Compute - GPU
Virtual machines equipped with NVIDIA Quadro GPUs. Ideal for complex processing and GPU-optimized workloads.
Scientists, artists, and engineers need access to significant parallel computational power. Linode offers GPU-optimized virtual machines accelerated by the NVIDIA Quadro RTX 6000, harnessing the power of CUDA, Tensor, and RT cores to execute complex processing, deep learning, and ray tracing workloads.
GPU plans are ideal for highly specialized workloads that would benefit from dedicated NVIDIA GPUs, including maching learning, AI, and data/graphics processing.
On-demand
Between purchasing, installing, and maintaining GPUs, the cost of ownership is often high. Linode GPUs allow you to leverage the power of GPUs while benefiting from the main value proposition of cloud: turning a CapEx into an OpEx.
Market Leading Hardware
Linode GPUs are NVIDIA Quadro RTX 6000 units, currently considered one of the best in market GPUs. With CUDA, Tensor, and RT cores in each unit, these GPUs support any use cases associated with parallel processing, deep learning, or ray tracing.
One GPU card isn’t enough for your projected workloads? Not a problem. Linode GPU plans offer up to four cards per instance, depending on how much horsepower you need.
Dedicated Competition-Free Resources
A GPU Compute Instance’s vCPU cores are dedicated and accessible only to you. Because the vCPU cores are not shared, no other Compute Instances can utilize them. Your instance never has to wait for another process, enabling your software to run at peak speed and efficiency. This allows you to run workloads that require full-duty work (100% CPU all day, every day) at peak performance.
Recommended Workloads
GPU Compute Instances are suitable for specialized workloads that are optimized for GPUs:
- Machine learning and AI
- Big data processing
- Video encoding
- Graphics processing
See the Use Cases for Linode GPU Instances guide to learn more about these use cases.
Availability
Atlanta, GA, United States; Frankfurt, Germany; Newark, NJ, United States; Mumbai, India; Singapore, Singapore
Plans and Pricing
Resource | Available Plans |
---|---|
GPU cards | 1-4 cards |
vCPU cores | 8-24 cores |
Memory | 32 GB - 128 GB |
Storage | 640 GB - 2560 GB |
Outbound Network Transfer | 16 TB - 20 TB |
Outbound Network Bandwidth | 10 Gbps |
Pricing starts at $1,000/mo ($1.50/hr) for a GPU Instance with 1 GPU card, 8 vCPU cores, 32 GB of memory, and 640 GB of SSD storage. Review the Pricing page for additional plans and their associated costs. See the Comparison of Compute Instances section below to learn more about other Instance types.
Additional Technical Specifications
In addition to the resources allocated to each available plan (outlined above), GPU Compute Instances have the following specifications:
- NVIDIA Quadro RTX 6000 GPUs
- Dedicated vCPU cores
- 100% SSD (Solid State Disk) storage
- 40 Gbps inbound network bandwidth
- Free inbound network transfer
- Dedicated IPv4 and IPv6 addresses (additional addresses available on request)
- Deploy using the many available Linux Distributions, Marketplace Apps, or Community StackScripts
- Direct console access through Lish
- Provisioning and management through the Cloud Manager, Linode CLI, or programmatically through the Linode API
- Multi-queue NIC support on plans with 2 or more vCPU cores.
Services Included at No Extra Cost
Linode bundles the following services with all Compute Instances:
- Always-on DDoS Protection
- Domain management through our DNS Manager
- Seamless firewall management with Cloud Firewalls
- Private Layer 2 networks with VLANs
- Metrics and monitoring through the Cloud Manager and Longview (free plan)
- Reusable deployment scripts through StackScripts
Complementary Paid Services
To help build and manage your applications, consider complementing your Compute Instance with the following compatible services:
- Automated daily and weekly backups with our Backups service
- Add additional storage drives with Block Storage
- Create and store reusable images with Custom Images
- Advanced metrics and monitoring through Longview Pro
- Automated service deployments with LKE (Linode Kubernetes Engine)
- Incident response (and more) with Managed Services
- Enable load balancing and horizontal scaling with NodeBalancers
- Add scalable storage to your application with Object Storage
Comparison of Compute Instances
Instance Type | CPU | Overview |
---|---|---|
Shared CPU Instance | Shared | 1 GB - 192 GB Memory, 1 - 32 vCPUs, 25 GB - 2840 GB Storage Starting at $5/mo ($0.0075/hour) Offers a balanced array of resources coupled with shared CPUs, which keeps costs down while still supporting a wide variety of cloud applications. Best for development servers, staging servers, low traffic web applications, and almost any modern web application. |
Dedicated CPU Instance | Dedicated | 4 GB - 512 GB Memory, 2 - 64 vCPUs, 80 GB - 7200 GB Storage Starting at $30/mo ($0.045/hour) Equipped with Dedicated CPUs, which provide competition free guaranteed CPU resources. Perfectly balanced for most production applications. Best for production websites, enterprise applications, high traffic databases, and any application that requires 100% sustained CPU usage. |
High Memory Instance | Dedicated | 24 GB - 300 GB Memory, 2 - 16 vCPUs, 20 GB - 340 GB Storage Starting at $60/mo ($0.09/hour) Optimized for memory-intensive applications and equipped with Dedicated CPUs, which provide competition free guaranteed CPU resources. Best for in-memory databases, in-memory caching systems, big data processing, and any production application that requires a large amount of memory while keeping costs down. |
GPU Instance | Dedicated | 32 GB - 128 GB Memory, 8 - 24 vCPUs, 640 GB - 2560 GB Storage Starting at $1000/mo ($1.50/hour) The only instance type that’s equipped with Nvidia Quadro RTX 6000 GPUs (up to 4) for on demand execution of complex processing workloads. Best for applications that require massive amounts of parallel processing power, including machine learning, AI, graphics processing, and big data analysis. |
See Choosing a Compute Instance Type and Plan for a full comparison.
This page was originally published on