Leveraging Dataiku DSS with Ryzen 9 Laptop: Exploring Possibilities and Performance
Hello forum members,
I would like to initiate a discussion regarding the utilization of Dataiku DSS, the collaborative data science platform, in conjunction with a Ryzen 9 laptop. Let's explore the potential benefits, performance considerations, and experiences of using these two together. Here are the details:
Laptop Specifications:
Processor: Ryzen 9
RAM: 8 GB Ram
Graphics Card: nvidia
Operating System Window 10
Objective:
The aim of this forum thread is to gather insights from the community regarding the compatibility and performance of Dataiku DSS on a Ryzen 9 laptop. We seek to explore the following aspects:
Performance Optimization: How does Dataiku DSS leverage the capabilities of a Ryzen 9 processor? Are there any specific configurations or settings that can be adjusted to optimize the platform's performance on this hardware?
Resource Utilization: Does Dataiku DSS effectively utilize the multithreading capabilities of a Ryzen 9 processor? How efficiently does it make use of the available CPU cores during data processing, model training, and other resource-intensive tasks?
RAM Considerations: Considering the memory requirements of Dataiku DSS, what are the recommendations for an optimal RAM configuration on a Ryzen 9 laptop? Are there any practical limitations or considerations to keep in mind?
User Experiences: If any forum members have experience using Dataiku DSS on a Ryzen 9 laptop, please share your insights, feedback, and any performance optimizations you have discovered. Additionally, if you have encountered any challenges or limitations specific to this combination, please share them as well.
By pooling our knowledge and experiences, we hope to create a valuable resource for anyone considering or currently using Dataiku DSS on a Ryzen 9 laptop. Let's discuss the possibilities and challenges, share best practices, and help each other make the most of this combination.
Thank you for your contributions and insights!
Operating system used: Window 10
Answers
-
tgb417 Dataiku DSS Core Designer, Dataiku DSS & SQL, Dataiku DSS ML Practitioner, Dataiku DSS Core Concepts, Neuron 2020, Neuron, Registered, Dataiku Frontrunner Awards 2021 Finalist, Neuron 2021, Neuron 2022, Frontrunner 2022 Finalist, Frontrunner 2022 Winner, Dataiku Frontrunner Awards 2021 Participant, Frontrunner 2022 Participant, Neuron 2023 Posts: 1,598 Neuron
Welcome to the Dataiku community, we are so glad you have joined us.
Interesting question. First I will note that the Dataiku team is very clear that configurations like this are not for production use, only test.
https://community.dataiku.com/t5/Setup-Configuration/Using-DSS-on-Windows-Computer-WSL2-Maybe-Or-VM/m-p/17543That said I have gotten value from running DSS on laptops for many years.
The configuration that I used 1-2 years ago now was:
A Ryzen 7 laptop from Dell, it had no name brand GPU to speak of. This was not a gaming laptop, this was a cheep lowish end office laptop.
That said it did have the ability to be upgraded to 24GB Ram. I immediately upgraded it to this Max amount of Ram. Not all of the RAM was available, to the OS and applications, because the main system RAM was shared with the video display. I believe that this memory upgrade was critical to ok performance, and it probably cost less than $100.
At the time there was no way to run Dataiku DSS directly on windows. (This is relatively recent innovation.) The typical way to run DSS, at the time, on windows 10 was to run a VM which itself ran Linux. I knew from past experiences that this would take a bunch of RAM and would run a bit slow. I also knew that accessing the nginx web site running inside the VM can be a bit challenging. I also wanted to run an instance of PostgreSQL on the same computer as an analytic database.
So I ended up jumping to Windows 11 late in it’s development cycle. Before it was finally released. The reason I did this was to run WSL2 (windows subsystem for Linux version 2). Which is way more feature rich that wsl1.
In wsl2 I installed the standard Linux version of Dataiku DSS with few challenges. I used Ubuntu 18.0.4 LTS because at the time Dataiku DSS was known to have problems with 20.0.4 LTS and there was still good support expected for a couple of years for 18.0.4.
I installed PostgreSQL in Ubuntu in WSL2. Setup a Dataiku connection as usual and Dataiku could get to the database.I used the browser in windows to access the DSS running inside WSL2. I was never able to access the PostgreSQL database from Windows. Likely because PostgreSQL defaults to a configuration that does not allow remote access.
Once setup things ran fairly smoothly. I would log into my wls2 account and had a Linux shell script that would startup Postgres’s and DSS.
this left me with a reasonable amount of RAM in which to run DSS visual models, graphics, data preparation flows and the like. I would run out of RAM memory from time to time when creating larger record linkage models. If I remember correctly these could create python processes that by themselves could grow to 12GB, 16GB. Before they would have problems.
So yes something like this is doable. No it is not recommended for anything production. And you should have realistic expectations. I am no longer using this configuration an no longer have access to this computer so there is not a lot more I can say about the configuration. There are other posts here in the Dataiku from that period you may find helpful.
hope this helps a bit.
-
Turribeach Dataiku DSS Core Designer, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 2,112 Neuron
I think if you are serious about using Dataiku you will need to move away from Windows since Dataiku only runs natively in Linux. Yes there are alternatives like VMs but these add overhead and do not support all the features like accessing GPUs directly. But it really depends on what you are tryign to achieve. Personally I think an Apple Silicon Mac will be a better machine to run Dataiku. While also unsupported Dataiku can run directly without a VM. And while it does use the Rosetta translation layer this has proven to be as fast as native x86 CPUs and sometimes even faster.
-
tgb417 Dataiku DSS Core Designer, Dataiku DSS & SQL, Dataiku DSS ML Practitioner, Dataiku DSS Core Concepts, Neuron 2020, Neuron, Registered, Dataiku Frontrunner Awards 2021 Finalist, Neuron 2021, Neuron 2022, Frontrunner 2022 Finalist, Frontrunner 2022 Winner, Dataiku Frontrunner Awards 2021 Participant, Frontrunner 2022 Participant, Neuron 2023 Posts: 1,598 Neuron
I have run DSS on Macs as well initially starting on an older core i7 intel MacBook air. Which worked surprising well.
Currently, I’m using a MacBook Pro 16 with a binned M1 Max chip and 32gb RAM. Other than lack of supported GPU it is working well.
Has anyone figured out how to run GPU or Neural engine based models on Apple Silicon? I note about a year ago that Apple’s Metal Team and PyTorch announced Accelerating PytTorch training on Mac. If this worked this could significantly improve the Apple silicon mobile Dataiku DSS experience. -
tgb417 Dataiku DSS Core Designer, Dataiku DSS & SQL, Dataiku DSS ML Practitioner, Dataiku DSS Core Concepts, Neuron 2020, Neuron, Registered, Dataiku Frontrunner Awards 2021 Finalist, Neuron 2021, Neuron 2022, Frontrunner 2022 Finalist, Frontrunner 2022 Winner, Dataiku Frontrunner Awards 2021 Participant, Frontrunner 2022 Participant, Neuron 2023 Posts: 1,598 Neuron
There was a prior thread like this on the forums. At the time the general idea from the Dataiku team was to run Linux on your laptop.
Here is a link to that thread.