[an error occurred while processing the directive]
RSS
Логотип
Баннер в шапке 1
Баннер в шапке 2
Project

Lenovo completed implementation of the supercomputer in Irkutsk Oil Company

Customers: Irkutsk Oil Company (IOC)

Product: Projects of creation of complex IT infrastructure

Project date: 2019/06  - 2019/12

2019: Project completion on implementation of a HPC-cluster

On January 14, 2020 it became known that the Lenovo company completed the project on implementation of a HPC-cluster of the Irkutsk Oil Company (IOC). A project objective — gain of IT infrastructure for acceleration of carrying out geological researches.

Irkutsk Oil Company — one of independent producers of a hydrocarbonic raw material in Russia. INK group is engaged in geological studying, exploration and production of a hydrocarbonic raw material on fields and subsoil plots in Eastern Siberia — in the Irkutsk region, the Sakha (Yakutia) Republic and Krasnoyarsk Krai.

Before project implementation specialists of INK used graphic stations at which they carried out engineering-geological calculations in work.

However graphic stations have the restrictions on the volume of calculations using digital models of fields connected with their performance. Use of these solutions was inefficient in terms of costs of time which is necessary for accomplishment of calculations at the certain station. Difficult models were calculated with a large number of cells long that resulted in low efficiency of acceptance of business solutions of high cost.

There was a need for an integrated geographically distributed system. Specialists of Lenovo suggested to connect graphic workstations of each object to a uniform HPC-cluster in which there will be a processing of huge data arrays. Communication between Lenovo and INK, logistics, delivery and installation of equipment was performed by the long-term partner of Lenovo — the Siberian center of information technologies (ISIB). After that specialists of Lenovo carried out setup of the equipment and tested it together with INK.

The cluster Lenovo Scalable Infrastructure computer system is equipped with 6 computing nodes with hybrid architecture of GPGPU (General-purpose computing on graphics processing units), 120 computing cores of CPU, 3,840 — CUDA, 30,720 — Tensor. Total capacitance of RAM of TruDDR4 2666MHz — 1,152 GB, and RAM on a core of CPU — 9.6 GB.

Thanks to implementation of the supercomputer Irkutsk Oil Company significantly increased computing powers. Specialists of INK note increase in productivity of separate models to 7 times. Besides, the HPC-cluster provided high flexibility and scalability of a system and also considerably simplified service of 19 fields.

File:Aquote1.png
to Each our customer we aim to provide the most optimal and effective solution, considering at the same time specifics of its activity and tasks. Together with INK we managed to construct a HPC-cluster,

which not only considerably accelerated process of calculations of digital models of fields, but also lifted limits on their dimension,

File:Aquote2.png

File:Aquote1.png
After implementation, start and setup we saw significant increase in speed. Some models received almost linear acceleration — their performance increased up to 7 times,
tells Alexander Ovchinnikov, the chief specialist on development of Irkutsk Oil Company
File:Aquote2.png

Specialists of INK summed up the main results of implementation of HPC:

  • There was almost linear acceleration of difficult models.
  • It was succeeded to release capacities of the graphic station (before implementation of the supercomputer use of graphic stations during calculations was difficult).
  • There was an opportunity to make multiple calculations — process accelerated more than by 6 times.
  • Employees without graphic stations had an opportunity to make high-performance calculations.

Now depending on complexity of a task specialists can use resources of both a separate node, and the whole cluster. Therefore, by means of HPC it was succeeded to lift limits on dimension of digital models.