Big data processing machine (can be expanded up to RTX6000 Ada x4) | Information dissemination media for research and development TEGAKARI

Case No.PC-11075A customer who saw this asked us about a machine for processing big data.
It is expected to be used for big data processing, MCMC processing, image processing and natural language processing using Python (PyTorch, Tensorflow), etc.

As a requirement for the specifications, we also received a request to install four NVIDIA RTX200 Ada, as it is okay to have a configuration that assumes a 6000V power supply environment.
Requests for specific specifications are as follows.

・GPU: NVIDIA RTX6000 Ada x4
・OS: Ubuntu 22.04
・Power supply: 200V environment is also possible
・Budget: Within 500 million yen

Based on the conditions you contacted us, we proposed the following configuration.

CPU	Intel Xeon Silver 4410 (2.00GHz 12 cores) x2
memory	256GB REG ECC
storage	960GB SSD S-ATA
video	NVIDIA RTX6000 Ada 48GB x2 (can be expanded up to 4 in total)
network	on board (10GBase-T x2)
Housing + power supply	Full tower type chassis + redundant power supply 1800W
OS	Ubuntu 22.04

We propose a configuration that meets the customer's requirements.

NVIDIA RTX6000 Ada is an expensive product for workstation video cards, so if you have four RTX6000 Adas, it will greatly exceed your budget.
Therefore, it is configured with only two pieces that can be installed within the desired budget range.

The configuration itself has been selected so that up to 4 NVIDIA RTX6000 Ada can be installed, so it is possible to add RTX6000 Ada in future upgrades.

If four RTX6000 Ada are installed, the TDP will exceed 4W with just the GPU.
Therefore, this configuration assumes operation in a 200V environment.

Regarding the CPU, we have selected a 4-CPU configuration equipped with the 4410th generation Xeon Scalable series Intel Xeon Silver 2 to maximize the number of cores within your budget.

Regarding memory, when using CUDA, it is not preferable that the amount of memory installed is less than the amount of VRAM installed.This is because when transferring data to VRAM for processing, it goes through memory once.
Therefore, considering that four RTX6000 Ada 48GB (total VRAM: 4GB) will be installed in the future, the amount of memory installed is 192GB.

The configuration of this case study is based on the conditions given by the customer.
We will flexibly propose machines according to your conditions, so please feel free to contact us even if you are considering different conditions than what is listed.

■ Keywords

・What is the MCMC method?

The MCMC (Markov Chain Monte Carlo) method is a statistical method for generating samples from probability distributions. MCMC can handle complex probability distributions and has applications in many fields such as Bayesian statistics and statistical physics.Specific algorithms include Metropolis method, Hamiltonian Monte Carlo method, Gibbs sampling, and the like.

・What is Python?
Python is an object-oriented programming language copyrighted by the Python Software Foundation (PSF).Its programming syntax is simple, making it highly readable, and it also features a wide variety of components, such as libraries and frameworks, that are suitable for different purposes.A popular language for programming beginners to advanced users.

reference:Python *Jumps to an external site

reference:[Feature Article] Programming language Python Why is it so popular? --Tools to accelerate Python programming * Jump to our owned media "TEGAKARI"

・What is the RTX 6000 Ada?
The RTX6000 Ada is a video card for AI computing released by NVIDIA and has the following features.

・Equipped with hardware functions specialized for AI workloads
・Equipped with 568 Tensor Cores, enabling high-speed deep learning operations
・Equipped with 48GB of GDDR6 memory, capable of handling large data sets
・Highly compatible with CUDA and cuDNN, NVIDIA's software development kits
・Supports virtualization technology and can run AI workloads on multiple virtual machines