Direct operator endpoint for VortexGrid's Nordic compute cluster (SE-STO-01). Attach to assigned physical NVIDIA RTX 6000 Ada and L40S workstation chassis over deterministic HTTP/3 QUIC and multiplexed HTTP/2 TLS 1.3 transport with sub-8.5ms frame delivery.
Single-tenant physical blades assigned via hardware FIDO2 seat binding. Cross-PoP Perforce vault replication links Stockholm directly with Warsaw and Frankfurt.
| Chassis ID | Location / Rack | Compute / CPU | GPU Accelerator | VRAM / Memory | Uplink Fabric | Allocation State |
|---|---|---|---|---|---|---|
| STO-BM-B1 | Stockholm SK1 // R-D02 | AMD EPYC 9124 (16C) | 1× NVIDIA RTX 6000 Ada | 48 GB ECC / 128 GB DDR5 | 2× 10GbE SFP+ | ACTIVE // LOCAL POP |
| STO-BM-B2 | Stockholm SK1 // R-D03 | AMD EPYC 9374F (32C @ 4.3GHz) | 2× NVIDIA L40S | 96 GB ECC / 256 GB DDR5 | 2× 25GbE SFP28 | RESERVED // STUDIO LOCK |
| STO-BM-B3 | Stockholm SK1 // R-D04 | AMD Threadripper PRO 7975WX (32C) | 2× NVIDIA RTX 6000 Ada | 96 GB ECC / 256 GB DDR5 | 2× 25GbE SFP28 | INVITE ONLY (1 SLOT) |
| WAW-BM-A1 | Warsaw LIM // R-C14 | AMD Threadripper PRO 7975WX (32C) | 2× NVIDIA RTX 6000 Ada | 96 GB ECC / 256 GB DDR5 | 2× 25GbE SFP28 | PEER POP // WAW-01 |
Interactive 120 Hz viewport frames and 1000 Hz HID device events ride over BBR-paced HTTP/3 QUIC (UDP 443) for zero head-of-line blocking, with automatic HTTP/2 TLS 1.3 fallback on restricted enterprise links.
Each Stockholm blade mounts an isolated LUKS2 NVMe-oF scratch volume synced continuously against our Warsaw and Frankfurt vault mirrors so artists resume Unreal Engine 5 builds without cold checkout delays.
Unauthenticated requests arriving at app.vortexgrid.site terminate cleanly at the edge controller. Studio operators must bind an organization seat token and FIDO2 / X.509 certificate before blade stream handoff.