TSI's HPCaaS featuring GigaIO's FabreX and TSI's CloudShell facilitating true Rack Scale Design - see below for a Super Computer Flyaway Kit for AI and Edge Computing mobile needs!
How to optimize your HPC delivery with 'Recomposable' pools of resources and self service
Use GigaIO's PCIe based network fabric to accelerate your HPC performance
Dynamically schedule, build and deploy your HPC resources with a self service portal
How to optimize your HPC delivery with 'Recomposable' pools of resources and self service
-
Easily share expensive resources between multiple servers, eliminating overprovisioning and reducing Capex
​
-
Precisely tune CPU/GPU/Storage ratios for each application
-
Lower Capex resulting from independent upgrade cycles for each compute/storage item
​​
-
Lower Opex with reduced power consumption and automated management​
-
Strips away conversion and overhead layer with 100% PCI Express network
​​
-
Full end-to-end latency below 200 nanoseconds
​​
-
PCIe Gen 4 Throughput of 256Gbit/sec (half duplex), now 512Gbit/sec in PCIe Gen 5
​
-
100% PCI SIG Compliant
-
Open-Source APIs
-
Forward and backward compatible across server generations
​
-
Easily upgrade accelerators and storage without changing the entire rack of servers and software
-
Add or upgrade only what you need and when you need it
-
Benefit from industry-wide acceptance, the huge variety of choice, interoperability, and innovation in the PCIe ecosystem
-
Greater agility in application development, provisioning and life cycle management
-
Self Service portal for
HPC Infrastructure
LaaS, IaaS, PaaS,
DevOps, Cloud ready
Automation
-
Vendor, Service, Cloud,
Application and Tool
Agnostic
-
Enterprise level control of user base
​
-
Open Source Automation Model support
-
Support both onpremise HPC and Cloud based HPC from single plane of glass
​