When it comes to choosing the best workstation for AI, there are several key factors to consider, including processing power, memory, storage, and GPU capabilities. While the CPU is still important, it is the GPU which does the heavy lifting.

While we regularly accommodate requests for AI servers with up to 10 GPUs such as the NVIDIA H100, the best workstation for AI is our most frequent request. Our most popular max-power-for-lowest cost option is our a-XP Threadripper Pro Workstation, but if going all out, our a-X2P Dual EPYC Workstation, configurable with an up to eye-watering 6TB of ECC RAM and 4 GPUs, is up to any task you can provide it 24/7.

With our portables, such as our a-X2P Portable Dual EPYC Workstation, cooling plays an even more important role than usual due to form factor. For this reason if you want multiple GPUs in our portables they must be RTX workstation series GPUs, of which the NVIDIA RTX 6000 Ada or NVIDIA RTX A6000 are flagship performers that run cool, too. A portable AI workstation is a specific use case, but very helpful for server room level power on the move.

To help you consider the best workstation for you, here’s a list of the components in our traditional and portable workstations and the role they play:

1. CPU: You want a workstation with a high-performance multi-core processor. Intel Xeon and AMD Ryzen Threadripper series are popular choices, but the AMD EPYC CPUs are the peak performers for thoroughly multi-threaded workloads. The more cores and threads, the better, as AI workloads often benefit from parallel processing.

2. GPU: Graphics Processing Units (GPUs) are crucial for AI tasks, especially deep learning. NVIDIA GPUs, like the GeForce RTX or the data center-focused Tesla series, are far and away the power leaders. You want a GPU with a high number of CUDA cores and ample memory (VRAM) to handle complex computations.

3. Memory: AI workloads often require abundant memory. A good minimum is still 64GB of RAM, but it’s best to consider more if your projects involve massive datasets or deep learning models. Though high-speed RAM can offer slightly better performance, spec speeds for the processors in question are what we always recommend to maximize stability, including the most stable ECC RAM with CPUs that support it.

4. Storage: Solid State Drives (SSDs) are recommended for faster data access. For AI, consider a combination of a fast NVMe SSD for storing the operating system and applications, and a larger HDD or SSD for data storage. Alternatively, consider NVMe RAID setups for even faster storage access.

5. Cooling and power supply: AI workloads can generate significant heat, so ensure your workstation has efficient cooling systems to prevent overheating. A robust power supply unit (PSU) is also necessary to handle the power demands of high-end components.

6. Connectivity: Check for high-speed connectivity options such as Thunderbolt or USB 3.1 Gen 2 for external device connections.

7. Future scalability: Consider a workstation that allows for component upgrades, such as additional RAM slots or PCIe expansion slots for adding more GPUs or specialized AI accelerators.

8. Software compatibility: Ensure that the workstation is compatible with the AI frameworks, libraries, and tools you plan to use, such as TensorFlow, PyTorch, or CUDA.

The best workstation for AI can vary depending on your specific needs and budget. It’s a good idea to consult with us for personalized recommendations based on your requirements if you’re not clear, which we can go over in a 30 or 60 minute consult by going through your workflow in detail.

If you are clear about your needs, and simply want a quote, give us a call or simply email us the specs of the unit you need along with your budget and we will get that quote out to you, typically within 24 hours.