LABS
NexusGrid
A distributed browser supercomputer. Compute tensor operations and matrix multiplications directly on client GPUs.
Castov NexusGrid
Decentralized WebGPU AI Compute Swarm. Pools idle browser GPU power via P2P WebRTC tunnels to train models without massive server farms.
Local WebGPU Adapter
Statusunsupported
VendorDetecting...
ArchitectureDetecting...
Max Buffer Size0 MB
Swarm Execution Logs
// Awaiting commands...
Local Node Power
0.00 GFLOPS
Memory Throughput
0.00 GB/s
Active P2P Nodes
0
Grid Compute Pool
0.0 PFLOPS
Global P2P Topology Matrix
Grid Disconnected.
Click "Join Global Compute Grid" to initialize WebRTC channels.
Click "Join Global Compute Grid" to initialize WebRTC channels.
Protocol: WebRTC DataChannelsEncryption: AES-GCM 256Engine: WGSL Compute Shaders
WebGPU Browser Compute
NexusGrid harnesses browser WebGPU compute shaders to run matrix operations locally at native hardware speeds.
How to use this tool
- Open the NexusGrid dashboard.
- Allocate GPU memory for matrix tensors.
- Execute compute kernels directly in your browser.