EdgeX-6000-E858000
The Arbor EdgeX-6000-E858000 is a compact multi-GPU edge AI server built on the AMD EPYC Embedded 8524P (24 cores / 48 threads, Zen 4, SP6) with 4× NVIDIA Quadro A1000 MXM GPUs and up to 576 GB of 6-channel DDR5 ECC memory — server-class AI inference in a 216 × 170 × 495 mm, 10 kg chassis rated 0 °C to +65 °C.
Storage combines 4× hot-swap-capable E1.S EDSFF NVMe SSDs with an M.2 NVMe boot drive and optional 2× 2.5″ SATA. Networking spans 2× 10GbE + 2× 2.5GbE, and a redundant 3200 W CRPS power system (2× hot-plug PSUs) keeps production inference running through a supply failure.
Key Features
- AMD EPYC Embedded 8524P: 24C/48T Zen 4 at up to 4.1 GHz boost, 225 W TDP, PCIe Gen 5 — with the EPYC Embedded 7-year supply commitment for long-lifecycle infrastructure.
- 576 GB 6-channel DDR5 ECC: 4800 MHz RDIMM (96 GB per DIMM); ECC corrects single-bit errors in real time — mandatory where a silently corrupted activation could yield a wrong inference.
- 4× NVIDIA Quadro A1000 MXM: Ampere GPUs with 4 GB ECC GDDR6 each (16 GB total), ISV-certified drivers, TensorRT/CUDA/ONNX support — parallel inference streams from a chassis smaller than 2U.
- Enterprise storage: 4× E1.S EDSFF 9.5 mm NVMe + M.2 NVMe 2280 + optional 2× 2.5″ SATA.
- Multi-tier networking: 2× 10GbE for AI data ingestion/output, 2× 2.5GbE for management and out-of-band — physical segregation without a switch.
- 3200 W redundant CRPS: N+1 hot-plug PSUs; a single failure never interrupts service; industry-standard replaceable modules.
- Full-service I/O: 2× DP 1.4a at 8K (7680 × 4320 @ 60 Hz), 3× RS-232/422/485, 6× USB 3.2; CE, FCC, CB certified.
Configuration Summary
| Component | EdgeX-6000-E858000 |
|---|---|
| CPU | AMD EPYC Embedded 8524P, 24C/48T, 225 W, SP6 |
| Memory | 6-ch DDR5 4800 ECC RDIMM, up to 576 GB |
| GPU | 4× NVIDIA Quadro A1000 MXM (16 GB ECC GDDR6 total) |
| Storage | 4× E1.S NVMe + M.2 NVMe; optional 2× 2.5″ SATA |
| Networking | 2× 10GbE + 2× 2.5GbE RJ45 |
| Power | 3200 W, 2× CRPS hot-swap redundant |
| Environment | 0 °C to +65 °C; 216 × 170 × 495 mm; 10 kg |
Application Use Cases
Multi-Camera AI Video Analytics
Four GPUs run parallel inference across dozens of camera streams whilst the 24 EPYC cores decode and orchestrate pipelines; 10GbE ingests feeds at wire speed and 576 GB ECC RAM buffers multi-stream queues.
Industrial Digital Twin and Simulation
CPU cores run physics solvers on live SCADA data whilst the Quadro GPUs render 3D visualisation at 8K via DisplayPort 1.4a — in a control-room cabinet rather than a server room.
Edge Data Centre Inference Node
Cloud-class GPU inference co-located at plants, ports and substations avoids cloud latency, egress cost and data-sovereignty risk; the redundant CRPS meets production SLA requirements.
Electronics Manufacturing AI Quality Control
Parallel AOI and wafer-inspection models run on separate GPUs at line cycle time; the EPYC cores manage recipes and results; 10GbE feeds the MES without backpressure.
Why Choose the EdgeX-6000-E858000
Server CPU with embedded supply commitment. The EPYC Embedded 8524P delivers genuine server-class multi-core throughput with AMD's 7-year availability commitment — a procurement guarantee consumer and mobile CPUs cannot offer for 10-year infrastructure lifecycles.
Four GPUs where PCIe cards need 2U+. The MXM architecture packs four discrete NVIDIA accelerators into a chassis smaller than a 2U server, maximising inference throughput per rack unit where space is the binding constraint.
ECC end to end. ECC DDR5 at the CPU and ECC GDDR6 at every GPU protect the full data path against silent corruption — a system-level requirement for safety-, finance- and quality-critical inference.
Frequently Asked Questions
What processor powers the EdgeX-6000-E858000?
The AMD EPYC Embedded 8524P: 24 cores / 48 threads (Zen 4), up to 4.1 GHz boost, 225 W TDP, SP6 socket, PCIe Gen 5 — with AMD's embedded long-term supply commitment.
Which GPUs are installed?
Four NVIDIA Quadro A1000 MXM modules (Ampere) with 4 GB ECC GDDR6 each — 16 GB total — with ISV-certified professional drivers for AI inference, simulation and digital-twin workloads.
What is the maximum memory?
576 GB of DDR5 4800 MHz ECC RDIMM across 6 channels (6× 96 GB DIMMs).
What storage does it provide?
4× E1.S EDSFF 9.5 mm NVMe SSD slots, 1× M.2 M-key NVMe 2280, and optionally 2× 2.5″ SATA SSD bays.
What networking is included?
2× 10GbE and 2× 2.5GbE RJ45 — four independent ports enabling physical separation of AI data, management and out-of-band networks without an external switch.
The Arbor EdgeX-6000-E858000 is a compact multi-GPU edge AI server built on the AMD EPYC Embedded 8524P (24 cores / 48 threads, Zen 4, SP6) with 4× NVIDIA Quadro A1000 MXM GPUs and up to 576 GB of 6-channel DDR5 ECC memory — server-class AI inference in a 216 × 170 × 495 mm, 10 kg chassis rated 0 °C to +65 °C.
Storage combines 4× hot-swap-capable E1.S EDSFF NVMe SSDs with an M.2 NVMe boot drive and optional 2× 2.5″ SATA. Networking spans 2× 10GbE + 2× 2.5GbE, and a redundant 3200 W CRPS power system (2× hot-plug PSUs) keeps production inference running through a supply failure.
Key Features
- AMD EPYC Embedded 8524P: 24C/48T Zen 4 at up to 4.1 GHz boost, 225 W TDP, PCIe Gen 5 — with the EPYC Embedded 7-year supply commitment for long-lifecycle infrastructure.
- 576 GB 6-channel DDR5 ECC: 4800 MHz RDIMM (96 GB per DIMM); ECC corrects single-bit errors in real time — mandatory where a silently corrupted activation could yield a wrong inference.
- 4× NVIDIA Quadro A1000 MXM: Ampere GPUs with 4 GB ECC GDDR6 each (16 GB total), ISV-certified drivers, TensorRT/CUDA/ONNX support — parallel inference streams from a chassis smaller than 2U.
- Enterprise storage: 4× E1.S EDSFF 9.5 mm NVMe + M.2 NVMe 2280 + optional 2× 2.5″ SATA.
- Multi-tier networking: 2× 10GbE for AI data ingestion/output, 2× 2.5GbE for management and out-of-band — physical segregation without a switch.
- 3200 W redundant CRPS: N+1 hot-plug PSUs; a single failure never interrupts service; industry-standard replaceable modules.
- Full-service I/O: 2× DP 1.4a at 8K (7680 × 4320 @ 60 Hz), 3× RS-232/422/485, 6× USB 3.2; CE, FCC, CB certified.
Configuration Summary
| Component | EdgeX-6000-E858000 |
|---|---|
| CPU | AMD EPYC Embedded 8524P, 24C/48T, 225 W, SP6 |
| Memory | 6-ch DDR5 4800 ECC RDIMM, up to 576 GB |
| GPU | 4× NVIDIA Quadro A1000 MXM (16 GB ECC GDDR6 total) |
| Storage | 4× E1.S NVMe + M.2 NVMe; optional 2× 2.5″ SATA |
| Networking | 2× 10GbE + 2× 2.5GbE RJ45 |
| Power | 3200 W, 2× CRPS hot-swap redundant |
| Environment | 0 °C to +65 °C; 216 × 170 × 495 mm; 10 kg |
Application Use Cases
Multi-Camera AI Video Analytics
Four GPUs run parallel inference across dozens of camera streams whilst the 24 EPYC cores decode and orchestrate pipelines; 10GbE ingests feeds at wire speed and 576 GB ECC RAM buffers multi-stream queues.
Industrial Digital Twin and Simulation
CPU cores run physics solvers on live SCADA data whilst the Quadro GPUs render 3D visualisation at 8K via DisplayPort 1.4a — in a control-room cabinet rather than a server room.
Edge Data Centre Inference Node
Cloud-class GPU inference co-located at plants, ports and substations avoids cloud latency, egress cost and data-sovereignty risk; the redundant CRPS meets production SLA requirements.
Electronics Manufacturing AI Quality Control
Parallel AOI and wafer-inspection models run on separate GPUs at line cycle time; the EPYC cores manage recipes and results; 10GbE feeds the MES without backpressure.
Why Choose the EdgeX-6000-E858000
Server CPU with embedded supply commitment. The EPYC Embedded 8524P delivers genuine server-class multi-core throughput with AMD's 7-year availability commitment — a procurement guarantee consumer and mobile CPUs cannot offer for 10-year infrastructure lifecycles.
Four GPUs where PCIe cards need 2U+. The MXM architecture packs four discrete NVIDIA accelerators into a chassis smaller than a 2U server, maximising inference throughput per rack unit where space is the binding constraint.
ECC end to end. ECC DDR5 at the CPU and ECC GDDR6 at every GPU protect the full data path against silent corruption — a system-level requirement for safety-, finance- and quality-critical inference.
Frequently Asked Questions
What processor powers the EdgeX-6000-E858000?
The AMD EPYC Embedded 8524P: 24 cores / 48 threads (Zen 4), up to 4.1 GHz boost, 225 W TDP, SP6 socket, PCIe Gen 5 — with AMD's embedded long-term supply commitment.
Which GPUs are installed?
Four NVIDIA Quadro A1000 MXM modules (Ampere) with 4 GB ECC GDDR6 each — 16 GB total — with ISV-certified professional drivers for AI inference, simulation and digital-twin workloads.
What is the maximum memory?
576 GB of DDR5 4800 MHz ECC RDIMM across 6 channels (6× 96 GB DIMMs).
What storage does it provide?
4× E1.S EDSFF 9.5 mm NVMe SSD slots, 1× M.2 M-key NVMe 2280, and optionally 2× 2.5″ SATA SSD bays.
What networking is included?
2× 10GbE and 2× 2.5GbE RJ45 — four independent ports enabling physical separation of AI data, management and out-of-band networks without an external switch.
Get a Quote
