A customer involved in sports and human body structure research asked us about a machine learning machine.
Since we are introducing a machine learning environment for the first time, we would like a starter configuration for the above purpose with a budget of 100 million yen.
Specific uses are planned for validation, transfer learning, and fine tuning of existing models, as well as model learning and estimation, and the amount of data handled is not expected to increase.
Additionally, the notebook PC you are using has a VRAM capacity of 6GB, and we have heard that you are concerned that there is a frequent lack of memory resources and processing stops. For this reason, we have received requests to place emphasis on VRAM capacity for newly introduced machines.
The OS is Ubuntu 22.04, and we have been asked about measures to prevent overheating when cooling the machine.
Based on the content of the consultation, we proposed the following configuration.
| CPU | Core i7-14700K (3.40GHz 8 cores + 2.50GHz 12 cores) |
| memory | 64GB |
| Storage 1 | 1TB M.2 SSD |
| Storage 2 | 4TB HDD S-ATA |
| video | NVIDIA Geforce RTX4090 24GB |
| network | on board(2.5GBase-T x1) Wi-Fi x1 |
| Housing + power supply | Middle tower type housing + 1500W |
| OS | Ubuntu 22.04 |
This configuration is equipped with the latest 2023th generation Core series as of December 12.
GPU selection for machine learning
Assuming that CUDA will be used for machine learning, the GPU is equipped with a high-end model NVIDIA GeForce RTX4090.
NVIDIA Geforce RTX4090 is currently the top model in the GeForce series, and its simple processing performance is among the best among video cards currently on sale. The video memory capacity is 24GB, which is slightly less than high-end models for workstations and servers.
When dealing with huge models such as large-scale language models, there may be a feeling of insufficiency, but for applications that deal with models that are not very large, as in the case of your consultation, you can expect high processing performance.
Also, in terms of budget, the Geforce RTX4090 is the product with the highest specs among the available GPUs.
About machine cooling
The cooling system in this configuration uses air coolers for both the CPU and GPU.
The CPU cooler is a high-end product with a large heatsink, so there will be no noticeable problems when using it with the default settings without manual overclocking.
In addition, at the time of manufacturing, we perform an operation test with full operation for 12 hours and confirm that there are no problems with temperature changes during operation before shipping.
When using the product, special attention must be paid to the room temperature of the installation location.
When we perform operation checks, we use a room temperature of 25°C. Exceeding this room temperature will not immediately cause problems, but if you regularly use it in an environment above 30 degrees Celsius, there is an increased risk of problems occurring due to inadequate cooling of the CPU and GPU.
In addition, if the machine is installed in an environment that prevents heat exhaust from the machine, such as in a place with poor air flow or if there are objects nearby that tend to get hot, there is a risk of problems occurring due to stagnation of exhaust air inside the housing.
We would appreciate it if you could consider the installation layout by creating a certain amount of space on the back side of the machine.
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 CUDA Toolkit?
|
|
■ Click here for details and inquiries about this PC case
* Please enter the name of the case or your desired conditions. |
