Optimizing JMP Performance on an AMD Ryzen Laptop

[#1]

Hello JMP Community,

I wanted to share my positive experience and seek further insights into optimizing JMP software on my new setup, especially when handling large datasets and running complex analyses.

I recently upgraded to a new laptop, and so far, I’ve been impressed by its performance. With its powerful processing cores, it has smoothly handled most tasks. However, I’ve noticed that when running more intensive analyses, such as large data imports or complex regression models, there is a slight lag in performance.

I’m using a laptop with 16 GB of RAM, which I expected to be ideal for running JMP. For the most part, it has worked very well, handling basic data manipulation and analysis seamlessly. However, for more demanding tasks like time-series forecasting or running advanced scripts, I occasionally notice a slight delay.

I’ve recently switched to an AMD Ryzen laptop with a Ryzen 7 processor, and the results have been generally excellent. This laptop offers great multitasking capabilities, and I’ve found it to be highly responsive, even when working with large datasets. It seems that with a bit of fine-tuning in terms of memory allocation and closing unnecessary background apps, the performance can be even smoother.

If anyone has experience optimizing JMP on AMD Ryzen laptops, I’d love to hear your suggestions! Any tweaks or settings you’ve used to ensure optimal performance would be greatly appreciated.

Additionally, I’m curious about how JMP utilizes multi-threading and GPU acceleration on AMD systems. Any recommendations on maximizing these features would be incredibly helpful.

Looking forward to your thoughts and suggestions!

[#2]

If you can add more memory, add it, max it out if possible.  

As data continues to grow, the need for RAM grows.  Remember that JMP does everything in memory, so memory is the precious commodity, often more so than processor speed.  Large data can lead to enormous matrix calculations which require RAM.

At present on Windows, only the Torch and XGBoost add-ins utilize GPUs, but I believe that is limited to CUDA which would be Nvidia GPUs.  You can utilize OpenCL which is supported on AMD GPUs and Nvidia GPUs through Python by installing pyopencl.  Pyopencl will let you write GPU programs that you can interact with JMP through the Python Integration.  This is possible even with JMP 14-17, but is a much better experience JMP 18 and beyond. 

JMP does take advantage of multi-threading on CPU cores. But at present does not utilize GPU acceleration except for the 3D graphics and the aforementioned add-ins.