computeFP32 fused arithmetic / storage output
Pending
Detected architecture
Inspecting GPU
Inspecting GPU- Device
- Pending permission
- Vendor
- Unknown
- Compute API
- Detecting
- Max storage
- Unknown
- Confidence
- Unverified
Supported GPU families
NVIDIA
AMD
Intel
Apple
NVIDIAAMDIntelApple
Verified reward mapping
NVDAx is the only planned reward for eligible server-timed sessions. NVIDIA, AMD, Intel, Apple, protected, and unknown GPU identities all map to NVDAx. Browser-reported hardware labels are descriptive, not cryptographic attestation.
memorystrided vec4 streaming / resident buffer
Pendingrenderprocedural ray field / 512 square
Pendingmatrix256 square SGEMM / storage output
Pendinginferencedeterministic dense forward pass
Pending0%woven
Elapsed00:00
InstrumentIdle
Dispatches0
Loom units0.00
Run length
Mining passTrial available
Mining is for $LOOM holders: keep at least $30 of $LOOM in your connected wallet. One one-minute run is free so you can confirm that your GPU works. A completed trial publishes to the global feed as a clearly labeled, no-reward record; it never enters a reward epoch.
Holder passx1.0$30+ of $LOOM
Operator passx1.25$300+ of $LOOM
Foundry passx1.5$3,000+ of $LOOM
Your tier is read when the pass is issued and the required holding is rechecked when a lease starts and completes. Tokens are not locked or staked. Every production run receives a one-use lease and must return one unpredictable server checkpoint per minute. Reward accounting uses server-observed duration rather than the browser's self-reported score.
Contract address TBA. Mining-pass issuance and production sessions are not live yet.