Open-Source Distributed Compute Network
Rent Distributed GPUs for 80% Less with Zero Lock-in
Aggregate idle NVIDIA & AMD compute from independent providers worldwide. Launch PyTorch, vLLM, and OCI containers with deterministic billing and sandboxed isolation.
Network Hardware & Capacity Telemetry SIMULATED PREVIEW (Offline Backend)
Active Compute Nodes
...
Simulated Cluster Preview
Aggregated Accelerators
...
Simulated GPU/CPU Fleet
Network Compute (TFLOPS)
...
Estimated Synthetic TFLOPS
Cost Efficiency Savings
...
vs Centralized Cloud Tiers
Intelligent Scheduler
Multi-objective optimization algorithm matches workloads against real-time latency, pricing, and hardware capabilities.
Sandboxed Isolation
Rootless container execution with cgroups memory limits, dedicated device passthrough, and zero host network exposure.
Verified Benchmarks
Hardware is verified via standardized FLOPS and memory bandwidth synthetic benchmarks before listing on the marketplace.
Real-Time Compute Market
Available Distributed Compute Nodes
SIMULATED INVENTORY (Real Hardware Verification Pending)
● ONLINESIMULATED TIER • us-east (Virginia)
8x NVIDIA H100 80GB HBM3
640 GB total VRAM • US-East-H100-OctaFleet
CPU Cores
96 Cores
System RAM
512 GB DDR5
NVMe Storage
4000 GB
Benchmark Score
985 / 1000
$14.50/hr
99.9% reliability
● ONLINESIMULATED TIER • eu-central (Frankfurt)
2x NVIDIA A100 80GB SXM4
160 GB total VRAM • EU-Central-A100-Dual
CPU Cores
64 Cores
System RAM
256 GB DDR5
NVMe Storage
2000 GB
Benchmark Score
890 / 1000
$3.20/hr
99.7% reliability
● ONLINESIMULATED TIER • ap-northeast (Tokyo)
4x NVIDIA GeForce RTX 4090
96 GB total VRAM • AP-Tokyo-RTX4090-Pro
CPU Cores
32 Cores
System RAM
128 GB DDR5
NVMe Storage
1500 GB
Benchmark Score
780 / 1000
$1.75/hr
98.9% reliability
● ONLINESIMULATED TIER • us-west (Oregon)
1x NVIDIA L40S 48GB
48 GB total VRAM • US-West-L40S-HighMem
CPU Cores
32 Cores
System RAM
128 GB DDR5
NVMe Storage
1000 GB
Benchmark Score
740 / 1000
$0.95/hr
99.4% reliability
Monetize Idle Hardware
Earn Passive Income with Your GPU & CPU Fleet
Install our lightweight, rootless agent on your gaming rigs, mining rigs, or datacenter servers. The platform automatically benchmarks, registers, and routes verified paying AI workloads to your hardware.
Zero Inbound Ports: Secure outbound-only WSS communication.
Strict Container Sandboxing: Rootless OCI execution isolates your host OS.
Automated Payouts: Hourly usage calculations credited straight to your wallet.
provider-agent-quickstart.sh
# 1. Pull & Run Agent via Podman
podman run -d \
--name compute-agent \
--device nvidia.com/gpu=all \
-e PAIRING_TOKEN=ptk_sec_9941a8 \
ghcr.io/distributed-compute/agent:latest
--name compute-agent \
--device nvidia.com/gpu=all \
-e PAIRING_TOKEN=ptk_sec_9941a8 \
ghcr.io/distributed-compute/agent:latest
# 2. Output Log Verification
[INFO] Detected 2x NVIDIA H100 80GB HBM3
[INFO] Synthetic FLOPS Benchmark: 985/1000
[INFO] Node registered successfully: node-01
[INFO] Listening for container workloads...
[INFO] Synthetic FLOPS Benchmark: 985/1000
[INFO] Node registered successfully: node-01
[INFO] Listening for container workloads...