I want to buy a machine to use my x2 GPUs (6000x RTX A2 or 100x A2).
The desired conditions are as follows.
- Desire for a configuration that operates in a 100V power supply environment
・We want to reduce power consumption as much as possible.
・Because it will be used in a living room, it is desirable to be quiet when not using the GPU, but operating noise when using the GPU is acceptable
・Usage is BERT Fine-tuning and NVIDIA Clara Parabricks
・Budget is 150 million yen
I don't mind if the budget exceeds the fiscal year, but I would be happy if I could get it as soon as possible.
We considered the configuration according to your request.
Considering power consumption, it is configured with Ryzen Threadripper.
The GPU to be installed is assumed to be RTX A6000.
Since the A100 does not have a cooling fan in the GPU itself, a chassis for a GPGPU server with a GPU cooling mechanism is essential, and it may be difficult to achieve within your budget.
The power supply unit is a 1600W compatible product, but up to about 100W can be used in a 1300V environment.Input is 100V/15A AC up to 80V/100A AC, there is a loss when converting to DC, and even TITANIUM with a maximum conversion efficiency of 90+ has a conversion efficiency of 100% at 15% operation. At 1500V/0.9A, the theoretical upper limit is 1350W as XNUMXW x XNUMX.
Based on the above, it is desirable for the CPU side to consume as little power as possible, so we proposed a Threadripper configuration with a single CPU configuration.In addition, since this configuration has a single CPU configuration, there is physically enough space inside the housing, so a total of three RTX A1s can be used when used in a 1V environment.
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.
For customers who want measures against operating noise,SI companyWe also accept proposals in combination with our silent racks.
Please feel free to contact us if you are considering installing a machine that emphasizes quietness.
Features of silent racks manufactured by SI
[1]Provide a specially designed rack that matches the user's environment and machine
[2] High-level compatibility between quietness and safe heat dissipation
[3] Because it is a manufacturer centered on specialized acoustic technology, it has high quietness technology
[4]Promise safe operation with technical service for machine adaptation
■FAQ
・What is BERT?
BERT (Bidirectional Encoder Representations from Transformers) is a natural language processing (NLP) model developed by Google.It can understand words based on a given context and is applied to a wide range of tasks in language processing.
Also, BERT consists of two phases: pre-training and fine-tuning.Pre-training creates a generic language model trained from a large corpus.Fine-tuning adjusts a model learned from a small dataset to apply it to a specific task.
It is characterized by showing higher accuracy than conventional NLP models and being able to handle complex tasks, and is applied to text generation, question answering, document classification, language translation, etc. Widely used as one.
・What is Fine-tuning?
Fine-tuning is the fine-tuning of a pre-trained machine learning model using task-related datasets in order to apply it to a specific task.For example, in the case of natural language processing, general language models trained from large corpora can be applied to specific text classification tasks (sentiment analysis, spam detection, topic classification, etc.).In this case, the model is learned from a small dataset relevant to a specific classification task and ultimately optimized for the specific task.
Fine-tuning allows existing generic models to be applied to specific tasks without the need for large amounts of data.
・What is NVIDIA Clara Parabricks?
NVIDIA Clara Parabricks is GPU-based high-speed genome analysis software.It utilizes GPU acceleration to significantly speed up genomic analysis tasks.
Used for tasks such as sequence alignment, variant calling, and genome assembly, it achieves high processing speed and high accuracy, and can process large amounts of data in a short time.
In addition, since it uses hardware-independent algorithms, it can run on various GPU platforms.
■ This article was posted on January 2014, 1, so the information content may be out of date. "I'm looking for a research PC that has as much memory as possible and is capable of high-speed computation. […see next]
A customer involved in vehicle and logistics research consulted us about a machine for conducting research on seismic motion.In order to perform large-scale data learning at high speed, memory and GPU […see next]
■This article was posted on August 2020, 8, so the information may be outdated. A customer has requested “Dee”, a software that analyzes animal behavior using deep learning. […see next]
