IRONBYTE

INTRODUCTION

We propose the creation of a Data Processing Center with a total capacity of at least 2000 PFlops. Using the IRONBYTE architecture for distributed launch and management of AI computing tasks, this data center focuses on ensuring high availability of the core system, responsible for all tasks and data storage orchestration.

Tasks That Can Be Solved

Learning

LLM (Large Language Models) training, fine-tuning, ML (Machine Learning)

Storage

Storage of models, datasets, AI software libraries

Scaling

Running models and forming a pipeline of model inferencing and scaling

Irregular Problems

Tasks can be combined on an IRONBYTE RIG if resources are available. During planned modernization, tasks can be designated for execution on nodes with new architecture accelerators. Legacy code support for inferencing models that remain operational for extended periods is also possible.

Comparison with Alternative Solutions

Type	IRONBYTE RIG	Nvidia server 8xA100	Nvidia server 8xH200
TFlops (FP32)	730	156	536
TFlops (FP8)	6,600	4,992	32,000
RAM GPU (GB)	240	640	1,128
Cost	$40,000	$160,000	$400,000
IRONBYTE RIG efficiency factor in synthetic load	100%	1900%	1400%
Cloud 2000 PFlops (synthetic load)	$115,000,000	$2,153,000,000	$1,567,000,000
Practical effectiveness of IRONBYTE RIG in a synthetic load		Over 40%	Over 60%
Analog of IRONBYTE RIG with consideration of common tasks in LLM	$115,000,000	$2,153,000,000 / $1,292,000,000	$1,567,000,000 / $635,000,000

Efficiency summary

Speed characteristics of inter-node exchange via NVLINK are compensated by task parallelization, based on the correlation separation of the dataset between nodes and the combination of model layers obtained on different nodes.

Memory capacity limitations in RIGs, particularly for larger models, are addressed through various mathematical solutions, demonstrating minimal impact on learning quality.

Efficiency in alternative tasks, in terms of "price-quality" ratio, exceeds 1000%.

The "price-quality" advantage, when comparing identical data centers in all parameters, is more than 2 times.

Efficiency in tasks specialized by competitors is 40/60% respectively.