200G DAC Cables

Add MCP1650-V00AE30 QSFP/CX7 Cable — 200G QSFP56 Passive DAC, 0.5m – NVIDIA DGX Spark Compatible to the cart
Add MCP7H60-W002 NVIDIA/Mellanox Compatible, 400G QSFP-DD to 2x200G QSFP56 Passive Breakout DAC Cable - 2m to the cart
Add MCP1650-V001E30 QSFP/CX7 Cable — 200G QSFP56 Passive DAC, 1m – NVIDIA DGX Spark Compatible to the cart
Add MCP7H60-W001 NVIDIA/Mellanox Compatible, 400G QSFP-DD to 2x200G QSFP56 Passive Breakout DAC Cable - 1m to the cart
Add MCP7H60-W01AR30 NVIDIA/Mellanox Compatible, 400G QSFP-DD to 2x200G QSFP56 Passive Breakout DAC Cable - 1.5m to the cart
Add MCP1650-V002E26 QSFP/CX7 Cable — 200G QSFP56 Passive DAC, 2m – NVIDIA DGX Spark Compatible to the cart
Add MCP1650-H002E26 Mellanox Compatible, 200G QSFP56 Passive DAC Cable - 2m to the cart
Add MCP1650-V003E26 QSFP/CX7 Cable — 200G QSFP56 Passive DAC, 3m – NVIDIA DGX Spark Compatible to the cart
Add MCP7H60-W003 NVIDIA/Mellanox Compatible, 400G QSFP-DD to 2x200G QSFP56 Passive Breakout DAC Cable - 3m to the cart
Add MCP7Y50-N001 Compatible 800G OSFP to 4x200G OSFP Breakout DAC Cable - 1M | NVIDIA / Mellanox to the cart
Add MCP7H50-H002R26 Compatible 200G QSFP56 to 2x100G QSFP56 Breakout DAC Cable - 2M | Mellanox / NVIDIA to the cart
Add MCP7H50-H001R30 Compatible 200G QSFP56 to 2x100G QSFP56 Breakout DAC Cable - 1M | Mellanox / NVIDIA to the cart
Add MCP1650-H001E30 Mellanox Compatible, 200G QSFP56 Passive DAC Cable - 1m to the cart
Four Acer Veriton GN100 GB10 systems clustered with a MikroTik CRS812 switch on the Resilient Tec test bench

Why engineers and teams are moving LLM inference and fine-tuning onto DGX Spark and GB10 systems: bigger models than one...

Dell PowerEdge R7525 2U GPU server with AMD EPYC processors, used for on-prem inference

Does the CPU matter for GPU inference? Dell R7525 (EPYC) vs R740xd (Xeon) compared on PCIe, memory bandwidth and GPU...

Four GB10 workstations and a MikroTik switch on a shop bench

When on-prem AI inference needs a data center switch, when a DAC or ring is enough, and what lossless RoCE...

Four Acer Veriton GN100 GB10 systems clustered with a MikroTik CRS812 switch on the Resilient Tec test bench

Why engineers and teams are moving LLM inference and fine-tuning onto DGX Spark and GB10 systems: bigger models than one unit can hold, private data that never leaves the building, and a desk-side box that matches NVIDIA's production stack.

Dell PowerEdge R7525 2U GPU server with AMD EPYC processors, used for on-prem inference

Does the CPU matter for GPU inference? Dell R7525 (EPYC) vs R740xd (Xeon) compared on PCIe, memory bandwidth and GPU count.

Four GB10 workstations and a MikroTik switch on a shop bench

When on-prem AI inference needs a data center switch, when a DAC or ring is enough, and what lossless RoCE Ethernet requires.

Keith L.