Deep learning machine with GeForce RTX 4090

A customer involved in speech synthesis research consulted us about a machine for machine learning of spoken language.

The budget is about 60 yen to 70 yen, and the desired specifications are as follows.

・Memory: 64GB or more
・Storage: 1TB SSD or more
・Video card: NVIDIA RTX3090 equivalent
・OS: Ubuntu (Please install familiar Ubuntu 18.04)

 

At the time of consultation, the impact of the exchange rate increase was significant, and it was difficult to propose a configuration that supports multiple GPUs within the budget, so we prepared a configuration that supports 1 GPU. It will be a configuration that does not support the addition of a second GPU.

【Main Specifications】

CPU AMD Ryzen5 7600X (4.70GHz 6 cores)
memory 64GB
storage 1TB SSD S-ATA
video NVIDIA Geforce RTX4090 (internal exhaust)
network on board (2.5GBase-T x1)
Housing + power supply Middle tower case + 1000W
OS Ubuntu 18.04

In addition, regarding the installation of Ubuntu 18.04, we have provided the following information and are considering the policy.

■ Points
・Ubuntu 18.04 is an older version, and operation has not been confirmed with the latest hardware configuration, so it is difficult to guarantee operation.

・There is no problem in trying to install Ubuntu 18.04.

・How about shipping with Ubuntu 18.04 if there is no fatal hardware recognition error, and changing to 20.04 or 22.04 if an error occurs and cannot be resolved?

・When Ubuntu 18.04 is installed, there is a possibility that an error may occur due to the low power consumption of LAN peripherals, CPU, and board systems.

From the customer "If you try installing Ubuntu 18.04 and have problems, feel free to change to 20.04 or 22.04”, so in this example, we have chosen an OS that meets your needs.

In addition, the GPU uses the new RTX4090.
Due to the large physical size and heat generation of the RTX4090, it is necessary to pay close attention to the airflow inside the housing.This example is a hardware configuration that supports up to 1 GPU, so it also supports the internal exhaust type RTX4090.Therefore, if there are no restrictions on the GPU on the software you use, you can expect higher processing performance than the old product RTX3090.

 

 

■FAQ

・What is deep learning?
Deep learning, or Deep Learning, is a machine learning technique in which computers autonomously make predictions and classifications by learning data.
For machine learning, please see related articles on our owned media "TEGAKARI".