29Mar, 2024
    Is A Copper Transceiver Or An Optical Transceiver, DAC, AOC Better?

    The copper transceiver is a communication interface that through copper cables (such as twisted pair network cables). It is also a type of optical transceiver.

    With the rapid development of communication technology, various transceivers have sprung up, among which copper transceivers, optical transceivers, DAC (Direct Attach Cable) and AOC (Active Optical Cable) have become hot choices in the industry. These transceivers have their own strengths and play important roles in different scenarios. To better meet communication needs, we need to deeply understand their characteristics and advantages in order to make wise choices in practical applications. Next, this article will focus on the comparison between copper transceivers, optical transceivers, DAC, and AOC, aiming to help readers better understand and choose suitable communication transceivers for themselves.

    Copper Transceiver VS DAC

    Both the copper transceiver and the DAC are suitable for short-distance transmission applications, and both are transmitted through copper cables. However, the transmission distance of the copper transceiver is larger than that of the DAC. The transmission distance of the DAC generally does not exceed 7m, while the copper transceiver can transmit 30m.

    Copper Transceiver VS AOC

    The copper transceiver uses copper cable for transmission, while the AOC uses optical cable. Therefore, the copper transceiver has a longer transmission distance, up to 300m. AOC has enhanced signal integrity and flexible deployment methods, and has better air flow and heat dissipation in the computer room cabling system, but its cost is higher than copper cabling at the 10G rate.

    Copper Transceiver VS Optical transceiver

    Compared with optical transceivers, copper transceivers use copper cables for wiring, while optical transceivers require optical fibers for wiring. If the original wiring system uses copper cables, using copper transceivers can reuse the previous copper cable wiring resources to complete the 10G network deploy. In terms of transmission distance, the transmission distance of the copper transceivers is relatively short, and the transmission distance of the optical transceiver can be as high as 120km according to different combinations.

    GIGALIGHT Copper Transceiver

    100M Copper SFP, including 100M single rate and 10/100M adaptive rate, adopts RJ45 interface, and can transmit up to 100m with Cat5/5e network cable. There are two options: commercial grade and industrial grade.

    1G Copper SFP, including 1G and 10/100/1000m adaptive rate, also uses RJ45 interface, can transmit 100m with Cat5/5e network cable, and the temperature range is commercial grade.

    10G Copper SFP supports 10G rate transmission, uses RJ45 interface, and can transmit up to 30m with Cat6/6A network cable, and the temperature range is commercial grade.

    From the latest perspective, copper transceivers are still widely used in Ethernet, especially for connections within data centers and local area networks. They are known for their reliability and cost-effectiveness, making them a popular choice for many organizations. However, as the demand for higher data rates and bandwidth continues to increase, some applications have turned to optical transceivers because of their ability to support higher data rates and longer transmission distances.

    Overall, copper transceivers continue to play a vital role in network infrastructure, providing a stable and efficient way to transmit data over copper cables. Their versatility and compatibility with existing network setups make them an essential component in modern communication systems.

    25Mar, 2024
    GIGALIGHT Innovatively Launches Linear 400G DR4 CPO Silicon Optical Engine

    Shenzhen, China, March. 25, 2024 – GIGALIGHT recently announced the launch of a 400G DR4 CPO silicon optical engine based on linear direct drive technology. This is a subset of Gigabyte’s innovative products for the next optical network era:

    1.The linear silicon optical engine 400G DR4 CPO is composed of SOCKET slot packaging and LPO linear direct drive technology. Compared with the traditionally defined NPO/CPO, linear CPO silicon photonic engines remove DSP, which can significantly reduce system-level power consumption and costs. The light emitted from the TX end of this engine is in the range of 0-3dBm, and the extinction ratio falls in the range of 4-5dB. TDECQ<2.5dB, generally falls within 1-2dB. Wavelength is 1304.5nm-1317.5nm; SMSR≥30dB; RX end OMA sensitivity <-5.5dBm. The eye diagram is as follows:

    2.Through the test board with DSP and external pluggable light source specially developed by GIGALIGHT, the linear 400G CPO optical engine can be completely restored.

    3.GIGALIGHT plans to upgrade the engine to 800G and 1.6T through higher-density advanced packaging technology.

    GIGALIGHT’s research on CPO originated from the Shenzhen Science and Technology Innovation Commission’s co-packaging CPO optical project undertaken by the company. Linear CPO is a relatively friendly technology form that can implement higher-speed and higher-density switches under the premise of limited low-nanometer manufacturing processes. Reduce power consumption and improve system interconnection reliability. Truly realize the goal of de-DSP for next-generation switches. Compared with the pluggable LPO technology that has fundamental flaws that the industry is currently working on, linear CPO is a stable technology. It should be said objectively that linear CPO is technically inspired by LPO and original CPO technology, mainly to more easily solve many problems such as system signal consistency and reliability. GIGALIGHT will delve into silicon optical linear CPO technology solutions.

    About GIGALIGHT

    As an open optical networking explorer, Gigalight integrates the design, manufacturing, and sales of both active and passive optical devices and subsystems. The company’s product portfolio includes optical modules, silicon photonics modules, liquid-cooled modules, passive optical components, active optical cables, direct attach copper cables, coherent optical communication modules, and OPEN DCI BOX subsystems. Gigalight focuses on serving applications such as data centers, 5G transport networks, metropolitan WDM transmission, ultra-HD broadcast and video, and more. It stands as an innovative designer of high-speed optical interconnect hardware solutions.

    25Mar, 2024
    Looking ahead to 2024 data center infrastructure

    Recently, DeLL’ORO GROUP released a forecast report, reviewing the outstanding trends in the 23-year data center infrastructure report and looking forward to the development of data center infrastructure in 2024. The following is the report content.

    Outstanding Trends Highlighted In 2023 Forecasts

    • Data center capex growth in 2023 has decelerated noticeably, as projected after a surge of spending growth in 2022. The top 4 US cloud service providers (SPs) in aggregate slowed their capex significantly in 2023, with Amazon and Meta undergoing a digestion cycle, while Microsoft and Google are on track to increase their greenfield spending on accelerated computing deployments and data centers. The China Cloud SP market remains depressed as cloud demand remains soft from macroeconomic and regulatory headwinds, although there are signs of a turnaround in second-half 2023 from AI-related investments. The enterprise server and storage system market performed worse than expected, as most of the OEMs are on track to experience a double-digit decline in revenue growth in 2023 from a combination of inventory correction, and lower end-demand given the economic uncertainties. However, network and physical infrastructure OEMs have fared better in 2023 because strong backlog shipments fulfilled in the first half of 2023 which lifted revenue growth.
    • We underestimated the impact of accelerated computing investments to enable AI applications in 2023. During that year, we saw a pronounced shift in spending from general-purpose computing to accelerated computing and complementary equipment for network and physical infrastructure. AI training models have become larger and more sophisticated, demanding the latest advances in accelerators such as GPUs and network connectivity. The high cost of AI-related infrastructure that was deployed helped to offset the sharp decline in the general-purpose computing market. However, supplies on accelerators have remained tight, given strong demand from new hyperscale.
    • General-purpose computing has taken a backseat to accelerated computing in 2023, despite significant CPU refreshes from Intel and AMD with their fourth-generation processors. These new server platforms feature the latest in server interconnect technology, such as PCIe 5, DDR5, and more importantly CXL. CXL has the ability to aggregate memory usage across servers, improving overall utilization. However, general-purpose server demand has been soft, and the transition to the fourth-generation CPU platforms has been slower than expected (although AMD made significant progress in 3Q23). Furthermore, CXL adoption is limited to the hyperscale market, with limited use cases.
    • Server connectivity is advancing faster than what we had expected a year ago. In particular, accelerated computing is on a speed transition cycle at least a generation ahead of the mainstream market. Currently, accelerated servers with NVIDIA H100 GPUs feature network adapters at up 400 Gbps with 112 Gbps SerDes, and bandwidth will double in the next generation of GPUs a year from now. Furthermore, Smart NIC adoption continues to gain adoption, though, mostly limited to the hyperscale market. According to our Ethernet Adapter and Smart NIC report, Smart NIC revenues increased by more than 50% in 2023.
    • The edge computing market has been slow to materialize, and we reduced our forecast in the recent Telecom Server report, given that the ecosystem and more compelling use cases need to be developed, and that additional adopters beyond the early adopters have been limited.

    According to our Data Center IT Capex report, we project data center capex to return to double-digit growth in 2024 as market conditions normalize. Accelerated computing will remain at the forefront of capex plans for the hyperscalers and enterprise market to enable AI-related and other domain specific workloads. Given the high cost of accelerated servers and their specialized networking and infrastructure requirements, the end-users will need to be more selective in their capex priorities. While deployments of general-purpose servers are expected to rebound in 2024, we believe greater emphasis will be made to increase server efficiency and utilization, while curtailing cost increases.

    Below, we highlight key trends that can enhance the optimization of the overall server footprint and decrease the total cost of ownership for end-users:

    Accelerated Computing Maintains Momentum

    We estimate that 11% of the server unit shipments are accelerated in 2023, and are forecast to grow at a five-year compound annual growth rate approaching 30%. Accelerated Servers contain accelerators such as GPUs, FPGAs, or custom ASICs, and are more efficient than general-purpose servers when matched to domain-specific workloads. GPUs will likely remain as the primary choice for training large AI models, as well as running inference applications. While NVIDIA currently has a dominant share in the GPU market, we anticipate that other vendors such as AMD and Intel will gain some share over time as customers seek greater vendor diversity. Greater choices in the supply chain could translate to much-needed supply availability and cost reduction to enable sustainable growth of accelerated computing.

    Advancements in Next-Generation Server Platform

    General-purpose servers have been increasing in compute density, as evolution in the CPUs is enabling servers with more processor cores per CPU, memory, and bandwidth. Ampere Computing Altra Max, AMD’s Bergamo are offered with up to 128 cores per processor, and Intel’s Granite Rapids (available later this year), will have a similar number of cores per processor. Less than seven years ago, Intel’s Skylake CPUs were offered with a maximum of 28 cores. The latest generation of CPUs also contains onboard accelerators that are optimized for AI inference workloads.

    Lengthening of the Server Replacement Cycle

    The hyperscale cloud SPs have lengthened the replacement cycle of general-purpose servers. This measure has the impact of reducing the replacement cost of general-purpose servers over time, enabling more capex to be allocated to accelerated systems.

    Disaggregation of Compute, Memory, and Storage

    Compute and storage have been disaggregated in recent years to improve server and storage system utilization. We believe that next-generation rack-scale architectures based on CXL will enable a greater degree of disaggregation, benefiting the utilization of compute cores, memory, and storage.

    forward from：Data Center Infrastructure—a Look into 2024, Baron Fung , Dell’Oro Group

    13Mar, 2024
    GIGALIGHT 400G QSFP-DD SR4, Exclusive Design To Help Data Center Interconnection

    Moving towards the supercomputing era, data centers have ushered in rapid development. With the continuous evolution of technology, the rise of AI supercomputing is leading the wave of innovation in the digital field. In this era with data as the core, the demand for data centers is not only the expansion of scale, but also the urgent desire for high-speed transmission. The transmission rate of 400G/800G, just like a powerful digital trend, is redefining data centers infrastructure.

    In this article, Gigalight introduces you to a brand new 400G product: 400G QSFP-DD SR4 optical module, and see how it is used in today’s data centers.

    Existing 400G network solutions

    The high-speed network in the GPU cluster mainly consists of high-speed network cards, optical connectors (optical transceivers and cables) and high-speed switches. Currently, servers equipped with H800GPU mainly use NVIDIA CX7 400G network cards, and there are many choices of switches on the market. These options include 64400G QDD switches based on BCM 25.6T Tomahawk4 (TH4) chips, 64800G switches based on BCM 51.2T chips, switches based on Cisco Nexus chips, and 400G/800G switches based on NVIDIA Spectrum chips. From a market perspective, 400G QDD switches are currently the dominant choice. The following figure shows the commonly used network architecture of clusters:

    It can be observed that the mainstream 400G switches in the market use 56G SerDes, while the CX7 network card uses 112G SerDes. In order to keep the network connected, this difference needs to be considered when selecting optical transceivers to ensure component compatibility and proper operation. As shown in the picture:

    In order to solve this problem, everyone currently chooses 400G QSFP-DD DR4 optical modules on the switch side, and uses 400G OSFP DR4 optical modules on the server side. The connection is shown below:

    This is a good solution within a connection distance of less than 500m, but if the connection distance does not exceed 100m, choosing the DR4 solution is a bit wasteful. Therefore, the short-distance 400G QSFP-DD SR4 using low-cost 850nm VSCEL laser is a good choice. It is much cheaper than DR4 and can meet the transmission distance requirement of 100m.

    400G QSFP-DD SR4 is inserted into the 400G QSFP-DD 8×50G PAM4 port, which can be interconnected with the 400G OSFP (4×100G PAM4) network card, and the interconnection between 400G QSFP-DD SR4 and 400G OSFP SR4 can be realized. . If shown:

    Gigalight’s newly launched 400G QSFP-DD SR4 electrical port uses 8×50G PAM4 modulation, and the optical port uses 4×100G PAM4 modulation. Compared with 400G QSFP-DD SR8, it reduces 4 lasers and achieves lower Power consumption and price, power consumption is less than 8W. Using MPO12 optical port, it can meet 60m transmission with OM3 multi-mode fiber, and 100m transmission with OM4 multi-mode fiber.

    The release of Gigalight’s 400G QSFP-DD SR4 fills the gap in short-distance 400G data center interconnection solutions and provides a more comprehensive choice for your data center.

    7Mar, 2024
    AI data center architecture upgrade triggers surge in demand for 800G optical transceivers

    The surge in demand for 800G optical transceivers directly reflects the increasing demand for artificial intelligence-driven applications. As the digital environment continues to evolve, the need for faster and more efficient data transmission becomes imperative. The deployment of 800G optical transceivers, coupled with the transition to a tier 2 leaf-spine architecture, reflects a strategic move to meet the needs of modern computing.

    Why do we need 800G optical transceivers?

    The surge in demand for 800G optical transceivers is closely related to changes in data center network architecture. The traditional three-tier architecture (including access tier, aggregation tier and core tier) has been the standard for many years. As the scale of AI technology and east-west traffic continue to expand, data center network architecture is also constantly evolving. In the traditional three-tier topology, data exchange between servers needs to go through access switches, aggregation switches and core switches, which puts huge pressure on aggregation switches and core switches.

    If the server cluster scale continues to expand according to the traditional three-tier topology, high-performance equipment will need to be deployed at the core tier and aggregation tier, resulting in a significant increase in equipment costs. This is where the new Spine-Leaf topology comes into play, flattening the traditional three-tier topology into a two-tier architecture. The adoption of 800G optical transceivers has promoted the emergence of Spine-Leaf network architecture, which has many advantages such as high bandwidth utilization, excellent scalability, predictable network delay, and enhanced security.

    In this architecture, leaf switches serve as access switches in the traditional three-tier architecture and are directly connected to servers. Spine switches act as core switches, but they are directly connected to Leaf switches, and each Spine switch needs to be connected to all Leaf switches.

    The number of downlink ports on a Leaf switch determines the number of Leaf switches, and the number of uplink ports on a Leaf switch determines the number of Spine switches. Together, they determine the scale of the Spine-Leaf network.

    The Leaf-Spine architecture significantly improves the efficiency of data transmission between servers. When the number of servers needs to be expanded, the scalability of the data center can be enhanced by simply increasing the number of Spine switches. The only drawback is that the Spine-Leaf architecture requires a large number of ports compared to the traditional three-tire topology. Therefore, servers and switches require more optical transceivers for fiber optic communications, stimulating the demand for 800G optical transceivers.

    Introduction to GIGALIGHT 800G optical transceivers

    In response to the surge in demand for 800G optical transceivers, Gigalight has successively launched short-distance 800G VR8/SR8 optical transceivers/AOCs based on VCSEL lasers and 800G DR8/DR8+/DR8++ optical transceivers based on silicon photonic technology. Gigalight’s silicon optical transceivers and multi-mode VCSEL optical transceivers constitute a complete interconnection solution for high-speed AI data centers.

    The multi-mode series 800G VR8/SR8/AOC is equipped with high-performance 112Gbps VCSEL laser and 7nm DSP. The electrical host interface is 112Gbps PAM4 signal per channel and supports CMIS version 4.0 protocol. Using QSFP-DD, OSFP form factor, VR8 supports 30 meters (OM3 MMF) and 50 meters (OM4 MMF) transmission distance; SR8 supports 60 meters (OM3 MMF) and 100 meters (OM4 MMF) transmission distance, with MPO16 and 2×MPO12 Two interfaces are available, suitable for short-distance data center application scenarios.

    Silicon optical transceivers focus on solving interconnection scenarios with links exceeding 100 meters. Gigalight silicon optical 800G DR8 supports QSFP-DD and OSFP two form factors. The DR8 transmission distance is 500m, DR8+ can transmit 2km, and DR8++ can transmit 10km. The  transceivers uses four 1310nm CW lasers; the maximum power consumption is less than 18W. The currently released version supports two connector architectures: MPO16/APC and Dual MPO12/APC. Compared with the 800G optical transceiver of the conventional 8-channel EML solution, the silicon optical transceiver can use fewer lasers. For example, 800G DR8 can use one laser to achieve lower power consumption. This year, Gigalight will also launch 800G QSFP-DD PSM DWDM8 and 800G OSFP/QSFP-DD 2×FR4 optical transceivers based on silicon photonics, providing customers with more choices.

    The adoption of 800G optical transceivers not only solves current challenges, but also provides forward-looking solutions to accommodate the expected growth in data processing and transmission. As technology advances, the synergy between artificial intelligence computing and high-speed optical communications will play a key role in shaping the future of information technology infrastructure.