at would have a lot of these same options so from what we say here you go and you figure out what the equivalent at your location is yeah and in this early in this first 10 15 minutes i'm trying to keep it general especially because i started studying computer science maybe i don't know 15 years ago 20 years ago and so i went through these basics but often i see colleagues who come from other fields from let's say psychology or natural sciences you know nobody has ever actually opened a computer in front of them and show the different pieces so even even though you know you don't need to know what's inside the car to drive a car it in some case it might be useful to know if you need to request a special engine if you need to use a special resource then it's good to know what they are and why that's different things so if i try to answer the what is computing on scientific computing the rich i was asking xkcds this nice picture that this is a machine learning system the data comes here you have a huge pile of linear algebra and then you have some answers so you know if we want to formalize this a little bit better when we often you know do research we might run an experiment or we simulate some data and so we end up with some raw data and what we need to do is to take this raw data filter reprocessive do some statistic fit some models with hypotheses that can come from the literature or other hypotheses that you have and basically all this process box here is where the computing happens so it's literally you know crunching numbers and then in the different steps we get you know preprocessed files or statistical maps figures plots etc which then will end up in your figures or posters or papers so that you can publish if you want to publish if you need to publish so this is kind of you know the process in a conceptual way but now when it goes to the actual you know how does it happen what's you know it's it's not just a black box that some data goes in and some pictures come out so in my opinion I still find it useful in 2021 to understand the components the physical components and also the more kind of software elements of a computing system so what you see in this picture here this is basically like I wrote the computing node but it can be a laptop it's also sometimes called a server but the idea is that there is some hardware and then on top of it there's some software running that lets you use the hardware so even though for some of you this might be trivial maybe you know this from high school I don't know when you learn about these things but it's still worth going through these little blocks of the hardware so inside the computing machine inside the server inside the computer there are CPUs now I don't really expand on the on the terms but you can read the mirror at the bottom so this is like where the the math happens there's some operation multiplication sums etc yeah and then of course you need some some kind of fast access memory it's like you know that you are doing some mental computation you want to do in your head two times two times two times two so you need to store some some of the intermediate numbers somewhere easily to find and then of course in a computer in a laptop in a server there's also a disk where you might want to store files so something a little bit bigger than what you would store in this RAM memory and here specifically I wrote SSD which is a type of disk that basically it's everywhere these days it's it's really fast it's what do you have for example in a USB stick so it's it's it has quite fast access to the files but then in some computers and in some servers you might also have a little bit more than CPUs you have the GPUs which is the graphical processing unit the GPUs were initially kind of used for you know processing graphics so that you can generate like beautiful images let's say that you want to play video game with a high frame rate but what it's it's interesting that that the architecture of the GPU is like many many CPUs layered in parallel so then you can really you know run you can use them also for computations for running many you know computations in in parallel and then often this is of course a very simple you know schematic of a hardware of a computing system but there's often a network system that you can because sometimes we need to access the internet you need to download stuff from the internet you need to access data that is more far away so here I wrote a big this kind of big bucket with the idea be that you get this big data it's way too big to be stored in this local hard disk and it might be stored in a remote location even you know far away on the on the planet I guess you could say things are getting a lot more complex like before there was CPU and now there's CPU and GPU that need to be programmed differently and before there was memory and now there's all these different types of storage that trade-off accessing it faster or being larger exactly and then on top of this you might even have other other limitations that maybe the data that you need to access is big and it's also sensitive then you know you can't really plug any cloud storage here you might have to use for example what your university has and this is also just just to remind the viewers that we have a HackMD so if you're if you're watching this from the from the link that we provided you go to the bottom and you click to HackMD if any of you has ever opened a computer and built one you can write it down and and tell your your experiences so but just just remember that you can ask questions in the HackMD and use it interactively all right but basically on top of the hardware of course there needs to be some you know software that we can use to interact with the hardware to actually use the hardware so every computing system has an operating system which is a piece of software that can basically you know talk with all the pieces of hardware and make sure that everything works together I guess you're all familiar with this and we don't really need to explain much more than it and on top of it you have software applications so now you for example you're running some web browser to watch this stream and you might already you might open I don't know MATLAB or R as we wrote here so another another distinction that I will also mention later when we talk about the different strategies of doing computing is that some of these applications have a graphical user interface with the idea being that you interact with the application with some buttons with you know I don't know you press a button to change the size of your phone to run an analysis but some other application which we focus especially on the second part today they use this command line interface now somebody might ask why do we still need to use command line interface in 2021 well if you think of the analogy of the of the coffee machine the coffee machine has few buttons and the coffee machine will just do what those you know five buttons do but if you really need to customize because you're a coffee you know expert and you really need to customize some timings of brewing or other things you need to start coding your executable bits of course you could start building your graphic or user interface it's you know it's if you have all the time in the world you cannot or layers on top of this but the first step is to start with the coding so that from the terminal you can launch you can start your executable and then we have basically I also mentioned your containers even though we won't really cover them in this course but maybe whether you already came across this term or you might come across this term the idea is that in the container it's a piece of software but inside you have you might have graphical user interface command line executable and another operating system so it's like that you carry a full software stack around and you can run it in multiple in multiple nodes so now a question that you should ask yourself or a question that comes obvious is like what do I need to make computing happen okay I have a piece of hardware or you know I got access to a HPC high performance computing cluster okay what do I need to do now well you know it's it's not easy to understand what are your needs it's not easy to know you know do I need more RAM do I need the better hard disk do I need the more CPUs I've wrote some questions here we don't really need to go through all of them but the typical example is that you get access to a supercomputer which has you know if the normal laptop has four CPUs supercomputer might you know you might have thousands you have access to thousands of CPUs but this doesn't mean that you're the software that runs on your laptop automatically sees these thousands of CPUs in the in the supercomputer it's actually it could even happen that something that runs on your laptop is actually faster when you when you write on your laptop rather than on the on the supercomputer which basically means that the code and the software needs to be adapted to use all the resources in the supercomputer another example that I often see is this GPUs it will be great if automatically work whatever code you wrote automatically can see okay there's a GPU in this machine I can use all the power of a GPU but in practice it's not like that GPUs need to have their own special language so the software that you have most likely needs to be rewritten or needs to be using library different libraries so that it can run on the GPU so you know I'm it's it's it's not simple and at the beginning might even feel frustrating to you know to kind of move from your local laptop that you maybe are used to to a to a remote system that is more more powerful but this is why we're here and this is what we kind of will cover in this in these three days yeah and this is sort of like the biggest barrier like you know you have your own computer and that's easy to use but you know the initial transition of changing to the big computer like it's as much an art as a science and well it's difficult to teach art so yeah and also I think that it might take time to understand what are your needs so at the beginning you don't know whether you need more memory or faster access to the data or more CPU so at the beginning it's a bit of a trial and error and it's okay and actually at least for those at Aldo we are here to help you in this process I think I think as I've been thinking back to things I learned more from working with people and see how they work than trying to you know read a book or read tutorials myself right now so like I really want that like value this sharing knowledge and co-working for things and it's excellent that you bring up the different way of working because you know in if there is no way this is not a single answer on how do I do scientific computing so some people they use these notebooks maybe some of you have used notebooks some people are happy with the command line interface that I mentioned earlier some people really need to use this high performance computing cluster some people use graphical interfaces like Matlab or RStudio and some other people they just write down the code you know with the with just a txt editor so finding your way is also part of this process it's also good like Richa was saying to learn a bit from the other you maybe sit down with somebody who is already been doing it for a while and see how they do it see if you like that way of working with the data and with computing and then eventually you know find your own your own way so here I'd finally to kind of wrap up this this bit is you know you started doing some computing maybe with your laptop and you're quite happy but you start feeling the limitations because your data is too big or your processing needs are growing so what is next where can you go where where can you take your code with you to a more powerful system so I kind of think you know this is just a personal thought there might be more categories but I think that we can kind of have three types of system so one system is like a remote desktop the idea of the remote desktop is like you know some of you might have an office and in this office you might have a workstation it's your own dedicated workstation it's much more powerful than your laptop so the idea is that you can connect to the workstation let's say that it's this one here in this cloud that I'm highlighting with the mouse and it's your own workstation there you can run you know your powerful whatever scripts and you know that nobody touches it and it's more powerful than your laptop another option is the virtual machine so we don't really need to explain or even know what is a virtual machine but the idea is that you can connect let's say through a website one example here is my binder that anyone can use and there through the graphical interface you can basically say you know what I need to run a notebook one of these Jupiter notebook and I need you know some power because I'm planning to run this and this analysis and so you can get one node meaning you can get one of these machines for your use until you are basically done with your analysis and then you can close and and continue with your with your local laptop or with your local desktop and then maybe the third type of kind of doing computing on remote is the i performance cluster where here it's if you see I wrote that the first two are a bit interactive in a sense that you connect to a system and you launch things and you know you get some answers and some pictures etc with the i performance cluster it's more like they have all many many many hundreds of these computing nodes waiting for something to do and here you just send to the queue so like you as I wrote the idea you submit a computing job so you tell the queue I would like to run this command this script for five days and when you're done let me know about it so then it goes to the queue and when it's your turn one of these computers picks up your your request executes it and then after whatever amount of time you get the answer so it's not interactive like the previous two workflow but then here you can really request you know because you know that let's say your code needs to run on multiple CPUs you you can have I don't know as many CPUs as you as your code needs so here's some literature yeah I guess these also sort of go up in the amount of difficulty like remote desktop is basically the same as what your laptop does but you're still limited to basically one computer and then these other the second level layer is still interactive which means it's easy to develop but you have more power and then once you get to the cluster you basically have to write a program to run your other program and that's what the Linux shell scripting tutorial is about coming up in 40 minutes it's a good point that you bring these apples on the on the levels there's also a level of kind of sharing a resource with the others that in this remote desktop it's your machine in your office or whatever it is and maybe you do what you want there but the more you go higher here you are sharing a big resource with many other people so you know you have to respect you have to somewhat follow the cluster etiquette when you when you start using these more powerful systems so here to wrap up at the end of this page that you should all have access to I listed I tried to find as many as possible services where you can you know click and start running a jupyter notebook or requesting a terminal and and doing some machine learning or whatever you need to do some of them of course as you might expect especially when you start reading things like Microsoft or Amazon some of them they are very limited with what you can do with the free or freemium whatever they call it but this is why many of the people watching here they also belong to a university meaning that you might have computing resources at your university and if I really go back to this picture here this is exactly what is seen at least in alto and in norway as well and in other universities in Finland so what would you say about the balance between free resources available to everyone and resources at your university like when do you recommend people to use the free resources sometimes or to stay within the university or what this is a very a very good point because for example I had to reply to a to a reviewer when I was when we were doing a scientific paper and I wanted to show interactively that you know we were right and they were wrong so I sketched I sketched you with the notebook using using google colab because I know that they can basically see the same system they don't need the password of my alto system in my case so in that case it was it was really convenient to share with others with us that I don't even know who they are because peer review is anonymous you know a little bit of code but then you know often it that's not the simple answer it's it's good or it's interesting to know and to get to know this system maybe you know one day you want to go and work in a company and maybe they use this amazon web services and elastic cloud systems and so it's good to learn and you can play with them for for free but when it comes to maybe computing power I would you know it's much easier to work with your own university and with the system that we have at at university they care like our university they take care of so much practicalities that you know if you if you start running your own amazon server and elastic cloud you basically also need to start learning about being an admin of a server and you know so it's yeah also there's some point like our university offers some of these cloud services like the microsoft azura for free but the difficulty of using that like in regular said you basically have to become your own system administrator to use that which is a few layers removed from typical research so that's why consulting with the experts at your university to find the right level of difficulty for you is good so um yeah any other comments here so we're constantly watching the hack md if you scroll to the bottom you see some questions there and um I guess we will well the questions are already being answered but we'll go through and answer some of them on the stream afterwards so I guess now should we split up into the university specific groups so if you're at alto university you can stay here and we will go over the different specific tools at alto so for example this computer has this data this computer has access this data and these are all connected this way if you're at oslo there is a separate zoom where you'll get uh oslo specific presentation and then at 45 minutes past the hour we will finish and begin a general question and answers back here on the stream where we'll go over the questions and hack and d and discuss them a little bit more and then um also hopefully work a break into there yeah that's very good so so maybe yeah let's I guess we can begin the alto thing right away and oslo can start once people get there okay since alto's here so once again you know this is a nice page and it's part of our website site com.alto.fi if you're not familiar with this website I you know I don't have anything against alto.fi alto.fi is a nice website and you can find solutions on alto.fi but here we try to make a little bit easier to find a solution that is related to computing so you can often try to find already a solution site com.alto.fi and one solution or one overview of all the computing system at at alto is here in this page that I'm using right now so there is no other page or at least I'm not aware of any other page in under alto that kind of gathers all these different types of workflows and all these different ways of doing computing at alto but what is nice is that you can edit this page if you feel that we're missing something you can submit edit request and we we can integrate it and you also get a credit of course credit as in karma not the not the study credit so anyway it it all goes down what I was already saying what's your style do you rather do you want to be independent with your laptop be you know your own administrator and decide what to run and what to install and you're happy with that and that's okay or do you really I don't know you're working a team that they have huge amount of data and you all need to use the same computing system so that's everything is possible in basically in the options that we have at alto do you think that there are one or the other is I guess also they're sort of like being able to use all of them like it's easy to develop code on your own computer and then you can scale up and like read on the cluster and then bring it back to your laptop so I guess portability is also very important somehow yeah it's in the end I mean you will find yourself using a bit of all these workflows that we that we mentioned in this page in general now this I'm more gonna focus on these remote workflows meaning that most likely you know how to use your laptop and maybe you know what to install or what you want to use and there you are your own boss and you can experiment and prototype and and work with your with your resources but things get interesting or get more a little bit more complicated when you start sharing the resources with other or as I mentioned some of the workflows in the in the page there sometimes you might have you might be working with some sensitive data that is that shouldn't leave this alto network that we see here so you know you can take a copy of the data to your own laptop and use it you should run everything or your you know your analysis at least with the real data should happen in on the within the alto network so if you think of these three systems that I mentioned earlier these remote machines there are some options that are open for everyone and I think these are actually for everyone even for for master student bachelor students right richer like anyone can connect to this yeah exactly the shell servers anyone can use and in fact I think that's they're basically recommended for students mainly yeah so this is a good starting point you connect to a remote machine and you can of course there will be other people connected so sometimes you might it might be a struggle to use as many resources that you can but this is a good starting point for everyone to to work with the remote desktop some people here they might be loaned to a department and they might have access to department workstation like I was saying earlier you know you have you have an office and you can connect to your workstation so yeah this is useful because you know that that's your workstation no one is going to touch it well sometimes maybe the administrator need to to reboot it or unplug it but in general you know this there is a I I wrote I wrote this picture that you know this is like the entry point so it's another computer but no computing happens in these computers this is just a gateway because if you would start running a huge you know python program on this computer you will start blocking the entrance for all the others so you enter in the gateway I wrote now some names here we have a list of those names in our pages but Maggie is the one for computer science department Amor is for mbe and then once you are in the gateway then you are in your department network and then you go to your workstation if you know the name of the of the workstation and now what is interesting here when you go to the workstation do you go buy SSH and shell or graphical or how does this connection work yeah this is a good question I didn't kind of like there are different ways of accessing this department workstation the SSH that we will show later the ritual we show later it's kind of like the terminal connection so that through the terminal you first SSH you first connect to this entry point to the gateway and then you SSH to your machine and then you can run terminal command but sometimes you can also or you might want to have a graphical interface so there is a way to get a graphical interface through this path but what it's interesting that you can also get a graphical interface to your workstation following this other path that I will demonstrate soon so but long long story short yes you can have both SSH or terminal interface and also the the graphical if needed okay thanks there's another there's another channel which is called SFTP which is like SSH but it's for file transfer so maybe sometimes you might also find yourself to use SFTP we have a nice page with all these options under psychom.alta.fi SFTP are sync scp you know we don't need to explain them now but you might find yourself yeah I guess there's so many different and I guess for people listening now you don't need to worry like realize there are all these options and we realize you'll probably need to thank for your particular case and ask someone which one you should use yeah and we'll go into this a little bit more about our Triton cluster next week in the high-performance computing kickstart things but anyway and now actually they were talking about file systems and accessing and transferring files it's worth mentioning that this is also something that is not transparent to the you know to people at Alta that there are actually different storage system and systems managed by different parties so you have your own kind of home folder which is this Alta ITS IT services and there you have your files your settings it's basically when you're logging in in a workstation what do you see there you know if you're at a workstation a department workstation or a student I guess the concept of network file system is something we should talk about too like there's the storage system but the same data is accessible on many different computers like for example no matter which Alta computer you're using if it's not a laptop it will have the same home directory so your same personal files there and then you can also do something called mounting and you mount the file system onto your laptop computer and you have access to these files this is really great because there's one central storage place which is shareable and backed up and snapshot it and really secure and you can use the data in all of these different places and this is something that may be really a non-obvious thing coming from working on your own computer and needing to manually move the file each time you need to and it's nice that you bring up the laptop because actually if you have a if you have Alta laptop they already kind of have the system so that the home folder in your laptop is synchronized with the workstation so you might be working on your laptop and then go to the department open the workstation see the same files so it doesn't always you know work perfectly but it's I can say that it's quite good especially if you're not dealing with huge really big files but so yeah this kind of how can I call it remote desktop or powerful machine workflow it's quite clear and what is interesting maybe what many don't know it's kind of using or accessing the Alta network and computing resources through some other systems I really like this vdi.alta.fi I'm gonna show it for you if you never used it because we still have some 10 minutes so you can visit this vdi.alta.fi there's also another link here in the what did it go in the page there is this mfa vdi.alta.fi the difference is that this mfa these are those rooms where those computer rooms that we have in um in various buildings at Alta so these are actually only windows machines instead in vdi you can I will show you now we can request a linux machine or a windows machine and what is nice is that you know it works to the web browser if you want of course you can install this vmware client so that you create this remote desktop connection with one virtual machine otherwise you know you might visiting your relatives and there's no way that you can install anything there but only through the web browser and your Alta account you can log in and then here you have some options so ubuntu 18 and ubuntu 18 in nvidia so these are linux machines and specifically this one as a gpu I would say that if you don't really need the gpu stick with the with the standard ubuntu just because there's more of them so there's no queue and then they're they're more stable and same story for the windows machine if you need these are also useful not just for computing you might need I don't know you might need access to microsoft world and you don't have microsoft world in your laptop you have it there and then so I'm going to click on the linux one and now this is exactly okay this is something that I left open from which this is also nice that that you have things you know that I was doing something yesterday and I didn't need to reopen everything I already have it there it stays there for 24 hours but here you know this is exactly what I meant you can this is the same view that you would have in a workstation at alto in a linux workstation at alto so you can start whatever software you need to write you need to you want to start a matlab or you know something else everything is is possible and this is something open for anyone at alto even beta students can can use this vdi system so you can even while I'm talking and while I'm showing you feel free to open a window go to vdi.alto.fi and try it out of course you know the amount of resources not infinite I think there are some hundreds of these virtual machines but eventually yeah but so far we never reached I never I'm using it almost daily I never found the day that I could not get access to a virtual machine and then to wrap up there is the triton cluster or you know the more the less interactive you can also actually we will show man that you can also interact with these notes meaning that you can get access to one of these computing notes and interact live sending comments and receiving answers but usually the way we work with this high performance computing clusters is to that we basically prepare the code that we need to run send it to a queue and then wait not too long and read the answers yeah so let's see what other options are there well maybe another option with this or yeah maybe just also to show some of you maybe through your coursework or other work you might be familiar with this jupyter notebooks that I also mentioned earlier if you're not familiar that's okay but it's it's something that you might come across so alto and especially triton the triton cluster we also provide this jupyter access to the triton to the triton notes this basically means that once again through a web interface so you don't need to install anything you can be at your grandparents with their whole computer you just need an internet connection you go to jupyter.triton.alto.fi and you can basically start computing you see the files that you would see when you connect to triton and you can run I can briefly show it if you so here you type your username and then wait that it starts so there you are I'm not going to start it now because otherwise it gets the the resources but here you can pick you know are you gonna need it for 10 hours four hours you know there's some different options and then when you click it's the it's the same kind of interface that you would if you're familiar with this jupyter yeah and this has all the same data as is available on triton so basically you can use jupyter.triton to develop and debug and so on and then you switch to the command line and submit a job and then you get well all the power you need this is also a good point about the triton data we will cover it also on Monday for those who follow the triton but you see that it's kind of independent from this alto ITS system so the difference is that the file system that we provide here is huge it's really huge I don't remember how many petabytes maybe two or three but the the other catch is that this is called scratch it means that it's not backed up yeah and occasionally we have someone to ask can you recover data and say well no but actually I think it's not I think it's a good thing that it's not backed up because the the idea would be that you know your original data like if I can go back to the to this schematic that I was showing here you know your raw data you should back that should be backed up or if it's a simulation you know those scripts that you were running to synthesize the data those should be backed up and you can ask your IT people what's the best way but then whatever comes after you know you don't want it to back it up because then you know that you back up the process to obtain those files so you're you know that you back up the the files that you need the the the scripts the code that you need but the output you know with you you don't have to care you don't have to start remembering okay this output which version was it do I is that the right one that I submit with the paper you can just rerun the process wait some days and and regenerate everything and this sort of like talking about the data things is a really interesting or trust really important concept which is the memory hierarchy so like Enrico said scratch is big but not backed up so the advantages are if someone says I need 200 terabytes of space for an experiment for a few months that's nothing like we can just give that to you but if you but it's not backed up so then you would have the original data on the alto IT services storage which we can also provide to you and then that is smaller so if you ask for hundreds of terabytes they'd sort of ask what you're doing and why but it's backed up it's snapshotted it's replicated to some other data center somewhere we don't know so that if the whole campus explodes you can still get that data back the downside is it's smaller and then you go to something like the hard drive on your own computer which well if not necessarily faster than these things but it's closer and you can use it yourself then you get to the memory on your computer which is the amount of RAM you have and then you get to the processor caches and once you get to like really when you're really kind of optimize code and make it run fast all of these considerations matter but I think they also matter from the general data management point of view like the idea of having the secure storage for the original data the scratch storage for the stuff you're working on lately the version control for the code is well it's also for memory hierarchy that we need to think about and these are the if you think back what I was saying earlier or you know when you start to understand the needs your computational needs then you start to understand okay I have a GPU GPUs they are really intensive they want lots of data and really fast maybe you that data then shouldn't leave a scratch maybe you need to move the data to the local disk of that node you know but these things we can you know it's a it's a learning that you will you will learn it by doing and by using it you know there's even a RAM file system where it uses the computer's main memory as what looks like a place you can store files but it's just stored in the computer's memory so once the computer reboots then it's gone and this can be useful for certain kinds of optimization but maybe we're going a little bit too deep so we should go back to the big summary picture oh i was asking about csc or is it time yeah so in this in this page here we also mentioned that this is what alto provides but in finland everyone who is affiliated with the research institution you can also use resources at csc csc has lots of training material and you know you can follow there but what is csc i'd never remember the acronym is it is it even officially back is it this center for scientific computing yeah i think i think you're right i think that's it i'm looking it up to see but anyway i guess we can say it's the center well the UPD says it's called csc hyphen it center for science so well anyway no one everyone just calls it csc but it's a government owned well government and university owned company which provides resources for science for education and research and most of its funding comes from the government not from the research project so you can use all of their things basically for free for research and teaching and we're actually being now just wanted to say that we've been actually working with them so that it's not too difficult to export and move your workflow from alto to the csc system so we can share some of the modules that we will mention on monday and other things that make like your life easier yeah the way you're saying yeah so each so maybe we can talk about sort of the general like give a summary now of the general tiers so you start with your laptop and then what would you say is the next higher level well then maybe you it's it's good to them try to move what can run on your laptop on your department workstation or virtual machine that i show with vdi yeah so that's okay and then you can go a little bit higher to jupiter hub like here so so then then you can start getting access to the tritone resources from a jupiter notebook you can submit the jobs the non-interactive jobs so it start to scale in a sense that you can get more you can get access to more resources with your with your computational needs here i also mentioned that you can actually have this interactive graphical or non-graphical sessions with the tritone cluster so the trick is basically to go through vdi now i still have the vdi open here so that um well i don't know how interesting this is of course for everyone but uh but the idea is that that we would show also later today that you can connect to the tritone cluster and then yeah with this you you can save the graphical interface so we're coming to the question time um but okay so next higher level is triton interactively and the last one is this non-interactive we basically what what i what i cover in the in the in the picture this is a long explanation of pros and cons it might be you know if you're starting with with tritone with our system it's it's worth spending the 20 minutes to go through this page and understand what is there so yeah let's show you have a look at the hack and deal is there any question it's highlighted by yeah so now it's question time so we have hack and deal here and we'll go through and scroll up and find any interesting questions to answer and you can continue um writing down there if you're in some of the other chats or zoom meetings you can also ask there and then um someone might relay them to us let's see so yeah also don't forget to take a break so at the next hour in 13 minutes we start the linux shell part so let's try to take a quick break before we get there okay i'm scrolling through the questions to see if there's anything yeah looks that our amazing helper i've already yeah reprised this is interesting what about the legality of code when using a free online platform this is uh uh it it always depends that's that's the simple answer i would say the platforms do not block any of your code nobody checks but you might use a code somebody in the license of the code of someone else's code that you might use they might they might add to the license you know you should not run this on amazon you know you can you can do what you want with your code so eventually you know you need to check the libraries that you're using yeah and i guess especially if you're doing something with innovations and so on the innovation office would say don't use free services that because then that's basically releasing it so that affects your ability to patent it or commercialize it yeah this is a very good point what i mentioned earlier that you can have sensitive data and store it in altosystem the sensitivity is also for the code you might have sensitive code because you're planning to patent or because it is sensitive that maybe does something you know important then then it's better that it stays in the alto network and on the alto system you don't want to carry it around right and here's some question about vdi well but for the previous question i guess if you like if it's sort of general science stuff and you talk to your supervisor and say yeah like i'm using this in colab to collaborate with people that's a pretty much like you know if you talk to your supervisor and say we're doing this i think many people do that and it's probably okay just make sure that there's no particular reason to keep it secret because basically you're releasing it to well your university security office would say that you're releasing it publicly even though it's not really public there's not security control there so i don't look that everyone got the answer some questions are specific to some of the systems that people are already using it's great that we're already actually familiar with this system the number of vdi machines with nvidia is limited yeah yes that's that's true i'm trying not to use them because i you know i don't basically do any gpu computing unless i'm just testing someone some of these errors but in general yeah that you know it's not the if you need to run more gpu computing then maybe triton has more more resources for for those and of course there's also this this classroom workstations in our documentation if you search there if you search there for panicky this is auto specific i know but in panicky you have access i hope my finish is correct you also have access to other gpu's i guess we should have said that also so that panicky classroom so sometimes students ask us okay we'd like to use gpu's for a project so like vdi well triton is what has most of the gpu's at alto but that is for research only so the panicky is a physical computer lab but you can also connect to it via ssh and when you connect there then you can use the gpu's for general student projects and connecting there via ssh well that uses linux shell which is what comes up next and this marie marie computer is actually what i mentioned earlier this um what was the name and everything this other other vdi option this so those are yeah mfa.vdi those are the marie the marie computers so the panicky computers are not only for computer science students anyone can use it well everyone at all yeah good point okay i think maybe it's good to stretch our legs and especially for the viewers should but you can keep on writing on the akmd we keep an eye on it then yeah we say that we are back at the sharp yeah sharp i guess basically sharp so see you then