Forum Discussion
PC advise Specification for Developing Dashboards for "Big Data" multiple DAX codes
When working with advanced DAX codes in tools like Power BI, a high-performance machine is crucial for smooth data processing and report development. Here's an updated list of ideal PC specs for developing reports for Big Data with high DAX codes:
Less than 500,000 rows:
- RAM Requirement: 64GB RAM
- Processor Core Requirement: Octa-core (e.g., Intel Core i7 or AMD Ryzen 7)
- Processor Clock Speed: 3.0GHz or higher
- Storage: 1TB NVMe SSD
500,000 to 1 million rows:
- RAM Requirement: 128GB RAM
- Processor Core Requirement: Dodeca-core (e.g., Intel Core i9 or AMD Ryzen 9)
- Processor Clock Speed: 3.5GHz or higher
- Storage: 2TB NVMe SSD
1 million to 5 million rows:
- RAM Requirement: 256GB RAM
- Processor Core Requirement: Hexadeca-core (e.g., Intel Xeon or AMD Ryzen Threadripper)
- Processor Clock Speed: 4.0GHz or higher
- Storage: 4TB NVMe SSD
5 million to 10 million rows:
- RAM Requirement: 512GB RAM
- Processor Core Requirement: Octadeca-core (e.g., AMD EPYC or Intel Xeon Scalable)
- Processor Clock Speed: 4.5GHz or higher
- Storage: 8TB NVMe SSD
Above 10 million rows:
- RAM Requirement: 1TB RAM
- Processor Core Requirement: Multi-socket server-grade processors
- Processor Clock Speed: 5.0GHz or higher
- Storage: 16TB NVMe SSD in RAID configuration
These specifications are tailored to handle the demands of developing reports for Big Data with high DAX codes in tools like Power BI. The combination of high RAM, powerful multi-core processors, fast clock speeds, and ample storage capacity ensures that your machine can efficiently process complex data transformations and calculations, resulting in optimized performance when working with advanced DAX codes.
Hi PowerBI_LOVER,
Even if you go for the service the refresh of 100m rows will take longer than 30 minutes specially if you are trying to do a lot of transformations and tables using dax.
If you are trying to load 100m rows to desktop has I referred previously has a best practice you should not do that.
The dax measures have no impact in terms of refresh since they are only calculated at the time you called them on your visuals.
Altough I understand what you want to achieve with this type of configuration large semantic model should be setup with only part of the data and then push the refresh to the service.
On top of this having this large semântica models you also should consider the usage of incremental refresh or aggregation tables, this will reduce the refresh time but also the performance when building your reports.
7 Replies
Hi PowerBI_LOVER ,
Not sure where you got this setup and number of rows comparision, but I can tell you from experience that I have worked with models with up to 100 Millions rows with a computer with 32GB of ram and an I7.
This is not only dependent on the computer performance, but also on the way you built your model and the calculations you do.
Off course that if you are working with millions of lines the loading time can take some time but not the dax.
However has a best practice if you have models with 10M rows I would not load everything into Power BI I would use some paremeters to crop the data and then on the service would do the full refresh.
For the DAX measures depending on what you are doing I also suggest to use external tools (tabular editor) to avoid the waiting time when you do the OK on the dax formula bar.
Hi Miguel Félix,
100 Millions rows with a computer with 32GB of ram and an I7 - If you have only dataset you are viewing maybe.
I have used 32GB on I5 and it can with dataset viewing.
I am taking about 600 lines of dax - measures and table codes.
32GB will refresh for over 30mins.
So my configuration above will work or alternatives to my configuration for large scale Dataset and Multiple Dax codes.
Thanks
Hi PowerBI_LOVER,
Even if you go for the service the refresh of 100m rows will take longer than 30 minutes specially if you are trying to do a lot of transformations and tables using dax.
If you are trying to load 100m rows to desktop has I referred previously has a best practice you should not do that.
The dax measures have no impact in terms of refresh since they are only calculated at the time you called them on your visuals.
Altough I understand what you want to achieve with this type of configuration large semantic model should be setup with only part of the data and then push the refresh to the service.
On top of this having this large semântica models you also should consider the usage of incremental refresh or aggregation tables, this will reduce the refresh time but also the performance when building your reports.
Hi PowerBI_LOVER,
We sincerely appreciate you for taking the time to share such detailed specifications and insights with the Microsoft Fabric Community. Contributions like yours are immensely valuable in assisting fellow users to comprehend and evaluate the hardware requirements for large-scale Power BI and DAX-based development environments.
We would also like to extend our gratitude to MFelix for your thoughtful and experience-based response. Your valuable insights on model optimisation, performance strategies, and the significance of DAX in enhancing report responsiveness greatly enrich this discussion.
Thank you.
Hi PowerBI_LOVER,
Thank you for initiating this important discussion. Both perspectives shared here contribute towards a comprehensive understanding of Power BI performance tuning, ranging from hardware provisioning to best practices in semantic modelling.
To enhance the visibility of this discussion within the forum, we kindly request you to mark the appropriate response as the accepted solution and extend kudos. This will greatly assist other members, who may have similar queries, in locating the relevant information more easily.
Thank you.Hi PowerBI_LOVER,
Thank you for initiating this valuable discussion. The perspectives shared here collectively provide a well-rounded understanding of Power BI performance tuning, covering aspects from hardware provisioning to best practices in semantic modelling.
To improve the visibility of this discussion within the forum, we kindly request you to mark the suitable response as the accepted solution and also extend kudos. This will greatly help other members with similar queries to identify the relevant information more conveniently.
Thank you.Hi PowerBI_LOVER,
Thank you for initiating this important discussion. Both perspectives shared here contribute towards a comprehensive understanding of Power BI performance tuning, ranging from hardware provisioning to best practices in semantic modelling.
To enhance the visibility of this discussion within the forum, we kindly request you to mark the appropriate response as the accepted solution and extend kudos. This will greatly assist other members, who may have similar queries, in locating the relevant information more easily.
Thank you.
