HPC & AI leader in FP64
Delivers 2.3× higher FP64 per watt versus Nvidia A100.
// gpu compute
AMD Instinct MI210 GPU Server
HPC-Ready GPU Power for Demanding AI, Scientific & Engineering Workloads. Built for Density, Speed & Performance.
- Online in 1 hour when in stock
- 20+ Tier III global data centers
- Unmetered 1/10/25 Gbps bandwidth
- Crypto payments & 24/7 support included
Starting FromLive Pricing
Low-Cost MI210 BuildAMD MI210
- CPU
- 2x Intel Xeon 5218
- Cores
- 32C / 64T
- RAM
- 128 GB
- Storage
- 2x960GB SSD
- Network
- 1 Gbps · 30 TB
- Location
- London, United Kingdom
$928.60€799£686.33A$1289.11S$1176.45₹87743.38/moDeploy Now
// mi210 pricing
AMD Instinct MI210 GPU Server Price
Single and multi-GPU MI210 builds on bare metal. No setup fee, full root access, and unmetered 1/10/25 Gbps uplinks.
// gpu compute
GPU Dedicated Servers
Enterprise NVIDIA GPUs on bare metal for AI and HPC.
GPUAMD MI21064 GB · 6,656 CUDAVRAM64 GBCUDA Cores6,656TFLOPS45.3AI, Compute
Server
RecommendedTop Deal · 50% Off
2x Intel Xeon 521832C / 64T · 2.3 GHzMemory128 GB
Storage2x960GB SSD
Network1 Gbps30 TB
LocationLondonUnited Kingdom
GPUAMD MI21064 GB · 6,656 CUDAVRAM64 GBCUDA Cores6,656TFLOPS45.3AI, Compute
Server
RecommendedTop Deal · 50% Off
2x AMD EPYC 741348C / 96T · 2.65 GHzMemory128 GB
Storage2x960GB SSD
Network1 Gbps100 TB
LocationAmsterdamNetherlands
GPUAMD MI21064 GB · 6,656 CUDAVRAM64 GBCUDA Cores6,656TFLOPS45.3AI, Compute
Server
RecommendedTop Deal · 50% Off
2x AMD EPYC 754364C / 128T · 2.8 GHzMemory128 GB
Storage2x960GB SSD
Network1 Gbps30 TB
LocationLondonUnited Kingdom
No servers match your filters.
Reservation Requested
An engineer will reach out shortly to confirm availability and next steps.
Bare Metal // Standard Equipment
All Bare Metal Plans Include
Every server ships fully dedicated: no shared resources, no usage meters, no surprises.
- Setup Cost
- Free
- Provisioning
- Instant & Automated
- Access
- KVM, IPMI, Root
- Protection
- DDoS Shield Included
- Cores
- Up to 128 (Dual Socket)
- Memory
- Up to 2TB RAM
- Storage
- Enterprise NVMe & SSD
- Support
- 24/7/365 Human Engineers
OS & Panels
* Select your OS and panel at checkout
// why redswitches for mi210
What You Get With Every AMD Instinct MI210 Server
Every RedSwitches AMD Instinct MI210 server is single-tenant bare metal with full root and IPMI access, unmetered 1, 10 or 25 Gbps bandwidth, no setup fee, no bandwidth overage and a flat monthly price. Stocked builds are online in about 1 hour, larger configurations up to 8 GPUs per node are built to order, volume and 6/12-month committed-term discounts apply, payments include crypto, and engineers answer 24/7.
- 1 hr
Online When in Stock
Stocked AMD Instinct MI210 builds are online in about 1 hour. Configurations not in stock are built to order, and an engineer confirms the lead time before you commit.
- $0
Setup Fee, Flat Monthly
No setup fee and a flat monthly price for the whole server. No per-hour meter and no surprise line items, so GPU spend is forecastable.
- Unmetered
Bandwidth, No Egress Bills
Unmetered 1, 10 or 25 Gbps uplinks are included, and whatever port speed you choose you can use all of it: no overage charges and no egress bills, ever. Moving datasets, checkpoints and model weights in and out costs nothing extra.
- 1 tenant
Bare Metal, Full Control
Single-tenant hardware with root and IPMI access. You choose the OS, drivers, CUDA or ROCm version, and the NVLink or MIG layout, and a private VLAN can link your RedSwitches servers.
- Up to 8x
Multi-GPU, Built to Order
Single and multi-GPU nodes, up to 8 GPUs per server with NVLink where the card supports it. Volume discounts on multi-GPU and multi-server orders, plus 6 and 12-month term savings.
- 24/7
Engineers, Not Bots
Live chat, Telegram and email answered by engineers around the clock. Pay by card, PayPal, bank wire or crypto with no KYC, in 20+ Tier III data centers across the EU, US and Asia.
// compared with cloud
RedSwitches MI210 Server vs a Typical Cloud GPU Instance
Same accelerator, different economics: what changes when the GPU sits in a dedicated server you control instead of a metered instance.
| RedSwitches MI210 server | Typical cloud GPU instance | |
|---|---|---|
| Billing | Flat monthly price per server | Per-hour or per-second metering |
| Bandwidth | Unmetered 1/10/25 Gbps, use the full port, no overage or egress fees | Egress billed per GB |
| Private networking | Private VLAN between your servers on request | Paid VPC and peering constructs |
| Setup fee | $0 | Varies by instance and region |
| Tenancy | Single-tenant bare metal | Shared, virtualised hosts |
| Access | Root and IPMI, your OS and drivers | Hypervisor-managed images |
| Multi-GPU | Up to 8 per node, NVLink where supported, built to order | Fixed instance shapes |
| Payments | Card, PayPal, bank wire, crypto | Card or invoice |
| Support | 24/7 engineers on chat, Telegram and email | Ticket tiers, paid support plans |
AMD Instinct MI210 Core Specifications
CDNA 2 silicon with 64 GB of ECC HBM2e and Infinity Fabric peer links, built for FP64 HPC density.
- Architecture
- AMD CDNA 2 (Aldebaran, 6 nm)
- Compute Units / Stream Processors
- 104 CUs, 6,656 stream processors
- Memory
- 64 GB HBM2e on 4096-bit bus (1.6 TB/s) with ECC
- Clock Speed
- 1.0 GHz base → 1.7 GHz boost
- FP64
- 22.6 TFLOPS
- FP32
- 22.6 TFLOPS
- FP16/BF16/INT8
- 181 TFLOPS/181 TOPS
- Form Factor
- Dual-slot, PCIe 4.0 ×16, 300 W TDP, passive cooling
- Infinity Fabric I/O
- Up to 3× 300 GB/s P2P links enabling 4-GPU hives with 600 GB/s
- Virtualization
- ECC memory + PCIe fabric, ROCm 5 support
Performance
// why mi210
Why Choose Instinct MI210
HPC-class double precision, massive HBM bandwidth, and an open ROCm software stack on single-tenant bare metal.
Massive HBM & I/O density
64 GB of HBM2e and 1.6 TB/s memory bandwidth in a PCIe card, unmatched in mainstream servers.
Scale with peer links
Build high-bandwidth GPU clusters via Infinity Fabric, 600 GB/s cross-GPU in a hive.
Energy-efficient dual-slot design
300 W TDP supports passive cooling, ideal for dense rack deployments.
Open compute stack
Seamless with ROCm, optimized containers, and Infinity Hub tools.
// use cases
Ideal Use Cases
From FP64 simulation to GPU hive acceleration, see where the MI210 changes what your team can ship.
Scientific & Engineering HPC
Ideal for FP64 workloads like CFD, molecular dynamics, climate modeling, fast, efficient, and dense.
AI Training & Mixed-Precision Compute
Supports FP32, BF16, INT8, great for mid-sized transformer fine-tuning, language models, and reinforcement learning.
Large In-Memory Compute & Simulations
Ideal for database caching, graph analytics, and processing large datasets in memory.
GPU Hive Acceleration
Up to 4 MI210s peer-linked for high-bandwidth multi-GPU training and compute bursts.
Private Cloud & Cloud Repatriation
Replace cloud GPU nodes with powerful, cost-effective bare metal featuring ECC and open-stack software.
Trusted by Enterprise Teams Worldwide
Read all RedSwitches reviews, or see them on Google, HostAdvice, Cryptwerk and Trustpilot.
// faqs
Deep Dive & FAQs
Common questions about MI210 density, Infinity Fabric, ROCm support, and cooling.
How many MI210 GPUs per server?
Typically 2–4 in PCIe racks, with peer linking for high-bandwidth compute clusters.
Is Infinity Fabric better than PCIe?
Yes, dedicated peer links deliver up to 300 GB/s between GPUs, far exceeding PCIe capabilities.
Does ROCm fully support this GPU?
Absolutely, ROCm 5 and Infinity Hub offer optimized frameworks, container images, and HPC tooling.
Is 64 GB HBM enough?
Yes, for large models, graph datasets, and full double-precision workflows, without compromising high throughput.
Passive cooling: any risk?
No, dual-slot design and 300 W TDP work ideally in airflow-optimized rack servers designed for HPC.
// need help
Not Sure Exactly What You Need
No problem. Our talented engineers will consult, architect, migrate, manage, and do whatever it takes to help your business grow and succeed.