Skip to main content
Optiscale
Log inContact us
Solutions
HPC Cluster Design · CPU supercomputersAI Supercomputer · GPU training & inference clustersParallel Storage · Spectrum Scale · Lustre · BeeGFS · VASTInterconnect · InfiniBand · RoCE fabricsSchedulers · Slurm · KubernetesApplication Workload · Reference diagrams by workloadSizing Calculator · 5-minute sizing + 5-year TCO
Products
Luxe Vision · Unified monitoring across heterogeneous resourcesLuxe Ray · Heterogeneous resource operations & unified controlLuxe Vantage · Usage & billing built on the AI/HPC schedulerLuxe Orbit · Heterogeneous software provisioning & parallel managementLuxe Series Overview · The unified 4-layer storyRequest a Live Demo · Vision · Ray · VantageRequest a Closed PoC · Orbit — 1:1 environment setup
Services
Architecture Consulting · Workload definition → specification designImplementation & PM · Vendor integration & project managementPerformance Tuning · Architecture & configuration optimization plus code-level performance gainsManaged Operations · Multi-year operations contracts
Resources
TCO Calculator · On-premises vs. the three major cloudsSelf-Check · 8-question workload assessmentApplication Workload Library · 6 workload diagramsWhite Papers · In-depth technical white papersBlog · Tech Notes · Engineering blogNewsletter · Biweekly infrastructure updates

Resources · RFP Template

HPC / AI Supercomputer RFP Template

Specification, verification, and evaluation standards ready to use within 30 days of issuing your RFP. It includes item-by-item guidance and benchmark acceptance criteria so that buyers can fill it in as-is.

v0.9·12 pages·Last updated 2026-04-22

Table of contents preview

  1. 01.

    Project Background and Objectives

    The buyer's rationale for adopting the infrastructure (workload types, scale, timeline) — focused on objective facts.

    • Primary workloads
    • Expected user scale
    • Operations start date
    • Operations term (3/5/7 years)
  2. 02.

    Scope of Work

    Definition of the consulting + deployment + operations + training phases. Specify the deliverables for each phase.

    • Compute / storage / network specifications
    • Management server, OS, and scheduler scope
    • Benchmark verification targets
    • Operations support (SLA and response times)
  3. 03.

    Technical Requirements — Compute

    CPU / GPU specifications, node count, memory, local storage. Measurable criteria (HPL, STREAM, NCCL).

    • CPU model / core count / node count
    • GPU model / NVLink topology
    • Memory per node / memory bandwidth
  4. 04.

    Technical Requirements — Network

    InfiniBand XDR/NDR vs. RoCEv2 decision. Fat-tree blocking ratio, adaptive routing.

    • Fabric type / speed
    • Topology (fat-tree / dragonfly / other)
    • Blocking ratio (1:1 / 2:1 / etc.)
    • Whether the management network is separated
  5. 05.

    Technical Requirements — Storage

    Parallel file system, metadata distribution, IOPS, backup and DR.

    • File system (Spectrum Scale / Lustre)
    • Capacity and throughput (GB/s sequential, metadata ops/s)
    • Tiers (Hot/Warm/Cold)
    • Backup and disaster recovery RPO/RTO
  6. 06.

    Software / Operations Tools

    Scheduler (Slurm/K8s), monitoring, chargeback, OS and image automation.

    • Slurm / Kubernetes / hybrid
    • Monitoring and log integration
    • User analytics / billing
    • OS deployment and provisioning (redundancy)
  7. 07.

    Acceptance Criteria

    Agree on benchmark items, conditions, and pass criteria in advance. The proposer submits the measurement tools and scripts.

    • HPL efficiency ≥ 95% theoretical
    • STREAM Triad ≥ 380 GB/s/node
    • NCCL all-reduce ≥ 94% peak
    • Spectrum Scale throughput ≥ X GB/s
  8. 08.

    Proposal Evaluation Criteria and Weights

    A standard model of 60% technical / 30% price / 10% company credibility (adjustable).

    • Technical evaluation detail (around 40 items)
    • Pricing methodology (5-year TCO)
    • Reference evaluation (similar-scale cases)
    • Review panel composition (including external experts)
  9. 09.

    Contract Terms and Schedule

    Contract type, payment terms, defect warranty, SLA, and penalties.

  10. 10.

    Submission Requirements (Proposer)

    Standard format for attachments beyond the main body (document template, PDF page count, quotation form).

  11. 0A.

    Appendix: The 5 Pitfalls of RFPs in the HPC/AI Market

    1) Over-weighting price → risk of a substandard build. 2) Verification criteria defined too late → disputes. 3) Specifications locked to a single vendor. 4) Omitted operations staffing costs. 5) Vague post-deployment support SLA.

  12. 0B.

    Appendix: Measurement Scripts and Tools

    Standard run options for HPL / STREAM / NCCL / IOR / mdtest, plus a results template.

HPC / AI Supercomputer RFP Template — A 30-Day Procurement Standard | Optiscale