You are so very correct. Especially for newer AI developers, long training times are not the norm. We use RTX through H100’s for most of our AI development- at least on the training side. However for coding, data sci work, inference and UI/UX we all use our favorite OS, whichever that is. One thing to keep in mind for pro level large parameter/data set AI dev, you will often be using a dedicated server running in the kilowatts with AI grade TPU/GPU’s (e.g. V100’S, H100’s, etc). Whether owned, hosted or otherwise, few jobs will be run locally.
I understand your point but I don't fully agree about your sentence when you say (with my words) "a CPU with just a load of RAM will be enough"... I'll explain why: Though you are right saying we have to prioritize RAM, but CPU is important too...try training a model with Weka workbench (java based) on you laptop or desktop computer... a fast CPU will help. Students will do deep learning and not necessarily limit themselves in machine learning with scikit or whatever framework. so... a) Having a lot of RAM yes but with a very good CPU too... most probably when working with MLmodels is because you are probably working on an application that requires many components where all are not necessarilly ML based. You could design a NodeJS driven UI that will interact with some back end that you still develop onto your computer and that will serve the model. In order to make it in a very efficient and organised way, you will endup with containers and there is why you'll need CPU and RAM (though they are lightweight). b) because of (a) you will probably start diving in both DEVOPS techniques and MLOPS paradigm. Both of them will require automation which will also consume CPU. Especially if you build a C++ or Java application that must be built. c) because of (a) and (b) your computer will start to gain some load just to work all these things. d) Though an NVIDIA RTX is quite expensive, it can help you a lot on doing deep learning tasks and allow TF to use the onboard GPU. there you will face interresting issues. You'll probably hit during training the VRAM limits and will have to work hard but learn in order to get a really good neural network architecture running on your machine. e) You talk about using cloud, yes I agree partly, this is only for experienced people. Other will get hard time to make it work (I am not talking about Google colab or any other fantasy stuff). Therefore you will travel from (a) to (d) on your local machine. Personally I follow you and agree on what you say about using open sytems and not using macs. 2 years ago I bought a Linux Laptop with an onboard NVIDIA RTX. Because of the budget I could only afford an RTX3600. But I could have a very good intel i7 16 vcpus and 32 GBRAM. All that for less than 1800€ with a wide screen (17"). But that was 2 years ago. Today I would go for a more robust RTX card and smash 64GB or 128GB RAM directly. The only thing I think I would recommend in that... is the battery, choose good ones and chose a laptop with spare batteries. Also, because you will work with Docker containers and perhaps have many versions of the virtual environments in Python... think about the disk space => I recommend today MINIMUM 2TB of SSD. If you can afford more, better it is. Then yes using a cloud solution is also elegant but you'll still need to consider an efficient laptop too because of (a) to (e)..
@@jayakumar4633 mine is a Clevo. Go to Laptop with Linux. But today, 2 years after, things have evolved a lot and i would consider run Tensorflow onto Mac OS. But i still haven't bought it. But certainly it Will be more expensive but there you get An M chip.
I got myself an m1 air a couple of months ago. One thing I dislike is that tensorflow has multiple issues with Mac. It's better to learn about scaling and deploying first, because clouds are always available, rather than throwing a large amount. As for whether it's worth it when you're very advanced in the field, I'll update when I get there 😂 Side note I have a 3070 but I realised model design, preprocessing plays more of a part in ML.
@@JesperDramsch thanks for replying, you made want to leave an update since the time I commented. The M1 air is very strong when it comes to smaller models, especially if you're working with scikit learn. Very snappy with pandas as well. But leave the LLM/ CV models to the cloud or a GPU. Just one month after the comment, I ssh/remote into my home PC for most of my ML work. Or do my training on my work server. Overall, I personally feel that one should not buy a laptop while considering for ML/AI. It's not worth it. Get what suits your use case. For me I'm quite happy with a smaller screen as I need to carry my laptop around and I work on the commute. I also do not game so mac is fine. Hope this helps someone out.
Great advice. Because I've been looking for a second machine for my deep learning research. Now, I will switch my strategy from a local machine to the cloud. Thanks.
The M1's run very cool and you don't need to worry about running long tasks even in the fan-less one. All 1st gen M1's have a maximum of 16GB of (V)RAM(including the iMac), M1 Ultra has a maximum of 128GB of (V)RAM. Unified Memory also removes the PCI bottleneck between VRAM and RAM, with a bandwidth of up to a combined 0.8TB/s.
How is the compatibility of python packages? As I see on the internet, many people still face issues. I'm willing to buy an m1 and pursuing a career in data science and ml, so I will be using python a lot. Also I'll use docker for some of my classes in school. I'm very very indecisive because of the incombatibility issues.
@@beytulk There are not any alternatives were you would have any type of portable device(w/o being plugged in all the time to run anything locally). I have not had any issues, I think the issue has been that pytorch was slow to update, by the time you have your system up I don't think anything will be missing. About 0.1% non M1 software on my system now, at launch it was about 40% so it has been a really fast transition from a dev perspective.
@@forlorn8025 M2 and M3 are just overclocked slightly more efficient M1 versions M4 will probably be the first worthy upgrade for ML with acceleration on CPU and GPU + better NPU
Already owned a Acer Nitro 5 with RTX 3070 mobile + R7 5800H. Still watch you full video :). And my laptop can train 90% types of model after I cramp up the virtual memory => 80 GB (from 16GB of RAM 😂). I'm very satisfied with my $1500 laptop
Just discovered your channel and I'm already a fan of you. I have recently started learning data Science, currently learning python and statistics. Looking forward to your guidance through this channel.
@@fianekosaputra417 Right now I'm using my old samsung np300e5c laptop to learn python. I'll be buying an AMD 6000 series laptop with rtx 3060 once it comes out
This was the first video of yours I ever watched and when I started I thought, naaah a new MacBook Pro could surely be fine for training models. I can't tell you how wrong I was. The hype is very different from reality and you are 100% correct. I have had to embed so many special cases into my training pipeline to support MPS (METAL) and even then support for torchvision is still incomplete in V2. I ended up going for an rtx4090 on a separate headless Linux server and it reduced training time on my use cases by an order of magnitude.
This is a really good video! i stumbled here and got close to buying NPU mini PC. However, I think I am okay running Gemma 2 Ollama and learning. Thank. Subscribed!
Hi! I am also starting to to learn AI and ML now. Can you please help me with a few things. 1) After what amount of time will I need a better laptop or can I do it on my current laptop? Right now I have an office laptop with intel i5 10gen U Processor with integrated graphics 2) Since I am starting to learn where should I start for AI & ML? 3) Is Asus ROG Flow X13 2023 a good option? It has Ryzen 9 7940HS and Nvidea RTX 4050 6GB (60W). I want this one because it is super portable and would also help in taking notes since it is touchscreen. Also is 16GB RAM enough in the laptop? It would be great if you could help me out a bit. Thanks!
This was really cool. I bought old server GPU and put it into a big desktop PC Case with a hydroponics fan to cool it down. Got a 24gb tesla card for usd$300 that is about as fast as a 1070ti
Agree with many things here, great video! However, using cloud GPUs is cheap at first sight, but letting a model train for days on cloud GPUs might be much more money than your electricity bill and cost you in the hundreds (with a sizeable model) and should be considered into the whole calculation. Cloud GPUs range from 0.2 €/hr (single 3090) up to 4€ /hr (multi-A100), a discrete GPU might pay off in less than a year depending on your project.
It's a consideration for sure. But if you're running a model for multiple days straight, I would try and dissuade from a laptop regardless and go for a desktop or server-solution. It'd be really annoying to have a laptop that has to stay on for days on a power outlet.
That's a good one, you can do deep network also on a cpu and for small models is even faster then a gpu. About using the cloud I have a different take, I'm using one computer for browsing and ome for training. 😂
@@JesperDramsch I ended up sticking to Colab and buying a Raspberry Pi to assemble a Bitcoin node instead. Indeed, no need to go too far on a hardware trip, especially for starters.
Should I consider buying a normal Laptop with Intel Iris Xe if i work in AI , ML, DL or some gaming laptops Nvdia RTX , GTX gpus ? Btw I am a non gamer
Technically g in gpu stants for general it was just used for graphics alot so it shifted now its returning to true aka general but you need cuda and tensor cores
Was looking for an EDC work laptop that can handle personal scoped AI ML on my own air gapped network at home. That way i can leave the models and data private and connect the laptop by cable to train the model when needed. Appreciate this extremely detailed information and how it all relates to AI and ML. My very own private air gapped AI 😀 sounds like I'm better off setting up a tower AI build and go with maybe a laptop with one of the new intel core ultras for otg edc.
Hi given that I will do the learning on cloud but it may not be always available while travelling. So, what minimum configuration would you recommend. I am a beginner and student . I travel so I have to also carry a work laptop. portability is a concern. I thought MacBook Air with m1 but ram could be concern. Can you suggest minimum ram, processor, gpu(if needed) and other.
I'm preparing to train on two laptops, Dell Precision 7550 with Quadro T1000 dedicated graphics, 64MB RAM and Windows or Dell Precision 7520 with 16GB RAM and Quadro M1200 (it's not even supported by Rapids, so we'll see). Maybe this year I will buy PC with nVidia 3070 or other new card, depending on the market offer (we'll see how Intel ARC will behave with AI).
The new Intel Xe gpus have support for machine learning via OpenVino. Intel seems to be pushing this really hard. In fact the whole Xe gpu architecture is designed around being multipurpose matrix specifically for doing lots of computation necessary for ml.
Minisforum MS-01 i9 13900H 32gb ddr5 ram (max 96gb) and 1tb for 900euro is a great base to start if you dont want a big machine ....(with cloud service)
I try the classical ML in my old Linux machine. It works very well. As you said in the video, it is a better idea to run Deep Learning in Cloud service. BTW, if I create and use my own data for creating a Deep Learning model in the cloud service, it will be guaranteed that my model and my train data set will belong to me?
Great explanation video! One thing I would have liked to hear more about is dominance of the Nvidia CUDA framework. It seems to me that a lot of ML python libraries are compiled to work with the CUDA framework and therefore one would need to run it on Nvidia hardware. That’s the advantage that Nvidia has because it started 20 years ago developing the CUDA framework and was miles ahead of everyone else in the field of deep learning. As you said, things like Tensorflow is just starting to have Apple silicon /aarch/arm64 architecture. But Nvidia continues to innovate with RAPIDS (CUDF vs. Pandas) and with their NVLink on their DGX A100 and DGX H100 (8 GPUs w 80GB VRAM each and all linked together). However, with respect to laptop for ML, would it make sense from the perspective of a DevOps use case? Rather than using the laptop to train a huge LLM (llama2 , falcon40b, mistral, etc.) what if I just want to test a few of the prepackage Nvidia NGC containers in docker and add some additional python packages/libraries to them and test training on a smaller dataset smaller model to confirm that things work and then move the container over to the Cloud like Amazon AWS and run it on Nvidia A100 or DGX A100 resources to do the full training? Would laptops with nvidia GPUs (for docker, kubernetes, VMware) for DevOps testing purposes be useful or not at all? Thanks.
I'm starting machine learning with a neural network for a school project, he 😅. First I start with Google Colab because there are too many things to install/setup and I wanted to omit that tedious step. But... I exceed the limit of GC, so I decided to test the model on my laptop (I'm kind of stingy). I looked how to use de GPU on the env created for this task (again I'm starting in ML, just knew a few of python) and I discovered more packages/libraries to install and deal with the compatilibily with windows, my Nvidia card, ... 🤯 (regular user). It's too overwhelming, I even considered to partion the disk for Linux OS or use a VM, 'cause I need some programs that don't run in Linux. Just too much. I'm too grateful for this video, it has been my lifeline in a moment of despair 🥹🥹🥹
@JesperDramsch I'm just starting out with some ML/AI projects, and I already know I'll be working with fairly large datasets. I'd be interested in hearing your opinion on whether it would make sense to use my current Mac Pro with these specs by installing an Nvidia RTX 4090 and running Linux: 3.3 GHz 12-Core Intel Xeon W 256 GB 2933 MHz DDR4 Or do you think it would be more efficient to build a separate setup? Obviously, building a separate setup would cost me more since I work with Macs and I wouldn’t want to sell my current Mac. What do you think?
Hi Jesper and tank you for your video, informative as usual. I'd like to ask you what you think of a laptop with a Ryzen 7 5825u and no GPU, but with the intention of connecting a "prosthetic" desktop GPU in the future through a pcie connector; I'm talking about something like the EXP GDC "THE BEAST". Or do you think it's easier and better just using an external GPU through thunderbolt 4?
Thanks for your recommendation with that cloud you suggested. I will try that! Though, I write deep learning networks also for my own projects in computer vision and would prefer not to use a cloud. Also because that requires internet access and I also enjoy working at places where I sometimes don’t have internet. Therefore, I’d appreciate a video about good laptop options for that purpose. P.S. It needs to be compatible w/ Fedora or Ubuntu distro
@@JesperDramsch to work for 2h outside it’s sufficient when my laptop is fully charged. Obviously, I will not let a very deep network learn for several hours if I just planned to go outside for 2h. In that time I would either build parts of the architecture or some test to try something out.. some ideas for architectures.. etc.
@@stellamn that makes sense. I'd go for portability anyways then. Build it offline. Make tests offline then train in the cloud later when you're back. Otherwise see that you get a laptop with a compatible Nvidia card and enough RAM if you still want to train on your laptop.
@@JesperDramsch don’t worry, I did already my own research and found some possible solutions. Thanks for your suggestions :) (I didn’t see the prior message)
I'd say that using macBook Air 15" and Ryzen based PC = both at price of single macBook Pro - does offer best value. And as far as upgrade is considered - first, do math Ph. D.
What's your take on new macbook pro models, am planning to get the 14 inch base model, will that be enough, should I upgrade anything in it like Ram, more CPU and GPU cores?
@@JesperDramsch I'm looking at picking up the new Mac book pros and my main task is learning ML and DL - I'm interested in computer vision object recognition, speech to text and NLP what would you recommend?
is it big difference between dedicated or integrated? I want to buy expertbook core i7 -32 RAM BUT it has iris xe .is this gpu enough for my works as you said???
I am working in java, python no-sql & bigdata. And want to learn NLP & deep learning. I'm looking for a durable laptop. My intention to buy a gaming laptop is learn to deep learning only. Not for gaming. I would like to buy Omen Ryzen 5800H with Rtx 3060 6gb or Legion Ryzen 5600H with Rtx 3050 4gb. I have two questions- 1. Will you suggest to buy a gaming laptop to learn ML/DL? 2. Suppose I am working on a java application, a non gpu task on that laptop. Will gpu run all the time? Will it effect the laptop longevity due to heat generation? Please advice.
Great video Jesper. Can I ask u? Wud be better laptop with i7 1260p and 64gb Intel Iris graphics xe or i7 12700h, 32gb, Intel Iris Xe too but having gpu RTX 3070 8gb? Second little more price about 250€. Thx for your help.
@@JesperDramsch thanks 👍 Great and informative video by the way. If you are buying the new Macs please do another one in depth like this. I believe people around the world are looking for this kind of content
@@JesperDramsch its weird that your video comes in my recommendation when I just faced the problem of my 32 gb ram not enough.. upgrading to 64 gb now.
Here's how you train a model in the cloud for free:
ruclips.net/video/Ld8vPbtyJWQ/видео.html
If vram is low then can extra ram get you over the line ? Like 16, 32 or 64GB ?
@@Matlockization No they're connected to a different kind of chip and aren't interchangeable.
@@JesperDramsch I see, thank you.
Help me should I buy black diamond from dell
@@JesperDramsch please help because I am new to these developments I want to be future proof so in near future I don't need to buy more
Man, this is so so helpful :) Many thanks for patiently covering all the key concepts !
Of course!
This channel needs way more subs! The content is high quality / well explained. :)
You are so very correct. Especially for newer AI developers, long training times are not the norm. We use RTX through H100’s for most of our AI development- at least on the training side. However for coding, data sci work, inference and UI/UX we all use our favorite OS, whichever that is. One thing to keep in mind for pro level large parameter/data set AI dev, you will often be using a dedicated server running in the kilowatts with AI grade TPU/GPU’s (e.g. V100’S, H100’s, etc). Whether owned, hosted or otherwise, few jobs will be run locally.
I understand your point but I don't fully agree about your sentence when you say (with my words) "a CPU with just a load of RAM will be enough"... I'll explain why:
Though you are right saying we have to prioritize RAM, but CPU is important too...try training a model with Weka workbench (java based) on you laptop or desktop computer... a fast CPU will help.
Students will do deep learning and not necessarily limit themselves in machine learning with scikit or whatever framework. so...
a) Having a lot of RAM yes but with a very good CPU too... most probably when working with MLmodels is because you are probably working on an application that requires many components where all are not necessarilly ML based. You could design a NodeJS driven UI that will interact with some back end that you still develop onto your computer and that will serve the model.
In order to make it in a very efficient and organised way, you will endup with containers and there is why you'll need CPU and RAM (though they are lightweight).
b) because of (a) you will probably start diving in both DEVOPS techniques and MLOPS paradigm. Both of them will require automation which will also consume CPU. Especially if you build a C++ or Java application that must be built.
c) because of (a) and (b) your computer will start to gain some load just to work all these things.
d) Though an NVIDIA RTX is quite expensive, it can help you a lot on doing deep learning tasks and allow TF to use the onboard GPU. there you will face interresting issues. You'll probably hit during training the VRAM limits and will have to work hard but learn in order to get a really good neural network architecture running on your machine.
e) You talk about using cloud, yes I agree partly, this is only for experienced people. Other will get hard time to make it work (I am not talking about Google colab or any other fantasy stuff).
Therefore you will travel from (a) to (d) on your local machine.
Personally I follow you and agree on what you say about using open sytems and not using macs. 2 years ago I bought a Linux Laptop with an onboard NVIDIA RTX. Because of the budget I could only afford an RTX3600. But I could have a very good intel i7 16 vcpus and 32 GBRAM. All that for less than 1800€ with a wide screen (17"). But that was 2 years ago. Today I would go for a more robust RTX card and smash 64GB or 128GB RAM directly.
The only thing I think I would recommend in that... is the battery, choose good ones and chose a laptop with spare batteries. Also, because you will work with Docker containers and perhaps have many versions of the virtual environments in Python... think about the disk space => I recommend today MINIMUM 2TB of SSD. If you can afford more, better it is.
Then yes using a cloud solution is also elegant but you'll still need to consider an efficient laptop too because of (a) to (e)..
Could you please recommend any old workstation laptops thank you
@@jayakumar4633 mine is a Clevo. Go to Laptop with Linux. But today, 2 years after, things have evolved a lot and i would consider run Tensorflow onto Mac OS. But i still haven't bought it. But certainly it Will be more expensive but there you get An M chip.
For PhD students, maybe it's better to access a University PC from your MacBook from home :D
Definitely, as long as the university has a GPU server or GPU desktops. It's exactly what I did actually!
I got myself an m1 air a couple of months ago. One thing I dislike is that tensorflow has multiple issues with Mac. It's better to learn about scaling and deploying first, because clouds are always available, rather than throwing a large amount. As for whether it's worth it when you're very advanced in the field, I'll update when I get there 😂
Side note I have a 3070 but I realised model design, preprocessing plays more of a part in ML.
Agreed. Thanks for the follow-up!
@@JesperDramsch thanks for replying, you made want to leave an update since the time I commented.
The M1 air is very strong when it comes to smaller models, especially if you're working with scikit learn. Very snappy with pandas as well. But leave the LLM/ CV models to the cloud or a GPU. Just one month after the comment, I ssh/remote into my home PC for most of my ML work. Or do my training on my work server.
Overall, I personally feel that one should not buy a laptop while considering for ML/AI. It's not worth it. Get what suits your use case. For me I'm quite happy with a smaller screen as I need to carry my laptop around and I work on the commute. I also do not game so mac is fine. Hope this helps someone out.
@@waynelau3256 love this Wayne. It's similar to my experience. Happy to train scikit-learn on my phone. Not gonna get a laptop for DL / LLM
Great advice. Because I've been looking for a second machine for my deep learning research. Now, I will switch my strategy from a local machine to the cloud. Thanks.
The M1's run very cool and you don't need to worry about running long tasks even in the fan-less one. All 1st gen M1's have a maximum of 16GB of (V)RAM(including the iMac), M1 Ultra has a maximum of 128GB of (V)RAM.
Unified Memory also removes the PCI bottleneck between VRAM and RAM, with a bandwidth of up to a combined 0.8TB/s.
How is the compatibility of python packages? As I see on the internet, many people still face issues. I'm willing to buy an m1 and pursuing a career in data science and ml, so I will be using python a lot. Also I'll use docker for some of my classes in school. I'm very very indecisive because of the incombatibility issues.
@@beytulk There are not any alternatives were you would have any type of portable device(w/o being plugged in all the time to run anything locally).
I have not had any issues, I think the issue has been that pytorch was slow to update, by the time you have your system up I don't think anything will be missing.
About 0.1% non M1 software on my system now, at launch it was about 40% so it has been a really fast transition from a dev perspective.
@beytulk which laptop did u buy?
@@forlorn8025 M1 Air 16GB and thermal modded it at launch, no issues
@@forlorn8025 M2 and M3 are just overclocked slightly more efficient M1 versions
M4 will probably be the first worthy upgrade for ML with acceleration on CPU and GPU + better NPU
ur right my laptop is really aerodynamic... i find myself playing frisbee with it all the time
Ultimate frisbee or fetch with a retreiver dog?
@@JesperDramschThe ultimate throw and regret kind
I am a complete computer novice and this video was soo helpful. Thankyou
I'm so glad to hear this!
Thank you
You just saved me from getting broke 😅
I was thinking of getting a pc but confused on what graphics card to get
Already owned a Acer Nitro 5 with RTX 3070 mobile + R7 5800H. Still watch you full video :). And my laptop can train 90% types of model after I cramp up the virtual memory => 80 GB (from 16GB of RAM 😂). I'm very satisfied with my $1500 laptop
How do u integrate 80gb ram into a laptop 🤭
We're due for an updated video after the m1x Mac
Isn't the successor to the M1 the M1 Pro and M1 Max after all instead of m1x?
Just discovered your channel and I'm already a fan of you.
I have recently started learning data Science, currently learning python and statistics.
Looking forward to your guidance through this channel.
So, what laptop is your choice?
@@fianekosaputra417 Right now I'm using my old samsung np300e5c laptop to learn python. I'll be buying an AMD 6000 series laptop with rtx 3060 once it comes out
Thanks so much Aamir! Good luck on your journey.
How far have you gone since?
This was the first video of yours I ever watched and when I started I thought, naaah a new MacBook Pro could surely be fine for training models. I can't tell you how wrong I was. The hype is very different from reality and you are 100% correct. I have had to embed so many special cases into my training pipeline to support MPS (METAL) and even then support for torchvision is still incomplete in V2. I ended up going for an rtx4090 on a separate headless Linux server and it reduced training time on my use cases by an order of magnitude.
I appreciate that feedback!
METAL is worthless IMHO - having a mac here in multiboot OS, the Radeon chips in the imac dont really hold a candle to the nvidia GPU's for ML.
Thanks a lot for making such a helpful video man.
Thanks for stopping by! Glad it's so helpful!
Video recommended days after I bought a full Desktop Computer
Don't feel bad. I'm still rocking a full desktop PC and I wouldn't change for the world.
Nowadays with LlaMA models we might can run some models on our laptops
Llama has 65 billion parameters. Are you sure?
Loved this video, Jesper. Thank you!!! I’m happy I came across your channels
Thank you, Rebeca! Welcome in.
This is a really good video! i stumbled here and got close to buying NPU mini PC. However, I think I am okay running Gemma 2 Ollama and learning. Thank. Subscribed!
Hi!
I am also starting to to learn AI and ML now.
Can you please help me with a few things.
1) After what amount of time will I need a better laptop or can I do it on my current laptop? Right now I have an office laptop with intel i5 10gen U Processor with integrated graphics
2) Since I am starting to learn where should I start for AI & ML?
3) Is Asus ROG Flow X13 2023 a good option? It has Ryzen 9 7940HS and Nvidea RTX 4050 6GB (60W). I want this one because it is super portable and would also help in taking notes since it is touchscreen.
Also is 16GB RAM enough in the laptop?
It would be great if you could help me out a bit.
Thanks!
Thank you Jesper!
Nice coverage of laptop configurations for ML.
Thanks!
I trained a CNN on the MURA dataset (40k mri’s). My MacBooks battery swelled and the whole top case had to be changed 😅
Btw! I used my macs gpu with plaidML, but it doesn’t look like it’s being maintained anymore 😫
That's terrifying. I hope it was under warranty 😱
@@JesperDramsch yes! Or else I would be crying still 🥴 now I write most of my code locally, then upload to colab for training.
Very relatable.
Thank you for this. I'm interested in ML for advancement in my career and your explanation is helpful.
I'm glad! Thanks for stopping by!
Wow thank you a lot i feel like i understand a lot better
Nice video. Very clear explanation
Thank you!
Excellent information!
Thanks a lot, man! Very useful tips
Glad I can help
0:42 I felt that
Thank you for this. Is there a need to offer an update to this video given that it is now two years later?
Maybe replace M1 with M2 that's it.
Laptops are aerodynamic 🤣🤣🤣🤣🤣🤣...
💸
I liked the expression 😅
Very very helpful 👏. Thanks alot .
Glad to help!
Very helpful! Thank you so much.
Glad it helped!
That was really helpful. Many thanks
Glad it helped!
Thank you very much!
This was really cool. I bought old server GPU and put it into a big desktop PC Case with a hydroponics fan to cool it down. Got a 24gb tesla card for usd$300 that is about as fast as a 1070ti
That's very cool!
great video, thank you!
Thanks!
is it good to use the x-elite notebooks for machine learning and deep learning?
good video! keep up the good work
0:43 lmao I hope that's edited, cause my soul damn near left my body when I heard that sound
This shall be forever a mystery 🙃
That is really helpful!
Thank you!
Agree with many things here, great video! However, using cloud GPUs is cheap at first sight, but letting a model train for days on cloud GPUs might be much more money than your electricity bill and cost you in the hundreds (with a sizeable model) and should be considered into the whole calculation. Cloud GPUs range from 0.2 €/hr (single 3090) up to 4€ /hr (multi-A100), a discrete GPU might pay off in less than a year depending on your project.
It's a consideration for sure. But if you're running a model for multiple days straight, I would try and dissuade from a laptop regardless and go for a desktop or server-solution. It'd be really annoying to have a laptop that has to stay on for days on a power outlet.
@@JesperDramsch you are right - I kind of ignored the fact your specificly talking about laptops - more used to SSHing next to my GPU
Agree there
Such a helpful video! I need an update for 2024 products. Can't decied on which laptop to buy.
Thanks! Honestly it mostly still holds 😅
What about RTX 4090 for deep learning?
If Im a pyshician and need to have data privacy? I can't just use an online notebook and lie about it in the paper 😅
Could you please recommend any old workstation laptops
That's a good one, you can do deep network also on a cpu and for small models is even faster then a gpu. About using the cloud I have a different take, I'm using one computer for browsing and ome for training. 😂
What a great video!
Very good, thanks.
Thanks hazel!
@@JesperDramsch I ended up sticking to Colab and buying a Raspberry Pi to assemble a Bitcoin node instead. Indeed, no need to go too far on a hardware trip, especially for starters.
Should I consider buying a normal Laptop with Intel Iris Xe if i work in AI , ML, DL or some gaming laptops Nvdia RTX , GTX gpus ? Btw I am a non gamer
Which laptop did you buy
@Sci-techsaga Asus tuf 😅
Technically g in gpu stants for general it was just used for graphics alot so it shifted now its returning to true aka general but you need cuda and tensor cores
Most people keep pondering whether to buy an SUV when they actually need is a cycle.
Tensorflow has issues with M1/M2 macbooks.
What about m1 now in 2023
Same story. Better support for metal but not everything's quite there.
Thanks man
Nice video!
your GEFORCE RTX is an NVIDIA card right? got confused!
Was looking for an EDC work laptop that can handle personal scoped AI ML on my own air gapped network at home. That way i can leave the models and data private and connect the laptop by cable to train the model when needed.
Appreciate this extremely detailed information and how it all relates to AI and ML.
My very own private air gapped AI 😀 sounds like I'm better off setting up a tower AI build and go with maybe a laptop with one of the new intel core ultras for otg edc.
this high-pitch sound you heard most likely came from capacitors
Possibly. Might also be coil whine though.
Best comments ever on buying laptop for ML
Thanks Raj!
How about running Mindspore on Macs and laptops with Nvidia?
you tube algo gods finally decided to give me a relevant video rather than click baits
Neat! Welcome!
Does the new AMD GPU work with deep learning ?
It's getting better, but often it doesn not.
How is rtx 2050
Excellent
Thank you!
For windows, when you say ram.. Is refer to RAM or gpu ram?
RAM is the key. At least, from my experience in ML.
So, what is your suggestion for a person who is planing to get a laptop for ML/DL workloads?
Don't. Buy MacBook Air, decent display, Ryzen based PC with 64 GB of RAM or more and any NVidia GPU.
Cool video. I had a choice to make between 8GB and 16GB mac. I will take a 16GB now.:) Thank you!
Love to hear it!
Hi given that I will do the learning on cloud but it may not be always available while travelling. So, what minimum configuration would you recommend. I am a beginner and student . I travel so I have to also carry a work laptop. portability is a concern. I thought MacBook Air with m1 but ram could be concern. Can you suggest minimum ram, processor, gpu(if needed) and other.
I'm preparing to train on two laptops, Dell Precision 7550 with Quadro T1000 dedicated graphics, 64MB RAM and Windows or Dell Precision 7520 with 16GB RAM and Quadro M1200 (it's not even supported by Rapids, so we'll see). Maybe this year I will buy PC with nVidia 3070 or other new card, depending on the market offer (we'll see how Intel ARC will behave with AI).
Were they any good? (The dell precision laptops)
2024 is this advice still relevant? Eg using only nvidia for deep learning
Mostly yes. Situation is better with AMD and LLMs make GPUs more central to the equation
I am a Data Science student and my laptop specs are low but can I use only cloud for the projects?
At least in the beginning you should, yes.
@@JesperDramsch Okay Sir, thank you !
So for a windows user the lenovo ideapad 5 would be a perfect option you think?
The new Intel Xe gpus have support for machine learning via OpenVino. Intel seems to be pushing this really hard. In fact the whole Xe gpu architecture is designed around being multipurpose matrix specifically for doing lots of computation necessary for ml.
Minisforum MS-01 i9 13900H 32gb ddr5 ram (max 96gb) and 1tb for 900euro is a great base to start if you dont want a big machine ....(with cloud service)
Any market recommendations? I waa told the Lenovo Yoga is a great buy, I am new on this topic. What do I need to get on memory etc?
I try the classical ML in my old Linux machine. It works very well. As you said in the video, it is a better idea to run Deep Learning in Cloud service. BTW, if I create and use my own data for creating a Deep Learning model in the cloud service, it will be guaranteed that my model and my train data set will belong to me?
That's awesome. Normally yes, if you go with any of the mentioned services it's your property
helpful video thanks, are you a german native ?
Yeah, I'm from the North. Thanks!
Great explanation video! One thing I would have liked to hear more about is dominance of the Nvidia CUDA framework. It seems to me that a lot of ML python libraries are compiled to work with the CUDA framework and therefore one would need to run it on Nvidia hardware. That’s the advantage that Nvidia has because it started 20 years ago developing the CUDA framework and was miles ahead of everyone else in the field of deep learning. As you said, things like Tensorflow is just starting to have Apple silicon /aarch/arm64 architecture. But Nvidia continues to innovate with RAPIDS (CUDF vs. Pandas) and with their NVLink on their DGX A100 and DGX H100 (8 GPUs w 80GB VRAM each and all linked together). However, with respect to laptop for ML, would it make sense from the perspective of a DevOps use case? Rather than using the laptop to train a huge LLM (llama2 , falcon40b, mistral, etc.) what if I just want to test a few of the prepackage Nvidia NGC containers in docker and add some additional python packages/libraries to them and test training on a smaller dataset smaller model to confirm that things work and then move the container over to the Cloud like Amazon AWS and run it on Nvidia A100 or DGX A100 resources to do the full training? Would laptops with nvidia GPUs (for docker, kubernetes, VMware) for DevOps testing purposes be useful or not at all? Thanks.
I'm starting machine learning with a neural network for a school project, he 😅. First I start with Google Colab because there are too many things to install/setup and I wanted to omit that tedious step. But... I exceed the limit of GC, so I decided to test the model on my laptop (I'm kind of stingy). I looked how to use de GPU on the env created for this task (again I'm starting in ML, just knew a few of python) and I discovered more packages/libraries to install and deal with the compatilibily with windows, my Nvidia card, ... 🤯 (regular user). It's too overwhelming, I even considered to partion the disk for Linux OS or use a VM, 'cause I need some programs that don't run in Linux. Just too much. I'm too grateful for this video, it has been my lifeline in a moment of despair 🥹🥹🥹
Yeah, it can be a nightmare to install everything. Glad this was helpful!
how about the performance of RTX2050 in deep learning? I'm considering a I7-12700H + RTX2050 laptop
How many gb of ram should the laptop have for neural networks
Lovely video! Can you make a video about Open/Free Website and Database to test/learn Machine Learning Model? Please...
Definitely. Check out my shorts for sme inspiration on that already!
@JesperDramsch I'm just starting out with some ML/AI projects, and I already know I'll be working with fairly large datasets. I'd be interested in hearing your opinion on whether it would make sense to use my current Mac Pro with these specs by installing an Nvidia RTX 4090 and running Linux:
3.3 GHz 12-Core Intel Xeon W
256 GB 2933 MHz DDR4
Or do you think it would be more efficient to build a separate setup? Obviously, building a separate setup would cost me more since I work with Macs and I wouldn’t want to sell my current Mac. What do you think?
Hi Jesper and tank you for your video, informative as usual. I'd like to ask you what you think of a laptop with a Ryzen 7 5825u and no GPU, but with the intention of connecting a "prosthetic" desktop GPU in the future through a pcie connector; I'm talking about something like the EXP GDC "THE BEAST". Or do you think it's easier and better just using an external GPU through thunderbolt 4?
Thanks for your recommendation with that cloud you suggested. I will try that!
Though, I write deep learning networks also for my own projects in computer vision and would prefer not to use a cloud. Also because that requires internet access and I also enjoy working at places where I sometimes don’t have internet. Therefore, I’d appreciate a video about good laptop options for that purpose. P.S. It needs to be compatible w/ Fedora or Ubuntu distro
Do you have power where you don't have internet?
@@JesperDramsch to work for 2h outside it’s sufficient when my laptop is fully charged. Obviously, I will not let a very deep network learn for several hours if I just planned to go outside for 2h. In that time I would either build parts of the architecture or some test to try something out.. some ideas for architectures.. etc.
@@stellamn that makes sense. I'd go for portability anyways then. Build it offline. Make tests offline then train in the cloud later when you're back.
Otherwise see that you get a laptop with a compatible Nvidia card and enough RAM if you still want to train on your laptop.
A pc like that will not be very portable though oftentimes. They get heavy and it can very power draining.
@@JesperDramsch don’t worry, I did already my own research and found some possible solutions. Thanks for your suggestions :) (I didn’t see the prior message)
I'd say that using macBook Air 15" and Ryzen based PC = both at price of single macBook Pro - does offer best value. And as far as upgrade is considered - first, do math Ph. D.
I tend to buy a laptop with ryzen 7 pro 6850U. Is AMD good for who begin learning machine learning ? Hope your response.
What's your take on new macbook pro models, am planning to get the 14 inch base model, will that be enough, should I upgrade anything in it like Ram, more CPU and GPU cores?
What is your main use case? NLP, computer vision, tabular data?
@@JesperDramsch I'm looking at picking up the new Mac book pros and my main task is learning ML and DL - I'm interested in computer vision object recognition, speech to text and NLP
what would you recommend?
could you help me , ERAZER MEDION laptops are good for neural networks?
Is that legion 7 you are using. Im a student and i am planning to get one.
I'm afraid not, sorry.
is it big difference between dedicated or integrated? I want to buy expertbook core i7 -32 RAM BUT it has iris xe .is this gpu enough for my works as you said???
I am working in java, python no-sql & bigdata. And want to learn NLP & deep learning. I'm looking for a durable laptop. My intention to buy a gaming laptop is learn to deep learning only. Not for gaming. I would like to buy Omen Ryzen 5800H with Rtx 3060 6gb or Legion Ryzen 5600H with Rtx 3050 4gb. I have two questions- 1. Will you suggest to buy a gaming laptop to learn ML/DL? 2. Suppose I am working on a java application, a non gpu task on that laptop. Will gpu run all the time? Will it effect the laptop longevity due to heat generation? Please advice.
6Gb and 4GB is very small for modern-day applications, like I said in the video.
Hi I am in the same situation like you - please suggest best laptop. Can I go for gaming laptop.
Which laptop did you buy @@himavanthkumar5269
Great video Jesper. Can I ask u? Wud be better laptop with i7 1260p and 64gb Intel Iris graphics xe or i7 12700h, 32gb, Intel Iris Xe too but having gpu RTX 3070 8gb? Second little more price about 250€. Thx for your help.
I would also like advice on this. Please.
Good price/performance Macs are only base models, if you upgrade just 1 thing it’s gonna cost you a fortune.
This is correct
Dude i dont know is it enough for me? For beginner = rtx3060 , 32gb ram , i7 12th gen , 512gb ssd (Hp Victus 16) . Is it ok for me? What do u think?
That's a pretty beefy PC. I'd hope it works for most consumer models
@@JesperDramsch Will a i5-12450h with a gtx 1650 and 64gb ram be okay?
Is jumping to a 32 gb RAM from 16gb worth it if you are going to be training a lot of Deep Neural nets (mostly Natural Language processing)
Probably. It just means you can fit twice the data in loading/preprocessing. Not having enough RAM can be annoying
@@JesperDramsch thanks 👍
Great and informative video by the way.
If you are buying the new Macs please do another one in depth like this.
I believe people around the world are looking for this kind of content
@@Duge6124 will do!
@@JesperDramsch its weird that your video comes in my recommendation when I just faced the problem of my 32 gb ram not enough.. upgrading to 64 gb now.
@@DaarioNeharis Google knows 😂