Direct Storage GPU Solution Bundle.
Featuring WekaFS, high-performing Supermicro Twin Servers and the latest processors.
SOLUTION BRIEF
Equus Computes’ Direct Storage GPU Solution Bundle
We take a consultative approach to solving your storage infrastructure challenges. With a deep understanding of translating business to knowledge of the latest technology and workloads, we prescribe complete end-to-end solutions that solve your problems.
The Challenge
Designing and implementing Artificial Intelligence (AI) and High Performance Computing (HPC) infrastructure can be time consuming and daunting as workload demands resemble high-performance computing compared with traditional data center applications. These workloads benefit from large amounts of data from gigabytes to petabytes including a more demanding metadata intensive workloads than traditional HPC systems were designed to serve.
Enhancement and Design
NVIDIA GPUDirect, a Direct Path Between Storage and GPU Memory
NVIDIA GPUDirect enhances data movement and access for NVIDIA data center GPUs. Using GPUDirect, network adapters and storage drives can directly read and write to/from GPU memory, eliminating unnecessary memory copies, decreasing CPU overheads and reducing latency, resulting in significant performance improvements.
The Integration
Weka™ software combined with Supermicro’s servers provide incredible performance, protection, and data management for Deep Learning, High-Performance Compute, and high- throughput low-latency storage workloads. WekaIO removes your computational storage bottlenecks by leveraging the power of NVMe and task-optimized servers, along with software designed for performance, scalability, and flexibility.
The Equus Solution
A complete preconfigured stack that delivers the right price performance orchestrated with our delivery services, so that you get the finished product ready for production.
The combined Supermicro and WekaIOTM solution integrates the latest PCIe Gen4 servers; the latest high performance networking to create a tightly integrated and validated solution for streamlining the most challenging HPC and AI workloads.
New features and benefits include:
- Intel-based systems can now access data faster with twice the NVMe Storage & I/O utilizing PCIe 4.0
- Up to 40 cores per socket, higher instructions per clock, and two 512-bit FMA units
- Up to 20% power efficiency
We put it all together with our integration process.
Design
Review RequirementsApplication AnalysisPower BudgetBOM CreationRack LayoutCabling DiagramNetwork Design ReviewAssemble
Material Planning Node AssemblyRack & StackNetwork & PowerCable & LabelSubcomponentSerialization CaptureConfigure
Bios SettingFirmware ManagementProvision Switch & IP AddressOS, Customer Image & Software Load
Test
Component PretestingVendor InteroperabilityFull Rack Burn-IN & QA Acceptance TestingFull Rack Test Report & Audit ReportEnd-To-End Performance ReportLogistics
Asset Labeling & DocumentationPrepack AssessmentCrating or KittingSpecialty Packaging (Air-Ride, Anti Tip) White Glove ServicesInstall
Optional Site-Survey & AssessmentOnsite Installation Manual CreationInstallation (Global Smart Hands) Onsite Troubleshooting and Diagnosis The Equus End-to-End Management Advantage
Equus Compute Solutions is an end-to-end Nvidia solution provider; a leading systems integrator for more than 30 years and has delivered more than 3.5 million custom- configured computer systems worldwide. We are developing smarter solutions to increase performance for better outcomes.
We take a consultative approach to solving your storage infrastructure challenges. With a deep understanding of translating business to knowledge of the latest technology and workloads, we prescribe complete end-to-end solutions that solve your problems.
Program Management
Rack Integration
Warehousing
Logistics & Deployment
Smart Hands
