AI service machine | Information dissemination media for research and development TEGAKARI

Case No.PC 10880We received a request for a PC for conducting Deep Learning experiments from a customer viewing this.
The assumed conditions are as follows.

・CPU: Core i9
・ Memory: 128GB
・Storage 1: 1TB SSD
・Storage 2: 2TB HDD
・GPU: RTX 4090 x 2 units
・OS: Ubuntu 20.04

・Software used: R, Rstudio, Python, BayoLinkS
・Budget: Within 100 million yen Even if it is difficult, I would like to know how much it will cost
・Other: Use PyTorch 1.9 or 2.0

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

【Main Specifications】

CPU	Xeon Silver 4314 (2.40GHz 16 cores)
memory	128GB
Storage 1	1TB SSD S-ATA
Storage 2	4TB HDD S-ATA
video	NVIDIA Geforce RTX4090 x2
network	on board (1GbE x1 10GBase-T x1)
Housing + power supply	Tower type housing + 1600W
OS	ubuntu20.04
Others	200V compatible cable

It is a custom configuration based on the reference example according to your budget.

It is a configuration with two RTX4090s.Unlike the PC-2 you saw, it works with a single power supply unit.However, we recommend using a 10880V power supply, as 1V power supply will not have sufficient power supply capacity during full operation.

Also, in order to install two RTX4090s, it is necessary to prepare two PCI-E x2 slots. If the CPU is Core i16, it is difficult to secure the required number of PCI-E lanes, so Xeon specifications are used.

The configuration of this case study is based on the conditions given by the customer.
Please feel free to contact us even if you are considering different conditions from what is posted.

■FAQ

・What is PCI Express (PCI-E/PCIe)?
PCI Express is an interface standard for high-speed data transfer in systems such as PCs and servers.Data is sent and received in units called lanes. PCI-E consists of one differential pair for transmission and one differential pair for reception. The transfer speed per lane varies depending on the version, but PCI-E 1 can transfer 1 Gbps per lane.
Also, PCI-E x1 slots have one lane and PCI-E x1 slots have four lanes.Therefore, a PCI-E x4 slot has four times the transfer speed of a PCI-E x4 slot.
Video cards require fast data transfer, so they are usually plugged into a PCI-E x16 slot, which has the maximum number of lanes.

・What is Deep Learning?
DeepLearning is a type of machine learning that uses multilayer neural networks to perform advanced pattern recognition and prediction.Since it generally requires a large amount of data, it is considered to be an effective method when data are abundant.
DeepLearning is also widely used in fields such as image recognition, speech recognition, and natural language processing.Because it can learn complex features and relationships, it can achieve higher accuracy than traditional machine learning methods.

reference:[Special article] What is machine learning? * Jump to our owned media "TEGAKARI"

・What is PyTorch?
PyTorch is an open source machine learning library written in Python.It aims to enable fast and flexible development in the construction and training of neural networks.In addition, it is possible to handle multidimensional arrays called Tensors, and it is possible to perform high-speed calculations using GPUs.
In addition, it supports many model architectures, allowing you to easily build various kinds of neural networks.In addition, since it supports GPUs, it is possible to efficiently perform training using large-scale datasets.

reference:PyTorch *Jumps to an external site