Field notes from the EMARQUE Lab.
Reference architectures, client guides, and decision matrices, written by the specialists who scope, build, and run the systems. New articles land here as deployments wrap and client permissions allow.
Reference architectures
Validated stacks (topology, components, software) for the deployments EMARQUE sets up most often.
Client guides
How EMARQUE solutions architects scope a project before committing hardware.
How to size an on-prem AI system
The eight numbers EMARQUE asks for before quoting hardware — model size, concurrency, latency budget, deployment site, compliance, timeline, and two we usually have to dig for.
Workstation vs Server vs Rack-Scale
Pick the right class before drilling into specs. Sized by users, model class, deployment shape, and budget.
H200 vs B200 vs B300 — which GPU?
The three NVIDIA data-centre GPUs you'll be quoted in 2026. HBM, FP4, NVLink, and which EMARQUE / DGX systems carry each.
Generations: Hopper → Rubin
What each generation actually buys you, what's shipping now, and how to plan a multi-year refresh.
We publish field notes from real deployments.
Case studies, benchmarks, and reference architectures land here under customer permission. If you're scoping something interesting and willing to be referenced — talk to us.
Get in Touch with Us
Tell us about your workload. We reply within one business day with a quote sized to fit.
- 01
Key Account Manager
+6012 627 2280 - 02
Request for Quotation
business@emarque.co