NVIDIA MFS1S50-H010E 200G to 2x100G QSFP56 AOC Low Latency Splitter Cable
商品の詳細:
| ブランド名: | Mellanox |
| モデル番号: | MFS1S50-H010E |
| 書類: | MFS1S50-H0xxV.pdf |
お支払配送条件:
| 最小注文数量: | 1個 |
|---|---|
| 価格: | Negotiate |
| パッケージの詳細: | 外箱 |
| 受渡し時間: | 備蓄に基づいた |
| 支払条件: | T/T |
| 供給の能力: | プロジェクト/バッチによる供給 |
|
詳細情報 |
|||
| 可用性: | ストック | 保証: | 1年 |
|---|---|---|---|
| 状態: | 新しい、オリジナル | テクノロジー: | インフィニバンド |
| データレート: | 最大200GB/s | コネクタの種類: | QSFP56 |
| 直径: | 3 +/- 0.2mm | 最小曲線放射: | 30mm |
| ニアエンドの出力の目の高さ: | 70mvpp | ニアエンドESMW: | 0.265UI |
製品の説明
The NVIDIA® MFS1S50-H010E is a high-performance active optical splitter cable (AOC) that enables 200Gb/s to 2x100Gb/s connectivity in modern data centers. Designed for cost-sensitive yet demanding environments, this QSFP56 VCSEL-based AOC provides a seamless interface between a 200Gb/s switch port and two separate 100Gb/s servers or compute nodes. Compliant with SFF-8665 and SFF-8679 standards, the MFS1S50 series delivers robust signal integrity, low latency, and superior power efficiency. The 10-meter length (H010E) is ideal for top-of-rack (ToR) to server connections, simplifying cabling while maintaining high bandwidth.
Each splitter cable integrates EEPROM on both QSFP56 ends for real-time product monitoring and status reporting, fully compatible with NVIDIA InfiniBand and Ethernet switches. With rigorous production testing and NVIDIA’s unique quality design, this AOC ensures plug-and-play reliability and extended lifecycle in enterprise, HPC, and AI cluster deployments.
Enables 200Gb/s switch port to serve two 100Gb/s endpoints, improving port density and reducing infrastructure cost.
VCSEL-based multi-mode fiber architecture ensures ultra-low latency for latency-sensitive HPC and AI workloads.
Real-time monitoring: Tx/Rx optical power, bias current, supply voltage, case temperature, warning/alarm thresholds.
Operating case temperature 0°C to 70°C, storage -40°C to 85°C, suitable for demanding data center environments.
Tx input equalization, Rx output amplitude, pre-emphasis; Tx/Rx CDR control for 100GbE operation.
Standard 2-wire management interface, ModSelL, IntL, ResetL, LPMode for full host control.
The MFS1S50-H010E utilizes Vertical Cavity Surface-Emitting Laser (VCSEL) arrays and high-sensitivity PIN photodiodes over four multi-mode fiber pairs. At the 200Gb/s QSFP56 side, all four lanes operate at 50Gb/s PAM4 (53.125GBd) to achieve 200Gb/s aggregate bandwidth; the breakout ends each handle two lanes at 100Gb/s (50Gb/s PAM4 per lane). This active optical design eliminates copper signal degradation and extends reach up to 50 meters while maintaining bit error rates below 1e-12.
The cable features SFF-8636-compliant control signals, including module select, low-power mode, and interrupt line. On-chip digital diagnostic monitoring (DDM) provides real-time insight into link health, simplifying network management and troubleshooting. Additionally, the transceiver supports rate select for flexible line rates: 10.3125G NRZ, 25.78125G NRZ, and 26.5625G PAM4 to accommodate backward compatibility where needed.
- Top-of-Rack (ToR) switch to dual server connectivity in 200GbE or InfiniBand HDR fabrics.
- High-performance computing (HPC) clusters requiring low latency and high-density cabling.
- AI/ML training infrastructure with 200G uplinks to GPU servers (e.g., NVIDIA Quantum or Spectrum switches).
- Data center spine-leaf architectures where port efficiency and simplified cable management are critical.
- Enterprise core aggregation reducing transceiver count while increasing bandwidth per rack unit.
Engineered for NVIDIA Quantum InfiniBand switches (e.g., QM8700, QM9700 series) and NVIDIA Spectrum Ethernet switches. The MFS1S50-H010E also interoperates with any standard QSFP56 port supporting 200GbE or InfiniBand HDR/NDR-compatible link training. The AOC’s EEPROM provides unique product identification, making it fully managed by NVIDIA networking platforms. For non-NVIDIA hosts, the cable complies with the QSFP56 MSA (SFF-8665) and the electrical interface follows SFF-8679. Always confirm host port configuration for breakout support.
| Platform / Product Family | Supported Speed | Port Type |
|---|---|---|
| NVIDIA Quantum HDR InfiniBand Switches (QM8700, QM8790) | 200G HDR | QSFP56 |
| NVIDIA Spectrum-2/3 Ethernet Switches (SN3700, SN4600) | 200GbE | QSFP56 |
| NVIDIA ConnectX-6 HDR / ConnectX-6 Dx | 100Gb/s per tail | QSFP56 |
| Third-party 200GbE switches with breakout support | 200G→2x100G | Compliant QSFP56 |
| Parameter | Details |
|---|---|
| Product Model | MFS1S50-H010E |
| Form Factor | QSFP56 to 2x QSFP56 active optical splitter cable |
| Data Rate | 200Gb/s (main end) / 100Gb/s per breakout branch |
| Signaling Rate per Lane | 53.125 GBd PAM4 (nominal) |
| Cable Length | 10 meters (tolerance: +300mm / -0 for lengths <50m) |
| Fiber Type | Multi-Mode Fiber (MMF) 50/125µm, Aqua jacket |
| Minimum Bend Radius | 30 mm |
| Supply Voltage (Vcc) | 3.135V – 3.465V (Typ. 3.3V) |
| Power Consumption (200G end) | Typ. 4.5W, Max 5.0W |
| Power Consumption (100G end per branch) | Typ. 3.0W, Max 3.5W |
| Operating Case Temperature | 0°C to 70°C |
| Storage Temperature | -40°C to 85°C |
| Relative Humidity (non-condensing) | 5% to 85% |
| ESD Tolerance | Compliant with typical data center ESD handling; class I laser safety |
| Diagnostic Features | DDM (Tx/Rx power, bias, voltage, temperature), LOS/LOL, Tx fault, programmable equalization |
| Regulatory Compliance | CE, FCC Class A, ICES, RCM, VCCI, CB, cTUVus, RoHS (per NVIDIA standards) |
| Ordering Part Number | Description | Length |
|---|---|---|
| MFS1S50-H003V | NVIDIA AOC, 200G to 2x100G, QSFP56 to 2xQSFP56 | 3 meters |
| MFS1S50-H005V | NVIDIA AOC, 200G to 2x100G, QSFP56 to 2xQSFP56 | 5 meters |
| MFS1S50-H010E | NVIDIA AOC, 200G to 2x100G, QSFP56 to 2xQSFP56 | 10 meters |
| MFS1S50-H015V | NVIDIA AOC, 200G to 2x100G, QSFP56 to 2xQSFP56 | 15 meters |
| MFS1S50-H020V | NVIDIA AOC, 200G to 2x100G, QSFP56 to 2xQSFP56 | 20 meters |
| MFS1S50-H030V | NVIDIA AOC, 200G to 2x100G, QSFP56 to 2xQSFP56 | 30 meters |
For custom lengths or bulk requirements, please contact our sales team. The "E" suffix denotes specific regional or factory coding; electrical and optical performance identical to standard H0xxV series.
- Genuine Product Assurance: 100% authentic NVIDIA cables, fully traceable and backed by manufacturer quality.
- Technical Expertise: Our engineers provide pre-sales compatibility validation and post-sales support.
- Global Logistics: Fast worldwide shipping with proper anti-static packaging and ESD protection.
- Competitive Pricing & Volume Discounts: Ideal for data center buildouts and infrastructure refresh projects.
- Extended Warranty Options: Standard 1-year warranty with extended coverage available upon request.
Hong Kong Starsurge Group provides end-to-end service: technical specification consulting, interoperability testing assistance, and RMA support. Our team is available for remote debugging, installation guidance, and firmware compatibility validation. Every cable is visually inspected before shipment to ensure clean optical connectors and proper labeling. For mission-critical projects, we offer on-site replacement services in select regions. Contact your account manager for SLA options.
- Confirm that your switch and NIC ports support breakout mode (200G → 2x100G).
- Verify required cable length and bend radius constraints in your rack layout.
- Ensure both host systems at the 100G ends have QSFP56 cages with proper airflow and power budget.
- Check whether your environment demands additional regulatory certifications (e.g., China RoHS).
- Consider ordering spare units for identical link symmetry in high-availability clusters.
- Electrostatic discharge (ESD) sensitive: Use grounded wrist straps and ESD-safe workstations during installation.
- Do not exceed absolute maximum supply voltage (3.6V) or data input voltage to prevent irreversible damage.
- Avoid tight bending below 30mm radius to maintain optical performance and prevent fiber breakage.
- Always insert QSFP modules gently into cages; forced insertion may damage connectors or cage latches.
- Operate within case temperature range 0–70°C; ensure adequate airflow in high-density environments.
- In case of any visible damage to the splitter boot or jacket, replace the cable to avoid signal integrity issues.
- NVIDIA MFS1S00-HxxxV – 200G to 200G QSFP56 active optical cable (point-to-point)
- NVIDIA MCP1650-HxxxE – 100GbE QSFP28 to 2x50GbE SFP56 splitter AOC
- NVIDIA Quantum HDR QM8700 Switch – 40-port 200Gb/s InfiniBand switch
- NVIDIA ConnectX-6 Dx Adapter Card – 100Gb/s dual-port QSFP56 Ethernet/IB adapter
- LinkX Memory Map Application Note (MLNX-15-5926) – EEPROM register details
- NVIDIA Cable Management Guidelines and FAQs (MLNX-15-3603)
- SFF-8636 Rev 2.9 – Management Interface for QSFP Modules
- Best Practices for Deploying 200G Breakout Cables in Data Centers
Hong Kong Starsurge Group Co., Limited is a technology-driven provider of network hardware, IT services, and system integration solutions. Founded in 2008, the company serves customers worldwide with products including network switches, NICs, wireless access points, controllers, cables, and related networking equipment. Backed by an experienced sales and technical team, Starsurge supports industries such as government, healthcare, manufacturing, education, finance, and enterprise. The company also offers IoT solutions, network management systems, custom software development, multilingual support, and global delivery. With a customer-first approach, Starsurge focuses on reliable quality, responsive service, and tailored solutions that help clients build efficient, scalable, and dependable network infrastructure.







