shutterstock_1162236571.jpg

HIGH PERFORMANCE COMPUTING SERVICES

Corvid delivers efficiency, low cost and high performance HPC.

Gray_metallic_background_edited.jpg

Our focus on efficiency, low cost, and high performance as seen by the implementation of the company’s own supercomputer, makes Corvid one of the top contenders in the high-performance computing industry.  Contact us for a quote tailored specifically to fit your needs at competitive price. 

  • Highly competitive prices for compute time on our water and air-cooled HPC systems​

  • Redundant clustered storage and daily incremental backups as additional add-ons for data generated using our high-performance computer​

  • Customer support for 3rd party solver integration on our HPC systems​

  • Customized HPC hardware and software recommendations based on customer needs, applications, and budget​

  • Specialize in HPC cluster configuration consisting of InfiniBand fabric, scheduler, provisioner, and nodes​ 

  • Upgrades of cluster provisioner and scheduler upon new releases with minimal downtime​ 

  • Develop and troubleshoot custom job submit and batch scripts for various solvers, both in-house and 3rd party (ex. ANSYS, RAVEN, VELODYNE, Fluent, CFX, CTH, Taos, Star-CCM+, ALE3D, etc.)​ 

  • Compile optimized software with proprietary compliers (ex. Intel compilers, Portland)​ 

  • Troubleshoot slow simulation performance, failed/dropped/hung jobs, and network issues on HPC cluster​

  • Configure HPC cluster security through network segmentation, job access restriction, and system hardening​ 

  • Consult customers on HPC best practices, common problems, and implementations 

DETAILS

70% trans blue forward.png

HPC HARDWARE SPECIFICATIONS 

  • Dual 10GBE LAN Interfaces Per Node - Intel X550  

  • 192GB ECC DDR4 REG 2933MHz, 12 x 8GB, Per Node

  • Mellanox EDR100GB IB ConnectX(5)

  • Intel 6242 Cascade Lake Processors 16C at 2.8GHz, 22MB L3 Cache

  • Custom 4 Nodes Per 2U Platform- 128 Cores Per 2U Total - Intel C621 Chipset

SUPERCOMPUTER STATS 

  • 10-12x Faster Than Typical Cloud Computing Implementations  

  • 0.6 Microsecond Latency and 100G/s Bandwidth 

  • 4.6 Petabytes of Active Storage 

  • ​340 Terabytes of RAM  

  • 20 Racks

  • Over 50,000 Processor Cores