On-Demand vCPUs and vGPUs
A Unified Compute Fabric
The Distributive Compute Protocol (DCP) makes it easy to deploy parallel workloads across any infrastructure, from GPU servers to PCs. Its cross-platform processing layer can be created, shared, and scaled without buying expensive servers or sending data to centralized cloud providers.
DCP's technology creates an open cloud from data centres around the world, where workloads have no lock-in to specific vendors. The protocol is also used to create private compute networks by capturing idle computational power from any endpoint devices on a LAN or WAN.
These two services can combine to create a hybrid cloud model, minimizing cost while maximizing performance and security.
Step 1
Instruct your workload to use private, public, or hybrid infrastructure.
Step 2
Run your workload through the Compute API without managing any servers.
Step 3
Collect your results in real time, paying only for what you consume.
Build a private grid out of any endpoints on your intranet or LAN, without investing in new servers.
Learn More -->Access CPU and GPU resources from data centres on-demand, for below spot prices elsewhere.
Learn More -->Connect your private computing network with the cloud, and seamlessly migrate workloads.
Learn More -->Create private edge networks from any IT/OT hardware, and without changing legacy systems .
Learn More -->DCP makes it easy to power computationally-intensive applications with smart workload placement.
Workload Scalability
Access thousands of computers at a fraction of the cost. You can greatly accelerate parallel workloads without being constrained by service limits.
Highly Performant
Our worker nodes improve throughput, wait time, and other performance indicators compared to VMs. Get closer to bare metal results.
Advanced Scheduling
Your workload is algorithmically mapped to the right processors. Our scheduler also prevents common multi-tenant issues like stolen CPU.
Most Competitive Pricing
Access serverless compute resources for as low as $20 per CPU year or $250 per GPU. Other computing services don't even come close.
Saves Cloud Spend
Organizations let millions of dollars of compute go to waste on hardware they already own. Our technology puts this back to work.
Transparent Billing
Know all your costs well in advance for complex workloads. DCP has both flat rate and usage-based pricing.
Pays for Itself - Literally
Your organization can sell compute it doesn't need to recover your costs, or even profit. Our 'smart grid' for compute is open to all.
Upgraded On-Prem
Our Private Networks enable you to perform more compute locally. Sending your data to 'the Cloud' may no longer make sense for many workloads.
Homomorphic Encryption
Protect your data and algorithms with industry-leading homomorphic encryption. Workloads can stay completely encrypted from end-to-end.
Granular Security Controls
Determine which end users can view program data and algorithms, if any. Enforcing your preferred access controls is very straightforward.
Orchestration-less
Scale across any number of servers without any upfront design. Our system automatically manages all orchestration and load balancing tasks.
No Migrations
Leave behind the pain of migrating VMs. Our technology lets you shift workloads between on-prem and off-prem resources at the press of a button.
Automatic Job Optimization
Our adaptive runtime lets you stop having to rightsize compute instances. Your workloads will find what they need, when they need it.
One Environment
Write code for one environment instead of planning for multiple. You can use hardware accelerators and more without modifying any source code.
Private Networks (Compute Groups) are private clouds made from heterogeneous computing hardware. They extend the serverless model to your own on-prem servers, workstations, and more.
DCP Cloud provides access to a multi-cloud federation of data centres, HPC sites, and other composable infrastructure. Its on-demand pricing is more competitive than any standalone commercial cloud provider.
Private Networks | DCP Cloud | |
---|---|---|
Infrastructure | Any x86, AMD, or RISC-V processors on your intranet or LAN. | Huge variety of public infrastructure - from AMD chips to RTX 3090s. |
Pricing | Fixed annual fee starting at $20/vCPU and $250/GPU. Unlimited use. | Consumption-based billing, averaging $100/vCPU and $1,250/GPU year. |
Security | Completely air gapped - data & algorithms stay behind your firewall. | SSH by default, with full support for homomorphic encryption (SEAL). |
Integrations | Microsoft Endpoint Manager and common network install methods. | Seamless support for multi-cloud and hybrid-cloud deployments. |
Best Suited For | Processing sensitive data & algorithms on-premises. | Massively parallel applications & hybrid cloud workloads. |
Every type of processor on devices you own.
Public data centres and edge networks
Fixed pricing at $20/vCPU and $250/vGPU.
Consumption-based at $100/vCPU and $1,250/vGPU
Completely air gapped, private, and/or on-premises
Protected by SSH and Homomorphic Encryption
Seamless multi-cloud and hybrid-cloud support.
All common network endpoint install methods.
Keeping workloads private, and pre-production testing.
Massively distributed, multi-cloud & hybrid-cloud applications.
Businesses consistently overprovision and overpay for their compute. DCP reduces compute bills by 90% or more.
Even the largest public cloud operators go down. But with DCP, you always have automated failovers.
Shorten the time to build, ship, and scale applications by only having to write code for a single environment.
DCP supports everything from command-line programs to beautiful web interfaces for non-developers.
It is compatible with almost every popular programming language, and multiple software packages used Industry Uses.
Get in touch with our experts, and test your enterprise edge network today!