事業概要

市場環境

経営資源・競争優位性

成長戦略

2024年6月期第2四半期連結決算

資本政策・配当についての考え方

Introduction

Cybersecurity is important for all types of computing systems, including HPC, but what makes security for HPC different?

In a lot of HPC clusters, there's no kernel access allowed for the users. So how do you get control of a system without kernel access?

In the paper you created on security for NIST you mentioned the four zones of an HPC

Okay. So there are going to be standards or suggested standards for each one of these zones. Has that been defined or is that to be done?

Can an AI application look at the behavior in the cluster and identify where there may be security risks?

Can the AI architectures that are being developed for the enterprise be transferred over for use in HPC?

Will there need to be some security policies with regard to containers?

One of the reasons why Singularity was developed is because Docker can have some security flaws or security vulnerabilities in it. Singularity was designed by Greg Kurtzer to provide security and lock in the user from becoming a bad actor, right?

Do you have ideas about HPC storage and the security involved or how security might be designed for storage?

Where's this going to go? Are there going to be conferences or work streams to define policy around security for HPC? We need more granularity in policies around the four different zones. Will more users collaborate in the community to give their feedback?

What is COMBIME?

Parallel Applications

Login to HPC system

Welcome message

Change Directory

Examples of UNIX commands

Batch systems

Loading a specific Python version

Background execution

MPI Compiler

Overall Design Challenge

High Performance Big Data

Benchmarks

Hadoop 2 Point X Distribution

Hbase

Hdfs Design

Tachyon

Common Benchmarks in the Spark

Distributed Deep Learning Toolkit

Challenges in Accelerating the Big Data

What is HPC?

History of HPC

Computer clusters

Distributed Memory

Distributed memory computers and message passing

Summary

Types of HPC solutions

Canonical solutions for HPC

Conclusion

Intro

Table of contents

What is high performance computing (HPC)?

Why use HPC/HPC Challenges

How does it work?

How to build an HPC environment on Google Cloud?

Security

Use cases

What is HPC?

What are supercomputers used for?

How big can a supercomputer be?

6400 Compute Nodes

How does HPC work?

What is a scheduler or workload manager?

What is Alto Accelerator?

How jobs fit together on the supercomputer

What is Altair Access?

Future of HPC

Complete Data Lake Management Platform

Powerful Data Engines Automatic Parallel Work Stream

Interfaces

Tracks Compliance and Data Lineage for GDPR and CCPA

High Performance and Scalable Data Delivery

Easily Integrate End User BI Tools and Workflows

Machine Learning Library

Deep Learning Support Built on top of Tensorflow

WAYNE ECKERLE VP of Research and Technology Cummins

MIKE HUGHES Modeling & Simulation Services Manager Cummins

DAN LAUER Engineering Simulation Advisor Cummins

Traditional Hpc

Life Sciences

Power Scale Platform

Latest with Power Scale

Highest Performance Powerscale F900 Node

Nfs Over Rdma

Power Scale

Power Scale Integrations

Simplified Management