Server rack and modulator device and exchange box thereof

Abstract

The present invention is directed to heat dissipation of interconnected transceivers applied to a server system. The server rack includes a rack structure for supporting one or more rack servers and a switch box mechanically mounted on the rack structure. The switch box is configured to receive a removable optical interconnect and includes a transceiver module and a cooling structure coupled to the transceiver module. The transceiver module is configured to convert an input signal to an output signal and generate heat when converting the input signal. The cooling structure is configured to inject a coolant through the inlet and output the coolant through the outlet, thereby allowing the coolant to at least partially take away heat generated by the transceiver module. In some embodiments, the cooling structure includes a metal plate that contacts the transceiver module through a contact surface to absorb heat generated by the transceiver module.

Description

Technical Field

[0001] This invention relates to cooling technology in electronic systems, including but not limited to methods, devices, structures, apparatuses, and systems for heat dissipation, which targets the heat generated by high-speed interconnect switching and conversion circuits with a compact form factor in server systems. Background Technology

[0002] Interconnect transceivers in server racks are hardware components used to enable high-speed data transmission between servers, switches, and other network devices. They are typically small, modular devices that plug into the ports of network hardware. When used in server racks, these interconnect transceivers can face several challenges, primarily related to heat dissipation, signal integrity, and physical wear and tear. Because these transceivers operate at high data rates, they generate significant amounts of heat, which can pose thermal management challenges within the limited space of a server rack. If not properly managed with appropriate cooling solutions, overheating can lead to performance degradation or even hardware failure. Furthermore, maintaining signal integrity during high-speed operation is crucial; any signal degradation caused by electromagnetic interference (EMI), low-quality cables, or connectors can lead to data transmission errors, resulting in network instability and increased latency.

[0003] Another issue is the physical durability of the transceivers and their connections. Frequent plugging and unplugging of transceivers for maintenance or upgrades can cause wear and tear on connectors and ports, leading to poor connections or complete interconnect failure. Accumulation of dust and debris in server racks can also affect connectivity and transceiver performance. Furthermore, ensuring compatibility between different types and brands of transceivers and network equipment can be quite complex, requiring thorough testing and verification to avoid interoperability problems. These challenges necessitate careful planning, regular maintenance, and appropriate environmental control to ensure the reliable and efficient operation of high-speed server interconnects. Summary of the Invention

[0004] Several embodiments of the present invention relate to methods, apparatuses, structures, devices, and systems for heat dissipation targeting high-speed interconnect transceivers with a compact form factor. For example, interconnect transceivers may operate at data rates up to 1.6 terabits per second (Tb / s) or 3.2 Tb / s in data centers performing artificial intelligence tasks, thus generating significant amounts of heat requiring efficient and timely cooling. In some embodiments, the optical engines of individual communication channels are decoupled from their associated optical fibers and integrated into a transceiver module contained within a switch box mechanically mounted in a rack structure. A cooling structure is disposed within the switch box and coupled to the transceiver module. The transceiver module may include optical engines for multiple communication channels associated with multiple optical fibers (e.g., 32 or 64 fibers). A coolant is configured to flow through the body of the cooling structure to at least partially remove the heat generated by the transceiver module. In these ways, the transceiver module can provide a compact form factor compared to separately packaged optical engines, optical fibers, or associated ports, while benefiting from the effective cooling effect achieved by the cooling structure.

[0005] Based on at least some embodiments disclosed herein, it is recognized that the integration of fiber optic cables with transceiver ports limits the port density of the switching enclosure, and that heat sinks or cooling plates (which dissipate heat for the entire switching enclosure) may be bulky and insufficient to dissipate the heat generated by the fiber-optic-coupled transceivers. In particular, when data centers perform artificial intelligence (AI) or high-performance computing (HPC) tasks, the data transmission rates of the associated servers exceed 1.6 Tb / s and 3.2 Tb / s, thus requiring efficient heat dissipation on the fiber-coupled transceivers. In some implementations, the fiber optic cable is separated from the associated optical engine and / or switching application-specific integrated circuit (switching ASIC), allowing the fiber optic cables to be densely packed to increase port density. The switching ASIC can be effectively cooled by cooling structures, thereby supporting the signal-to-noise ratio required to achieve the desired data transmission rates (e.g., 1.6 Tb / s and 3.2 Tb / s).

[0006] In one aspect, some embodiments include a server rack. The server rack includes a rack structure for supporting one or more rack servers, and a switching box mechanically mounted on the rack structure. The switching box further includes (e.g., surrounding) a transceiver module and a cooling structure coupled to the transceiver module. The switching box is configured to receive multiple detachable optical interconnects, and the transceiver module is configured to convert multiple input signals into multiple output signals, generating heat in the process of converting the multiple input signals. The cooling structure includes an inlet and an outlet, and is configured to inject coolant through the inlet and discharge coolant through the outlet, thereby allowing the coolant to at least partially remove the heat generated by the transceiver module.

[0007] In some embodiments, the cooling structure includes a metal plate with a contact surface that contacts the transceiver module to absorb heat generated by the transceiver module. Additionally, in some embodiments, the metal plate includes a coolant channel sealed within the metal plate, with an inlet and an outlet each coupled to a respective edge of the metal plate and connected to a respective end of the coolant channel, the coolant channel extending generally parallel to the contact surface from the inlet to the outlet.

[0008] In some implementations, the switch box further includes multiple ports configured to receive multiple detachable electrical interconnects, each port configured to exchange electrical signals with a corresponding rack server mounted on the rack structure.

[0009] In some implementations, the switch box further includes multiple ports configured to receive multiple detachable electrical interconnects, a first subset of which is coupled to multiple rack servers on one or more alternative server racks, each alternative server rack including at least one rack server electrically coupled to a corresponding port of the first subset.

[0010] In another embodiment, some embodiments include a modulator device that further includes a transceiver module enclosed in a switching box and a cooling structure coupled to the transceiver module. The switching box is configured to receive multiple detachable optical interconnects, and the transceiver module is configured to convert multiple input signals into multiple output signals, generating heat in the process of converting the multiple input signals. The cooling structure includes an inlet and an outlet, and is configured to inject coolant through the inlet and discharge coolant through the outlet, thereby allowing the coolant to at least partially remove the heat generated by the transceiver module.

[0011] In another aspect, a method for providing a server rack is provided. The method includes providing a rack structure for supporting one or more rack servers, and providing a switching box mechanically mounted on the rack structure. The switching box is configured to receive a plurality of detachable optical interconnects. Providing the switching box includes providing a transceiver module configured to convert a plurality of input signals into a plurality of output signals and generate heat in the process of converting the multiple input signals. Providing the switching box further includes providing a cooling structure coupled to the transceiver module. The cooling structure includes an inlet and an outlet, and is configured to inject coolant through the inlet and discharge coolant through the outlet, thereby allowing the coolant to at least partially remove the heat generated by the transceiver module.

[0012] In another aspect, a method is implemented in a server rack comprising a rack structure for supporting one or more rack servers and a switching box mechanically mounted on the rack structure. The method includes receiving multiple detachable optical interconnects by the switching box. The switching box includes transceiver modules and a cooling structure coupled to the transceiver modules. The method further includes receiving multiple input signals by the transceiver modules; converting the multiple input signals into multiple output signals by the transceiver modules; and generating heat by the transceiver modules during the conversion of the multiple input signals. The method further includes injecting coolant through an inlet of the cooling structure and discharging coolant through an outlet of the cooling structure, thereby allowing the coolant to at least partially remove the heat generated by the transceiver modules.

[0013] These exemplary embodiments and references to implementation methods are not intended to limit or define the invention, but rather to provide examples to aid understanding. Additional embodiments are discussed in the detailed description, and further description is provided therein. Attached Figure Description

[0014] To better understand the various implementation methods described herein, please refer to the following detailed description in conjunction with the following figures, wherein the same reference numerals refer to the corresponding parts in each figure.

[0015] Figure 1A This is a front view of an example server rack according to some embodiments, which supports one or more servers.

[0016] Figure 1B This is an example data center according to some embodiments, including a server hierarchy organized on multiple server racks.

[0017] Figure 2 This is a block diagram of an example system module in a typical computer device according to some embodiments, which can be applied as the server in Figure 1.

[0018] Figure 3A and Figure 3B Two simplified front views of an example server rack including a switch box, according to some embodiments.

[0019] Figure 4 This is a perspective view of an example switching box motherboard according to some embodiments and an enlarged view of a transceiver module disposed on the motherboard.

[0020] Figure 5A An unfolded perspective view of an integrated module including a cooling structure and a transceiver module coupled to the cooling structure, according to some embodiments.

[0021] Figure 5B For coupling to according to some embodiments Figure 5A Top perspective view of an example cooling structure for the transceiver module shown.

[0022] Figures 6A-6C This is a cross-sectional view of an example exchange box according to some embodiments.

[0023] Figure 7 This is a block diagram of a transceiver module coupled between a fiber optic port and a server-side port in a switching box, according to some embodiments.

[0024] Figure 8 This refers to a server group comprising multiple servers, according to some embodiments.

[0025] Figure 9A This is a rear view of an example server rack including a server cooling system according to some embodiments. Figure 9B This is a front view of another example server rack, including a server cooling system, according to some embodiments.

[0026] Figure 10 This is a flowchart of a method for providing a server rack according to some embodiments.

[0027] Figure 11 This is a flowchart of a method for managing input data implemented on a server rack according to some embodiments.

[0028] The same reference number in each figure refers to the corresponding part.

[0029] The attached figures are labeled as follows:

[0030] 100: Server rack 100A: Server rack

[0031] 100B: Replacement server rack; 102: Frame

[0032] 104: Slot 104-1: First tray

[0033] 104-2: Second tray; 104A: Server tray

[0034] 104B: Empty slot; 106: Computing device module

[0035] 108: Firewall Module; 110: Switch Box

[0036] 112: Display device; 114: Keyboard

[0037] 116: Data storage device; 116S: Solid State Drive (SSD)

[0038] 116N: Network Attached Storage (NAS) 118: Uninterruptible Power Supply (UPS)

[0039] 120: Server 120A: First rack server

[0040] 120C: Core Server; 120S: Spine Server

[0041] 120SA: Server 120L: Leaf Server

[0042] 120LA: Servers; 150: Data Center

[0043] 152: First communication path; 154: Second communication path

[0044] 156: Third communication path; 200: System module

[0045] 202: Processor module; 204: Memory module

[0046] 206: Input / Output (I / O) Controller; 208: Network Interface

[0047] 210: Memory controller; 212: Solid-state drive (SSD)

[0048] 214: Hard Disk Drive (HDD) 216: Power Supply Unit (PSU)

[0049] 218: Power Management Integrated Circuit (PMIC) 220: Graphics Module

[0050] 222: Audio module; 240: Data bus

[0051] 302: Interconnection 304: Interconnection

[0052] 312: Input signal; 312A: Electrical signal

[0053] 312B: Optical signal; 314: Output signal

[0054] 314A: Electrical signal; 314B: Optical signal

[0055] 330: Corresponding edge 400: Motherboard

[0056] 402: Fiber Optic Port; 404: Interconnector Port

[0057] 410: Transceiver Module; 500: Integration Module

[0058] 502: Entrance 504: Exit

[0059] 506: Coolant; 508: Metal plate

[0060] 510: Cooling structure; 512: Coolant passage

[0061] 600: Cross-sectional view; 602: Server side port

[0062] 602A: Port Group 1; 602B: Port Group 2

[0063] 606: Optical Engine; 608: Application-Specific Integrated Circuit (ASIC) for Switching

[0064] 610: Optical fiber cable; 612: Digital signal processing (DSP) circuit.

[0065] 612A: First DSP block; 612B: Second DSP block

[0066] 614: Electrical Switching Cable; 614-620: Cross-sectional View

[0067] 640: Sectional View 700: Conveyor

[0068] 704: Laser diode; 706: Laser driver circuit

[0069] 708: Optical sensor; 710: Intermediate electrical signal

[0070] 712: Electrical Signals; 800: Server Cluster

[0071] 900: Server cooling system; 902: Fan

[0072] 904: Cooling Distribution Module (CDM) 906: Connector

[0073] 908: Coolant pump; 910: Coolant controller

[0074] 912: Entrance 914: Exit

[0075] 1000: Method 1002: Operation

[0076] 1004: Operation 1006: Operation

[0077] 1008: Operation 1010: Operation

[0078] 1100: Method 1102: Operation

[0079] 1104: Operation 1106: Operation

[0080] 1108: Operation 1110: Operation

[0081] d: Separation distance Detailed Implementation

[0082] Detailed reference will now be made to specific embodiments, examples of which are illustrated in the accompanying drawings. In the following detailed description, numerous non-limiting details are set forth to aid in understanding the subject matter presented herein. However, it will be apparent to those skilled in the art that various alternatives may be used without departing from the scope of protection of this invention, and that the invention may be practiced without these specific details.

[0083] Figure 1A This is a front view of an example server rack 100 (also referred to as a rack mount, rack cabinet, or simply rack) according to some embodiments, which supports one or more servers 120. The server rack 100 includes a frame 102 and a plurality of slots 104, which can be used in data centers, server rooms, or network rooms to support, organize, and manage multiple computing device modules 106 (e.g., servers 120, storage devices such as solid-state drives (SSDs) 116S and network-attached storage (NAS) 116N, network devices, and other types of hardware). Each slot 104 of the server rack 100 is configured to receive and support a corresponding computing device module 106. In some embodiments, the plurality of slots 104 includes at least one unused blank slot 104B that can receive a computing device module 106 when needed. In some implementations, the server rack 100 has a predefined width of 19 or 23 inches, a height of up to 84 inches or more, and a depth selectable from 24, 32, 40, or 48 inches.

[0084] Examples of computing device modules 106 supported by multiple slots 104 of server rack 100 include, but are not limited to, firewall module 108, switch box 110, server 120, display device 112, keyboard 114, solid-state drive (SSD) 116S, network attached storage (NAS) 116N, and uninterruptible power supply (UPS) 118. Each computing device module 106 plays a corresponding role in maintaining the network and computing environment. In some embodiments, firewall module 108 is a network security device that monitors and controls inbound and outbound network traffic according to predetermined security rules, thereby establishing a barrier between trusted internal networks and untrusted external networks. Firewall module 108 may be placed near the network entry point to protect server rack 100 from unauthorized access, malware, and network attacks. In some embodiments, firewall module 108 includes packet filtering, stateful inspection, virtual private network (VPN) support, and intrusion prevention system (IPS). In some embodiments, the switch box 110 and firewall module 108 are placed near the network entry point and configured to receive input signals and forward them (e.g., convertible to electrical signals) to different servers 120 mounted on server rack 100. The switch box 110 is used in server rack 100 to minimize cable length and ensure efficient network traffic management. The switch box 110 supports different speeds (e.g., 800 gigabits per second (Gbps), 1.6 Tb / s, 3.2 Tb / s), has multiple ports (24, 48, etc.), and provides features such as Virtual Area Network (VLAN) support, Power over Ethernet (PoE), and managed or unmanaged functionality.

[0085] The multiple computing device modules 106 of server rack 100 may include multiple servers 120, each server 120 configured to provide data, resources, services, or programs to other client devices via one or more wired or wireless communication networks. Each server 120 is mounted in a slot 104 of server rack 100 and configured to provide one or more services (e.g., web hosting, database management, and application support). Servers 120 mounted on server rack 100 can provide higher processing power, larger storage capacity, redundant power supplies, and hot-swappable components compared to individual client devices to achieve high availability and reliability. In some embodiments, one or more rack servers 120 include multiple graphics processing units (GPUs) configured to operate in a data center 150 associated with machine learning tasks. Figure 1B Machine learning operations are performed in ).

[0086] Solid-state drives (SSDs) 116S and network attached storage (NAS) 116N are configured to provide storage space for server 120 mounted on server rack 100. SSDs use flash memory to store data and, compared to hard disk drives (HDDs), offer high speed, low latency, durability, and low power consumption, and are available in various capacities and form factors. Conversely, the network attached storage (NAS) 116N is a dedicated file storage device that provides data access to the network, allowing a large number of different types of client devices to retrieve data from a centralized disk capacity. In some embodiments, the network attached storage (NAS) 116N may have high capacity, redundant array of independent disks (RAID), support for multiple file sharing protocols (NFS, SMB / CIFS, FTP), user management, and backup functions. In some embodiments, the solid-state drives (SSDs) 116S are storage drives for speed, for example, for use within server 120 located on the same server rack 100, while the network attached storage (NAS) 116N is configured for file sharing, data backup, and remote access.

[0087] In some implementations, the uninterruptible power supply (UPS) 118 is used to provide emergency power to other computing device modules 106 during a power outage, enabling them to operate long enough to safely shut down or switch to an alternative power source. For example, the UPS 118 may be installed in a server rack 100 or placed in a bottom slot to support weight, providing backup power to other computing device modules 106. The UPS 118 provides battery backup, surge protection, voltage regulation, real-time monitoring, management software, and / or capacity- and load-based runtime.

[0088] Server rack 100 further includes multiple mechanical structures configured to provide mechanical support or facilitate access to multiple computing device modules 106. These mechanical structures include one or more of the following: open-frame racks (e.g., doorless or side-panel-less), mounting rails, cable management features (e.g., arms, hooks, and trays), power strips, shelves, drawers, and blank panels. In some embodiments, the multiple mechanical structures also include rack housings (e.g., cabinets), lockable doors, and side panels to protect the computing device modules 106 from unauthorized access. For example, server rack 100 includes or is coupled to multiple panels configured to convert server rack 100 into a server cabinet. In some embodiments, server rack 100 further includes a cooling or ventilation system to facilitate heat dissipation. Using server rack 100 helps optimize space, improve cooling efficiency, simplify maintenance, and enhance the overall organization and management of information technology (IT) infrastructure.

[0089] Some implementations of server rack 100 include a rack structure (e.g., including frame 102 and multiple slots 104) for supporting one or more rack servers 120. Switch box 110 includes transceiver modules (e.g., Figure 3B The transceiver module 410 in the middle) and related cooling structures (e.g. Figure 3B The cooling structure 510 is included. In some embodiments, the switching box 110 completely surrounds the transceiver module and the associated cooling structure. The switching box is mechanically mounted on a rack structure. The transceiver module is configured to convert multiple input signals (e.g., optical signals) into multiple output signals (e.g., optical or electrical signals), generating heat in the process of converting the multiple input signals. The cooling structure is coupled to the transceiver module. The cooling structure includes an inlet and an outlet, configured to inject coolant through the inlet and discharge coolant through the outlet, thereby allowing the coolant to at least partially remove the heat generated by the transceiver module.

[0090] Figure 1B Example data center 150 according to some embodiments includes a hierarchy of servers 120 organized on multiple server racks 100. Data center 150 is used for cloud computing to perform different types of tasks, such as artificial intelligence (AI), high-performance computing (HPC), network and / or storage data management. Data center 150 may include physical facilities housing computing machines and their associated hardware, such as servers 120, data storage devices 116, and network devices. Data center 150 is used to provide cloud services.

[0091] In some embodiments, the server 120 hierarchy of the data center 150 includes a three-tier server structure (e.g., core server 120C, spine server 120S, leaf server 120L). At each tier, a corresponding server rack 100 includes a set of corresponding servers 120. The server rack 100 including the core server 120C is communicatively coupled to the server rack 100 including the spine server 120S via multiple first communication paths 152 (e.g., extending 2 km or less). The server rack 100 including the spine server 120S is communicatively coupled to the server rack 100 including the leaf server 120L via multiple second communication paths 154 (e.g., extending 100 m or less). Each leaf server rack 100L includes and organizes a set of leaf servers 120L. The switch box 110 of each leaf server rack 100L is communicatively coupled to another switch box 110 or leaf server 120L on an adjacent leaf server rack via multiple third communication paths 156 (e.g., extending 20 m or less). Each leaf server rack 100L's switch box 110 is communicatively coupled to the leaf server 120L on the same leaf server rack 100L via multiple fourth communication paths (e.g., approximately 2 meters or less in length). In other words, the communication paths can be applied to different levels of the data center 150, such as within each server rack 100 (e.g., from switch box 110 to...). Figure 1A The servers 120 in the middle, the adjacent server racks 100, from leaf server 120L to spine server 120S, and from spine server 120S to core server 120C.

[0092] Regardless of the layer of server 120, the corresponding first communication path 152, second communication path 154, or third communication path 156 has a signal-to-noise ratio (SNR) lower than the corresponding threshold for its target data transmission rate. For example, the target data transmission rate on the first communication path 152, second communication path 154, and third communication path 156 is 1.6 Tb / s, 3.2 Tb / s, or higher. Each communication path (e.g., path 154A) is coupled between the originating server (e.g., server 120SA) that generates the data to be transmitted and the destination server (e.g., server 120LA) that receives the data to be transmitted. The higher the data transmission rate, the greater the heat generated by the transceiver modules of the originating and destination servers. In several embodiments of the invention, the heat generated by the transceiver modules is effectively dissipated by directly using cooling structures on each transceiver module, so that the SNR can be controlled to maintain the target data transmission rate (e.g., 1.6 Tb / s, 3.2 Tb / s, or higher).

[0093] In some cases, large language models (LLMs), autonomous driving, generative artificial intelligence, and cloud services require significant bandwidth and data transfer rates from data center 150. For example, to support data center 150 (e.g., to implement Content Security Policy (CSP) or machine learning), in-rack and inter-rack data communication may require target data transfer rates of 1.6 Tb / s or 3.2 Tb / s. The data transfer rate growth of traditional pluggable optical modules has lagged far behind the growth rate of data center traffic. Global data center data rates may increase from 400 Gbps and 800 Gbps to 1.6 Tb / s, with greater data bandwidth and lower data latency. The gap between application requirements and the capabilities of traditional pluggable optical modules continues to widen. In some embodiments, co-packaged optics (CPO) or linearly driven pluggable optics (LPO) increase interconnect bandwidth density and energy efficiency by shortening electrical connection lengths, achieved through the packaging and co-optimization of electronic and photonic wafers. Further details on CPO and LPO solutions will be available in [reference needed]. Figures 6A-6C This will be discussed below.

[0094] In some cases, the speed of in-rack and inter-rack clustered Ethernet networks corresponds to the error rate caused by heat dissipation. Higher transceiver temperatures associated with fiber optics result in lower data communication efficiency and slower data transmission rates. In some embodiments, integrated electro-optic transceiver assemblies are located at each fiber optic port and operate at power consumption that cannot be effectively cooled, leading to higher bit error rates. For example, the power consumption of integrated transceiver assemblies is 5-17 W when data transmission rates are below 1 Tb / s. In some implementations, data transmission rates of 1.6 Tb / s and 3.2 Tb / s require power consumption of 25 W and 35 W, respectively. Given the heat that needs to be dissipated, these power consumption levels may limit the application of these transceiver assemblies in data centers with significantly high target data transmission rates (e.g., 1.6 Tb / s or 3.2 Tb / s). In some embodiments of the invention, the transceiver module may integrate the fiber-optic associated optical engine and / or associated switching ASIC in a switching box and away from the associated fiber optic port, allowing the cooling structure to absorb and transfer the heat generated by the transceiver module in a centralized manner.

[0095] Figure 2This is a block diagram of an example system module 200 in a typical computer device according to some embodiments, which may be applied as server 120 in FIG1. ​​The system module 200 in this computer device includes at least a processor module 202, a memory module 204 for storing programs, instructions, and data, an input / output (I / O) controller 206, one or more communication interfaces (e.g., a network interface 208), and one or more data buses 240 for interconnecting these components. In some embodiments, the I / O controller 206 allows the processor module 202 to communicate with I / O devices (e.g., a keyboard, mouse, or touchpad) via a Universal Serial Bus interface. In some embodiments, the network interface 208 includes one or more interfaces for Wi-Fi, Ethernet, and Bluetooth networks, each allowing the computer device to exchange data with external sources (e.g., a server or another computer device). In some embodiments, the data bus 240 includes circuitry (sometimes referred to as a chipset) for interconnecting and controlling communication between the various system components included in the system module 200.

[0096] In some embodiments, memory module 204 includes high-speed random access memory, such as dynamic random access memory (DRAM), static random access memory (SRAM), double data rate (DDR) dynamic random access memory (RAM), or other random access solid-state memory devices. In some embodiments, memory module 204 includes non-volatile memory, such as one or more disk storage devices, optical disk storage devices, flash memory devices, or other non-volatile solid-state storage devices. In some embodiments, memory module 204 or the non-volatile memory device within memory module 204 includes non-transitory computer-readable storage media. In some embodiments, a memory slot is reserved on system module 200 to receive memory module 204. Once inserted into the memory slot, memory module 204 is integrated into system module 200.

[0097] In some embodiments, system module 200 further includes one or more components selected from: memory controller 210, solid-state drive (SSD) 212, hard disk drive (HDD) 214, power supply unit (PSU) 216, power management integrated circuit (PMIC) 218, graphics module 220, and audio module 222. Memory controller 210 is configured to control communication between processor module 202 and memory components (including memory module 204) in the computer device. Solid-state drive (SSD) 212 is configured to store data in the computer device using integrated circuit components, and in many embodiments is based on NAND or NOR memory. Hard disk drive (HDD) 214 is a conventional data storage device used for storing and retrieving digital information based on electromechanical disks. Power supply unit (PSU) 216 is configured to receive external power and provide multiple DC power supplies (e.g., 12V, 54V). The power management integrated circuit (PMIC) 218 ​​is configured to modulate multiple DC power supplies to other required DC voltage levels, such as 5V, 3.3V, or 1.8V, to meet the requirements of various components or circuits within the computer device (e.g., processor module 202). The graphics module 220 is configured to generate output images according to desired image / video formats and feed them to one or more display devices. The audio module 222 is configured to facilitate audio signal input and output of the computer device under the control of a computer program.

[0098] It is worth noting that the data bus 240 also interconnects and controls communication between various system components included in the system module 200, including the memory controller 210 to the audio module 222.

[0099] Figure 3A and Figure 3B The following are two simplified front views of an example server rack 100 according to some embodiments, which includes a switch box 110. Some implementations of the server rack 100 include a rack structure (e.g., including frame 102 and multiple slots 104 in FIG. 1) for supporting one or more rack servers 120. The switch box 110 is mechanically mounted to the rack structure (e.g., in a top slot of the server rack 100) and configured to receive multiple detachable optical interconnects 302. The switch box 110 includes transceiver modules (e.g., ... Figure 4 The transceiver module 410 is configured to convert multiple input signals 312 into multiple output signals 314 and output multiple output signals 314.

[0100] refer to Figure 3AIn some embodiments, one or more rack servers 120 are configured to receive multiple output signals 314 (e.g., optical or electrical signals) via multiple output interconnects 304. Additionally, in some embodiments, multiple input signals 312 include a set of optical signals received via multiple detachable optical interconnects 302, and multiple output signals 314 include a set of electrical signals 314A configured to be transmitted to one or more rack servers 120. Alternatively, in some embodiments, multiple output signals 314 include a set of output optical signals 314B configured to be transmitted to one or more rack servers 120. The multiple output signals 314 may include only electrical signals 314A, only optical signals 314B, or a combination thereof.

[0101] Conversely, reference Figure 3B In some embodiments, one or more rack servers 120 are configured to provide multiple input signals 312 to the switch box 110. Furthermore, in some embodiments, the multiple input signals 312 include a set of electrical signals 312A provided by one or more rack servers 120, and the multiple output signals 314 include a set of optical signals transmitted via multiple detachable optical interconnects 302. Alternatively, in some embodiments, the multiple input signals 312 include a set of input optical signals 312B provided by one or more rack servers 120. The multiple input signals 312 may include only electrical signals 312A, only optical signals 312B, or a combination thereof.

[0102] Figure 4 Showing a perspective view of a motherboard 400 of an example switch box 110 according to some embodiments, and an enlarged view of a transceiver module 410 disposed on the motherboard 400. The switch box 110 is mechanically mounted on a rack structure (e.g., slot 104 of a server rack 100) for supporting one or more rack servers 120 (FIG. 1). The switch box 110 includes a transceiver module 410 disposed on the motherboard 400 and configured to receive a plurality of detachable optical interconnects 302. The transceiver module 410 is configured to convert a plurality of input signals 312 into a plurality of output signals 314 and generate heat in the process of converting the plurality of input signals 312. In some embodiments, the switch box 110 further includes a plurality of fiber optic ports 402 configured to receive the plurality of detachable optical interconnects 302. The plurality of detachable optical interconnects 302 are coupled to the transceiver module 410 and can provide optical signals as a plurality of input signals 312 or receive optical signals as a plurality of output signals 314.

[0103] In some embodiments, each of the plurality of detachable optical interconnects 302 includes an interconnect port 404 configured to mate with a corresponding fiber optic port 402 via a fixing structure and to mechanically fix the end of the corresponding optical interconnect 302 to the switching box 110. Optical signals can be exchanged between the transceiver module 410 and each detachable optical interconnect 302 via the corresponding interconnect port 404 and the corresponding fiber optic port 402. In other words, in some embodiments, the detachable optical interconnect 302 does not include an optical engine within its fiber optic port 402; the optical engine of the detachable optical interconnect 302 is integrated into the transceiver module 410 disposed on the mainboard 400 of the switching box 110.

[0104] When the optical engines of the detachable optical interconnects 302 are moved into the switching housing 110, the size of the interconnect port 404 of each optical interconnect 302 is reduced compared to the interconnect port 404 including the corresponding optical engine. This arrangement allows a greater number of interconnects 302 to enter the limited interface space of the switching housing 110, thereby increasing the port density of the switching housing 110. Furthermore, the transceiver module 410 has a compact form, and the heat generated by the transceiver module 410 can be dissipated through a cooling structure (e.g., a metal plate). Conversely, in some cases, for each detachable optical interconnect 302, even if the space of the interconnect port 404 can accommodate the corresponding optical engine, there is almost no heat dissipation mechanism to accommodate the space to effectively dissipate the heat generated by the transceiver module 410. In other words, in some implementations, the transceiver module 410 is configured to integrate the optical engines of the interconnect ports 404 of the detachable optical interconnects 302. When integrated into the switching housing 110, the transceiver module 410 and the cooling structure (e.g., a metal plate) are... Figure 5A and Figure 5B The transceiver module 410 is compatible with the structure 510 in the data center server system and provides centralized heat dissipation. In this way, the application of the transceiver module 410 increases the port density of the switching box 110 used in the data center server system and overcomes the thermal challenges of high-speed data transmission (e.g., 1.6 Tb / s, 3.2 Tb / s).

[0105] Figure 5A An unfolded perspective view of an integrated module 500 comprising a cooling structure 510 and a transceiver module 410 coupled to the cooling structure 510, according to some embodiments. Figure 5B For coupling to according to some embodiments Figure 5A The top perspective view of an example cooling structure 510 for the transceiver module 410 is shown. In some embodiments, both the cooling structure 510 and the transceiver module 410 are enclosed in a switching box 110. Figure 5AIn a server rack 100 (not shown), the switch box is mounted on the server rack 100 (FIG. 1). For example, the switch box 110 is mechanically mounted on a rack structure (e.g., the top slot 104 of the server rack 100) for supporting one or more rack servers 120 (FIG. 1), and the transceiver module 410 may be disposed on the motherboard 400 of the switch box 110. Figure 4 In some cases, the switch box 110 may be mechanically fixed to the rack structure and cannot be detached manually without tools. The switch box 110 is configured to receive multiple detachable optical interconnects 302. The transceiver module 410 includes multiple optical engines (e.g., Figure 6A The optical engine 606 in the middle is configured to convert multiple input signals 312 into multiple output signals 314. Figure 3A and Figure 3B ), and generates heat when converting multiple input signals 312.

[0106] In some implementations, the cooling structure 510 is coupled to the transceiver module 410. For example, the bottom surface of the cooling structure 510 may at least partially contact the top surface of the transceiver module 410 to absorb heat generated by the transceiver module 410. The cooling structure 510 includes an inlet 502 and an outlet 504, configured to inject coolant 506 through the inlet 502. Figure 5B The coolant 506 is output through outlet 504, thereby allowing the coolant 506 to carry away at least part of the heat generated and absorbed by the transceiver module 410.

[0107] In some embodiments, the cooling structure 510 may be coupled to the transceiver module 410 via adhesive or a fixing structure. In some embodiments, the transceiver module 410 and the cooling structure 510 cannot be manually separated without tools. At least one of the transceiver module 410 and the cooling structure 510 may be mechanically fixed to the switch box 110 and cannot be manually separated from the switch box 110 without tools. In some embodiments, the rack structure associated with the server rack 100 includes slots (e.g., Figure 9A The second tray 104-2 is used to receive the switch box 110 containing the transceiver module 410 and the cooling structure 510, allowing the switch box 110 to be removed from the server rack 100. The transceiver module 410 and the cooling structure 510 can be replaced individually or together on the motherboard 400 of the switch box 110.

[0108] In some embodiments, a plurality of fiber optic ports 402 of the switching box 110 are coupled to one or more edges (e.g., one, two, three, or four edges) of the motherboard 400. The transceiver module 410 is connected via a plurality of fiber optic ports 402 and a plurality of interconnect ports 404. Figure 4The transceiver module 410 is coupled to multiple detachable optical interconnects 302. The optical interconnects 302 can provide multiple input signals 312 or carry multiple output signals 314. In some embodiments, the multiple detachable optical interconnects 302 received by the interconnect port 404 have a data communication bandwidth greater than a threshold bandwidth (e.g., 1 Tb / s), and the transceiver module 410 has a power consumption level greater than 25 W. For example, the data bandwidth of the detachable optical interconnects 302 is 1.6 Tb / s or 3.2 Tb / s. Further details regarding signal transmission have been referenced. Figure 3A and Figure 3B As discussed above.

[0109] refer to Figure 5B In some embodiments, the cooling structure 510 serves as a radiator and heat dissipator, including a metal plate 508. The metal plate 508 contacts the transceiver module 410 via a contact surface to absorb heat generated by the transceiver module 410. In some embodiments, the metal plate 508 comprises one or more of the following: naval brass, aluminum, aluminum brass, carbon steel, copper, copper-nickel alloy 70 / 30, copper-nickel alloy 90 / 10, nickel-copper alloy (also known as Monel alloy), and stainless steel (e.g., duplex or super duplex grade). Furthermore, in some embodiments, the metal plate 508 includes a coolant channel 512 sealed within the metal plate 508, with an inlet 502 and an outlet 504 each coupled to a corresponding edge 330 of the metal plate and connected to a corresponding end of the coolant channel 512. The coolant channel 512 may have a serpentine shape and extend generally parallel to the contact surface from the inlet 502 to the outlet 504.

[0110] Furthermore, in some embodiments, the height of the metal plate 508 is greater than a threshold dimension, for example, equivalent to or greater than the length or width of the metal plate 508, forming a metal block. Coolant channels 512 may extend within the three-dimensional body of the metal block. In some implementations, the coolant channels 512 extend along multiple parallel layers, each layer being generally parallel or perpendicular to the contact surface between the metal plate 508 and the transceiver module 410. In particular, in an example not shown, the coolant channels 512 extend continuously from a bottom layer adjacent to and parallel to the contact surface to each upper layer above the bottom layer, which is parallel to the contact surface.

[0111] Figures 6A-6CThe figures 600, 620, and 640 are cross-sectional views of an example switch box 110 according to some embodiments. The switch box 110 includes a motherboard 400, a transceiver module 410 mounted on the motherboard 400, a fiber optic port 402 coupled to the motherboard 400, and a server-side port 602. In some embodiments, the fiber optic port 402 and the server-side port 602 are each coupled to a corresponding edge of the motherboard 400 and exposed from a corresponding side of the switch box 110. The fiber optic port 402 and the server-side port 602 are configured to receive corresponding detachable optical interconnects 302 and server-side interconnects 304 (e.g., carrying optical or electrical signals), respectively.

[0112] In some embodiments, the transceiver module 410 further includes an optical engine 606 and a switching ASIC 608. The optical engine 606 is coupled to each fiber optic port 402 via an optical cable 610 and is configured to transmit the input signal 312 ( Figure 3A The input signal 312 (i.e., an optical signal) is converted into an intermediate electrical signal, which is further processed by the switching ASIC 608 to generate an output electrical signal 314A. The input signal 312 is provided by a corresponding detachable optical interconnect 302 via a fiber optic port 402 and an optical cable 610 disposed in the switching box 110. Conversely, the switching ASIC 608 is configured to convert the input electrical signal 312A (i.e., an optical signal) into an intermediate electrical signal, which is further processed by the switching ASIC 608 to generate an output electrical signal 314A. Figure 3B This signal is converted into another intermediate electrical signal, which drives the optical engine 606 to generate an output signal 314 fed into the optical cable 610. The output signal 314 is an optical signal, which is provided to the corresponding detachable optical interconnect 302 via the fiber optic port 402 and the optical cable 610.

[0113] In some embodiments, the surface area of ​​transceiver module 410 is larger than that of transceiver module 410 and completely covers transceiver module 410. In some cases, heat is primarily generated by the switching ASIC 608, which may include digital signal processing (DSP) circuitry 612. Cooling structure 510 is aligned with the area corresponding to the switching ASIC 608 to dissipate heat generated by the switching ASIC 608. Cooling structure 510 relies on liquid cooling to dissipate heat generated by the switching ASIC 608. Reference Figure 6A In some embodiments, the optical engine 606 is coupled to the motherboard 400 along with the switching ASIC 608, while maintaining a separation distance d from the switching ASIC 608. The separation distance d reduces the impact of heat generated by the switching ASIC 608 on the operation of the optical engine 606.

[0114] refer to Figure 6BIn some embodiments, the optical engine 606 includes an optically pluggable module separate from the transceiver module 410 containing the switching ASIC 608. This optically pluggable module is mechanically coupled to the fiber optic port 402 in a pluggable manner. In some embodiments, the optically pluggable module corresponding to the optical engine 606 is inserted into the fiber optic port 402 and automatically aligned and connected to the electrical switching cable 614. After the optical engine 606 is inserted into the fiber optic port 402, a corresponding detachable optical interconnect 302 is further inserted into the fiber optic port 402 and automatically aligned and connected to the input of the optical engine 606. When the detachable optical interconnect 302 is removed, it can be selectively removed with or without the optical engine 606. In these implementations, the liquid cooling effect of the cooling structure 510 is further provided by the switching ASIC 608 and the optically pluggable module (e.g., Figure 6B The optical engine 606 in the ASIC is separated or isolated. The switching ASIC 608 can be cooled to a relatively low operating temperature (e.g., below 60°C) to avoid overheating issues that can affect the associated transmission error rate when data transfer rates exceed certain threshold rates (e.g., 1.6 Tb / s, 3.2 Tb / s, 6.4 Tb / s, and 12 Tb / s).

[0115] In some embodiments, the switching ASIC 608 includes a digital signal processing (DSP) circuit 612, which further includes a first DSP block 612A and a second DSP block 612B. Reference Figure 6C The first DSP block 612A is integrated into the switching ASIC 608, and the second DSP block 612B is coupled to or integrated with the optical engine 606 in the optical pluggable module. The second DSP block 612B is physically separate from the transceiver module 410. The second DSP block 612B is configured to receive and process the output signal of the optical engine 606 and provide the output signal to the transceiver module 410 via the electrical switching cable 614.

[0116] In other words, in some embodiments, Figure 6A The switching box 110 is formed according to a common packaged optics (CPO) scheme. Optical components (e.g., optical engine 606) and silicon components (e.g., switching ASIC 608) are integrated on a single package substrate (e.g., to address data bandwidth and power consumption issues). The single-package CPO substrate is configured to support fiber optics, digital signal processing (DSP), switching ASIC, and packaging, and can be used in data center and cloud infrastructure. Alternatively, in some embodiments, Figure 6B or Figure 6C The exchange box 110 is formed according to a linearly driven pluggable optics (LPO) scheme, which utilizes an optically pluggable module. Furthermore, in Figures 6A-6CIn various embodiments, varying degrees of physical separation between the switching ASIC 608 and the optical engine 606 help reduce the impact of heat generated by the switching ASIC 608 on the performance of the optical engine 606. A cooling structure 510 may be coupled to a CPO- or LPO-based substrate (e.g., motherboard 400), further contributing to reducing the impact of heat on the performance of the optical engine 606.

[0117] Figure 7 This is a block diagram of a transceiver module 410 coupled between a fiber optic port 402 and a server-side port 602 in a switching enclosure 110, according to some embodiments. In some embodiments, the transceiver module 410 is housed in the switching enclosure 110 and includes a switching ASIC 608 and a plurality of optical engines 606 (FIG. 6). Each optical engine 606 can exchange signals with a full-duplex optical interconnect (e.g., a full-duplex fiber optic cable). Alternatively, each optical engine 606 can exchange signals with multiple full-duplex optical interconnects in a time-division multiplexing manner. Each optical engine 606 includes a transmitter 700 (e.g., a laser diode 704 and laser drive circuitry 706) and a receiver (e.g., one or more optical sensors 708). In one example, the laser diode 704 includes a vertical-cavity surface-emitting laser (VCSEL) diode. In another example, the laser diode 704 is different from a VCSEL diode. One or more optical sensors 708 are configured to receive an input signal 312 and convert it into an intermediate electrical signal 710, which is processed by a switching ASIC 608 to generate an electrical signal 314A. In some embodiments, the switching ASIC 608 receives the input electrical signal 312A and converts it into an electrical signal 712. The electrical signal 712 controls a laser driving circuit 706 to drive a laser diode 704 to emit an output signal 314 (an optical signal).

[0118] In some embodiments, the server rack 100 includes a plurality of optical engines 606 configured to generate a plurality of output signals 314 (as an optical signal). The transceiver module 410 includes a plurality of transmitters 700, namely a plurality of laser diodes 704 and a plurality of laser drive circuits 706 coupled to the plurality of laser diodes 704. The laser diodes 704 are configured to emit a set of optical signals 314 configured to be transmitted via a plurality of detachable optical interconnects 302. The plurality of laser drive circuits 706 are configured to receive a plurality of input signals 312, provide electrical signals 712 to drive the plurality of laser drive circuits 706, and generate a set of optical signals 314.

[0119] In some embodiments, the output signal of the switching box 110 includes a set of electrical signals 314A, and the input signal of the switching box 110 includes a set of optical signals 312. The transceiver module 410 includes multiple receivers (e.g., optical sensors 708) configured to work with the switching ASIC 608 to convert the set of optical signals 312 into a set of electrical signals 314A. The set of electrical signals 314A can utilize multiple electrical interconnects 304 (… Figure 3A The data is then further transmitted to one or more rack servers 120.

[0120] Figure 8 This illustrates a server group 800 comprising a plurality of servers 120 according to some embodiments. The plurality of servers 120 are coupled to each other and distributed across a plurality of server racks 100. The plurality of server racks includes a first server rack 100A. In some embodiments, a switching box 110 of the first server rack 100A includes a first set of ports 602A configured to receive a plurality of detachable electrical interconnects 304, each port of the first set of ports 602A being configured to exchange electrical signals with a corresponding rack server 120 mounted on a rack structure of the first server rack 100A. Alternatively or additionally, in some embodiments, the switching box 110 of the first server rack 100A further includes a second set of ports 602B coupled to a plurality of rack servers 120 on one or more alternative server racks 100B. Each alternative server rack 100B includes at least one rack server 120 electrically coupled to a corresponding port of the second set of ports 602B of the first server rack 100A. It is worth noting that, in various embodiments of the present invention, the switching box 110 of the first server rack 100A includes only the first set of ports 602A, only the second set of ports 602B, or a combination thereof.

[0121] Figure 9A This is a rear view of an example server rack 100 including a server cooling system 900 according to some embodiments. Figure 9BThis is a front view of another example server rack 100 including a server cooling system 900 according to some embodiments. The server cooling system 900 relies on liquid cooling. In some embodiments, the server rack 100 including the server cooling system 900 is applied in a data center to perform machine learning tasks, such as training deep neural networks or performing large language models (LLMs). The server rack 100 includes a plurality of slots 104 for receiving and supporting corresponding computing device modules 106 (e.g., GPU servers 120). In some embodiments, the server rack 100 further includes a cooling distribution module (CDM) disposed between two adjacent slots 104. In other words, each CDM is disposed below the bottom plate of a corresponding upper slot or above the top plate of a corresponding bottom slot. Each CDM includes a corresponding coolant channel embedded in a heat sink. When coolant (e.g., water) flows through the corresponding coolant channel of each CDM, the coolant at least partially removes heat absorbed by the heat sink from the adjacent computing device module 106. Reference Figure 9A and Figure 9B In this example, the CDM can be selectively set between two GPU servers or between a GPU server and switch box 110.

[0122] Each CDM has an inlet 912 and an outlet 914, and is coupled to a cooling distribution unit (CDU), which acts as an engine driving coolant through the cooling system 900, allowing coolant to be injected into the inlet 912 of each CDM and collected from the outlet 914 of each CDM. The CDU can regulate and control the flow of coolant and maintain the desired temperature and flow rate. In some embodiments ( Figure 9A In this configuration, CDMs can be arranged in parallel and coupled between the inlet and outlet of the CDU. In some embodiments (not shown), CDMs can be arranged in series and coupled between the inlet and outlet of the CDU. (See reference...) Figure 9A In some embodiments, each GPU server 120 or CDU includes one or more respective fans 902 located within the associated free slot space, each respective fan 902 being configured to enhance air circulation and increase heat dissipation in the slot 104 via air convection.

[0123] In some embodiments, the transceiver module 410 and its associated cooling structure 510 are disposed within the switching box 110. Alternatively, in some embodiments, the transceiver module 410 and its associated cooling structure 510 are disposed between the switching box 110 and the cooling distribution module 904. (See reference...) Figure 9B The inlet 502 and outlet 504 of the cooling structure 510 can be connected to one of the cooling distribution modules 904 disposed on the server rack 100. Coolant is injected through the cooling distribution module 904 and the cooling structure 510 to jointly dissipate the heat generated by the transceiver module 410 and the server 120.

[0124] refer to Figure 9A In some embodiments, the cooling distribution module 904 is coupled to an adjacent server 120 and dissipates heat generated by the adjacent server 120 as a whole. The rack structure of the server rack 100 further includes a server tray 104A (also a server slot) configured to receive the first rack server 120A. When coolant 506 ( Figure 5B When the coolant flows through the cooling structure 510, the cooling distribution module 904, located directly above the first rack server 120A, injects the second coolant and outputs the second coolant in parallel with the cooling structure 510 associated with the transceiver module 410, thereby allowing the second coolant to at least partially remove the heat generated by the first rack server 120A.

[0125] In some embodiments, the switch box 110 is optically coupled to the server 120 on the same server rack 100 using a plurality of server-side interconnects 304 (e.g., corresponding to optical communication channels). Alternatively, in some embodiments, the switch box 110 is electrically coupled to the server 120 on the same server rack 100 using a plurality of server-side interconnects 304 (e.g., corresponding to electrical communication channels).

[0126] refer to Figure 9B In some embodiments, on the front side of server rack 100, cooling distribution module 904 has multiple connectors 906 coupled to a cooling structure within first rack server 120A. More specifically, cooling structure 510 associated with transceiver module 410 includes a first cooling structure, and server tray 104A further includes a second cooling structure configured to inject and discharge a second coolant parallel to the first cooling structure, thereby allowing the second coolant to at least partially remove heat generated by first rack server 120A. In some embodiments, the first coolant 506 flowing through first cooling structure 510 is separated from the second coolant flowing through first rack server 120A before entering inlet 502 (e.g., Figure 9A (position A in the middle), and after leaving outlet 504, it merges with the second coolant (e.g. Figure 9A Position B in the middle.

[0127] In some embodiments, a coolant pump 908 is coupled between an inlet 502 and an outlet 504. A coolant controller 910 is coupled to the coolant pump 908 and configured to control the coolant pump 908 to push coolant 506 into the inlet 502 of the cooling structure 510 and to extract coolant 506 from the outlet 504 of the cooling structure 510. Furthermore, in some embodiments, the coolant pump 908 is disposed in a first tray 104-1 (e.g., the bottom tray) of the server rack 100, and the transceiver module 410 is disposed in a second tray 104-2 of the server rack 100, which is different from the first tray 104-1.

[0128] Figure 10 This is a flowchart of a method 1000 for providing a server rack 100 according to some embodiments. A rack structure for supporting one or more rack servers 120 is provided (operation 1002). A switching box 110 is provided (operation 1004). The switching box 110 is mechanically mounted on the rack structure and configured to receive a plurality of detachable optical interconnects 302. A transceiver module 410 is provided in the switching box (operation 1006), the transceiver module being configured to convert multiple input signals (e.g., ...) Figure 3A Input signal 312, Figure 3B The electrical signal 312A in the middle is converted into multiple output signals (e.g., Figure 3A Electrical signal 314A in Figure 3B The output signal 314 in the transceiver module 410 generates heat when converting multiple input signals. A cooling structure 510 is provided in the switching box 110 (operation 1008). The cooling structure 510 is coupled to the transceiver module 410. The cooling structure 510 includes an inlet 502 and an outlet 504 (operation 1010), configured to inject coolant 506 through the inlet 502 and output coolant 506 through the outlet 504, thereby allowing the coolant 506 to at least partially remove the heat generated by the transceiver module 410.

[0129] In some embodiments ( Figure 5A and Figure 5B The cooling structure 510 includes a metal plate with a contact surface that contacts the transceiver module 410 to absorb heat generated by the transceiver module 410. Furthermore, in some embodiments, the metal plate includes a coolant channel 512 sealed within the metal plate, with an inlet 502 and an outlet 504 each coupled to a respective edge of the metal plate and connected to a respective end of the coolant channel 512. The coolant channel 512 extends generally parallel to the contact surface from the inlet 502 to the outlet 504.

[0130] In some embodiments, the switch box 110 further includes a first set of ports (e.g. Figure 8 The first set of ports 602A of the server-side port 602 is configured to receive a plurality of detachable electrical interconnects 304, each of the first set of ports being configured to exchange electrical signals with a corresponding rack server 120 mounted on the rack structure.

[0131] In some embodiments, the switch box 110 further includes a second set of ports (e.g. Figure 8 The second set of ports 602B of the server-side port 602 is configured to receive a plurality of detachable electrical interconnects 304. The second set of ports is coupled to a plurality of rack servers 120 on one or more alternative server racks 100B. Each alternative server rack 100B includes at least one rack server 120 electrically coupled to a corresponding port of the second set of ports.

[0132] In some embodiments, the transceiver module 410 and the cooling structure 510 cannot be separated from each other by manual operation without tools. The rack structure includes a first tray 104-1 (also a first slot) configured to receive a switch box 110 including the transceiver module 410 and the cooling structure 510, allowing the switch box 110 to be removed from the server rack 100 and the transceiver module 410 and the cooling structure 510 to be replaced in the switch box 110.

[0133] In some embodiments, at least one of the transceiver module 410 and the cooling structure 510 is mechanically fixed to the exchange box 110 and cannot be separated from the exchange box 110 by manual operation without the use of tools.

[0134] In some embodiments ( Figure 6A The transceiver module 410 includes multiple optical engines 606 and a switching ASIC 608. The multiple optical engines 606 are configured to exchange optical signals with multiple detachable optical interconnects 302, and the switching ASIC 608 is configured to exchange electrical signals with the multiple optical engines 606.

[0135] In some embodiments ( Figure 6B and Figure 6C The transceiver module 410 includes a switching ASIC 608, and the switching box 110 further includes a plurality of optical engines 606 that are separate from and electrically coupled to the transceiver module 410. Furthermore, in some embodiments ( Figure 6B and Figure 6C The switching box 110 is configured to receive multiple detachable optical interconnects 302 via multiple fiber optic ports 402, each optical engine 606 being detachably coupled to a corresponding fiber optic port 402. Furthermore, in some embodiments ( Figure 6C The switching ASIC 608 of the transceiver module 410 is electrically coupled to a plurality of optical engines 606 via an electrical switching cable 614. The plurality of optical engines 606 further include digital signal processing (DSP) circuitry 612 (such as a second DSP block 612B) configured to exchange digital electrical signals with the switching ASIC 608 via the electrical switching cable 614.

[0136] In some embodiments ( Figure 7 The transceiver module 410 further includes multiple laser diodes 704 and multiple laser driver circuits 706 coupled to the multiple laser diodes 704. The laser diodes 704 are configured to emit a set of optical signals configured to be transmitted via multiple detachable optical interconnects 302. The multiple laser driver circuits 706 are configured to receive multiple input signals and provide electrical signals to drive the laser diodes 704 to generate a set of optical signals.

[0137] In some embodiments, the multiple output signals include a set of electrical signals, the multiple input signals include a set of optical signals, and the transceiver module 410 further includes multiple receivers (e.g., Figure 7 The optical sensor 708 is configured to convert a set of optical signals into a set of electrical signals for transmission to one or more rack servers 120 using multiple electrical interconnects 304.

[0138] In some embodiments, the server rack 100 further includes one or more rack servers 120 configured to receive a plurality of output signals. The plurality of input signals include a set of optical signals received via a plurality of detachable optical interconnects 302, and the plurality of output signals include a set of electrical signals configured to be transmitted to the one or more rack servers 120.

[0139] In some embodiments, the server rack 100 further includes one or more rack servers 120 configured to provide a plurality of input signals. The plurality of input signals include a set of electrical signals provided by the one or more rack servers 120, and the plurality of output signals include a set of optical signals transmitted via a plurality of detachable optical interconnects 302.

[0140] In some embodiments, the server rack 100 further includes one or more rack servers 120 configured to receive a plurality of output signals. The plurality of input signals include a set of input optical signals received via a plurality of detachable optical interconnects 302, and the plurality of output signals include a set of output optical signals configured to be transmitted to the one or more rack servers 120.

[0141] In some embodiments, server rack 100 further includes one or more rack servers 120 configured to provide a plurality of input signals. The plurality of input signals includes a set of input optical signals provided by the one or more rack servers 120, and the plurality of output signals includes a set of output optical signals transmitted via a plurality of detachable optical interconnects 302. Furthermore, in some embodiments, the one or more rack servers 120 include a plurality of graphics processing units (GPUs) configured to perform machine learning operations.

[0142] In some embodiments ( Figure 9A and Figure 9BThe coolant 506 includes a first coolant 506. The rack structure further includes a server tray 104A configured to receive the first rack server 120A. The cooling structure 510 includes a first cooling structure, and the first tray 104-1 further includes a second cooling structure configured to inject and exit the second coolant parallel to the first cooling structure, thereby allowing the second coolant to at least partially remove the heat generated by the first rack server 120A. The first coolant 506 separates from the second coolant before entering the inlet 502 and merges with the second coolant after exiting the outlet 504.

[0143] In some embodiments, the server rack 100 further includes a coolant pump 908 coupled between an inlet 502 and an outlet 504. Figure 9A ), and a coolant controller 910 coupled to a coolant pump 908. Figure 9A The coolant controller 910 is configured to control the coolant pump 908 to push coolant 506 into the inlet 502 of the cooling structure 510 and to extract coolant 506 from the outlet 504 of the cooling structure 510. Furthermore, in some embodiments, the coolant pump 908 is disposed in a first tray 104-1 of the server rack 100, and the transceiver module 410 is disposed in a second tray 104-2 of the server rack 100, which is different from the first tray 104-1.

[0144] In some embodiments, the plurality of detachable optical interconnects 302 have a data communication bandwidth greater than 1 Tb / s, and the transceiver module 410 has a power consumption level greater than 25 W.

[0145] In some embodiments, server rack 100 includes or is coupled to a plurality of panels configured to convert server rack 100 into server cabinet.

[0146] In some embodiments, the switch box 110 houses the transceiver module 410 and the cooling structure 510.

[0147] Figure 11 This is a flowchart illustrating a method 1100 for managing input data implemented on a server rack 100 (FIG. 1) according to some embodiments. Server rack 100 (FIG. 1) Figure 1A and Figure 1B This includes (operation 1102) a rack structure for supporting one or more rack servers 120, and a switch box 110 mechanically mounted on the rack structure. The switch box 110 mechanically receives (operation 1104) multiple detachable optical interconnects 302 and converts (operation 1106) multiple input signals 312 into multiple output signals 314. Figure 3A and Figure 3BThe switching box 110 further includes a transceiver module 410 and a cooling structure 510 coupled to the transceiver module 410. The transceiver module 410 generates heat (operation 1108) during the conversion of multiple input signals 312. The cooling structure 510 injects (operation 1110) coolant 506 through its inlet 502. Figure 5B The coolant 506 is output through the outlet 504 of the cooling structure 510, thereby allowing the coolant 506 to at least partially remove the heat generated by the transceiver module 410.

[0148] The terminology used in the various embodiments described herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used in the description of the various described embodiments and the appended scope of protection, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should be further understood that the term “and / or” as used herein refers to and covers any and all possible combinations of one or more of the associated listed items. It should be further understood that the terms “comprising,” “including,” “having,” and / or “possessing” as used herein specify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. Furthermore, it should be understood that although the terms “first,” “second,” etc., may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used only to distinguish one element from another.

[0149] As used herein, the term "if" may be interpreted alternatively as "when" or "in response to a determination" or "in response to a detection" or "according to a determination," depending on the context. Similarly, the phrase "if a determination" or "if 'the condition or event' is detected" may be interpreted alternatively as "when a determination is made" or "in response to a determination" or "when 'the condition or event' is detected" or "in response to the detection of 'the condition or event'" or "according to the determination that 'the condition or event' is detected," depending on the context.

[0150] The foregoing description is for illustrative purposes and has been described with reference to specific embodiments. However, the above illustrative discussion is not intended to be exhaustive or to limit the scope of patent protection to the precise forms disclosed. In view of the foregoing teachings, many modifications and variations are possible. The embodiments were chosen and described to best explain the operating principles and practical applications, thereby enabling those skilled in the art to implement them.

[0151] While the various figures illustrate multiple logical stages in a specific order, stages independent of the order can be reordered, and other stages can be combined or broken down. Although some reorderings or other groupings are specifically mentioned, others will be obvious to those skilled in the art, and therefore the orders and groupings presented herein are not an exhaustive list of alternatives. Furthermore, it should be recognized that these stages can be implemented using hardware, firmware, software, or any combination thereof.

Claims

1. A server rack, characterized by Comprise: a rack structure supporting one or more rack servers; a switch box mechanically mounted on the rack structure, wherein the switch box is configured to receive a plurality of detachable optical interconnects, and further comprises: a transceiver module configured to convert a plurality of input signals into a plurality of output signals and generate heat while converting the plurality of input signals; and a cooling structure coupled to the transceiver module, wherein the cooling structure comprises an inlet and an outlet, and is configured to inject a coolant through the inlet and output the coolant through the outlet, thereby allowing the coolant to at least partially carry away the heat generated by the transceiver module. The cooling structure comprises a metal plate having a contact surface, the metal plate being in contact with the transceiver module through the contact surface to absorb the heat generated by the transceiver module.

2. The server rack of claim 1, wherein, The metal plate comprises a coolant channel sealed within the metal plate, the inlet and the outlet being respectively coupled to a corresponding edge of the metal plate and connected to a corresponding end of the coolant channel, the coolant channel extending substantially parallel to the contact surface from the inlet to the outlet.

3. The server rack of claim 2, wherein, The switch box further comprises a first set of ports configured to receive a plurality of detachable electrical interconnects, each port of the first set of ports being configured to exchange electrical signals with a corresponding rack server mounted on the rack structure.

4. The server rack of claim 1, wherein, The switch box further comprises a second set of ports configured to receive a plurality of detachable electrical interconnects, the second set of ports being coupled to a plurality of rack servers on one or more alternative server racks, each of the alternative server racks comprising at least one rack server electrically coupled to a corresponding port of the second set of ports.

5. The server rack of claim 1, wherein, The transceiver module and the cooling structure are not separable from each other by manual operation without using tools, the rack structure comprising a first slot configured to receive the switch box comprising the transceiver module and the cooling structure, allowing the switch box to be detached from the server rack and replaced in the switch box with the transceiver module and the cooling structure.

6. The server rack of claim 1, wherein, At least one of the transceiver module and the cooling structure is mechanically fixed to the switch box and cannot be separated from the switch box by manual operation without using tools.

7. The server rack of claim 1, wherein, The transceiver module comprises a plurality of optical engines configured to exchange optical signals with the plurality of detachable optical interconnects and a switch application specific integrated circuit configured to exchange electrical signals with the plurality of optical engines.

8. The server rack of claim 1, wherein, The transceiver module comprises a switch application specific integrated circuit, and the switch box further comprises a plurality of optical engines separate from the transceiver module and electrically coupled to the transceiver module.

9. The server rack of claim 1, wherein, The switch box is configured to receive the plurality of detachable optical interconnects through a plurality of fiber ports, each of the optical engines being detachably coupled to a corresponding fiber port.

10. The server rack of claim 9, wherein, The switch application specific integrated circuit of the transceiver module is electrically coupled to the plurality of optical engines through an electrical switch cable, and the plurality of optical engines further comprises a digital signal processing block configured to exchange digital electrical signals with the switch application specific integrated circuit through the electrical switch cable.

11. The server rack of claim 10, wherein, The plurality of output signals comprises a set of optical signals, and the transceiver module further comprises:

12. The server rack of claim 1, wherein, ​ a plurality of laser diodes configured to emit the set of optical signals for transmission through the plurality of detachable optical interconnects; and a plurality of laser driver circuits coupled to the plurality of laser diodes, wherein the plurality of laser driver circuits are configured to receive the plurality of input signals and provide electrical signals to drive the plurality of laser diodes to generate the set of optical signals.

13. The server rack of claim 1, wherein, the plurality of output signals comprise a set of electrical signals, the plurality of input signals comprise a set of optical signals, and the transceiver module further comprises: a plurality of receivers configured to convert the set of optical signals to the set of electrical signals for transmission to the one or more rack servers using a plurality of electrical interconnects.

14. The server rack of claim 1, wherein, further comprising one or more rack servers configured to receive the plurality of output signals, wherein the plurality of input signals comprise a set of optical signals received through the plurality of detachable optical interconnects, and the plurality of output signals comprise a set of electrical or optical signals configured to be transmitted to the one or more rack servers.

15. The server rack of claim 1, wherein, further comprising one or more rack servers configured to provide the plurality of input signals, wherein the plurality of input signals comprise a set of electrical or optical signals provided by the one or more rack servers, and the plurality of output signals comprise a set of optical signals transmitted through the plurality of detachable optical interconnects.

16. The server rack of claim 1, wherein: the coolant comprises a first coolant; the rack structure further comprises a server tray configured to receive a first rack server; the cooling structure comprises a first cooling structure, and the server tray further comprises a second cooling structure configured to inject a second coolant and output the second coolant in parallel with the first cooling structure such that the second coolant at least partially carries away heat generated by the first rack server; and the first coolant is separated from the second coolant before entering the inlet and combined with the second coolant after exiting the outlet.

17. The server rack of claim 1, wherein, further comprising: a coolant pump coupled between the inlet and the outlet; and a coolant controller coupled to the coolant pump, wherein the coolant controller is configured to control the coolant pump to push the coolant into the inlet of the cooling structure and pull the coolant out of the outlet of the cooling structure. wherein the coolant pump is disposed in a first tray of the server rack, and the transceiver module is disposed in a second tray of the server rack, the second tray being different from the first tray.

18. The server rack of claim 1, wherein: the plurality of detachable optical interconnects have a data communication bandwidth greater than 1 terabit per second, and the transceiver module has a power consumption level greater than 25 W; the server rack comprises or is coupled to a plurality of panels configured to convert the server rack to a server cabinet; and the switch box houses the transceiver module and the cooling structure. further comprising: ​ 19. A modulator device, characterized by ​ a transceiver module disposed within a switch box configured to receive a plurality of removable optical interconnects, the transceiver module configured to convert a plurality of input signals to a plurality of output signals and generate heat during the conversion of the plurality of input signals; and a cooling structure coupled to the transceiver module, wherein the cooling structure includes an inlet and an outlet, the cooling structure configured to inject a coolant through the inlet and output the coolant through the outlet, thereby allowing the coolant to at least partially remove the heat generated by the transceiver module.

20. A switching box, characterized by comprising: a transceiver module configured to convert a plurality of input signals to a plurality of output signals and generate heat during the conversion of the plurality of input signals; and a cooling structure coupled to the transceiver module, wherein the cooling structure includes an inlet and an outlet, the cooling structure configured to inject a coolant through the inlet and output the coolant through the outlet, thereby allowing the coolant to at least partially remove the heat generated by the transceiver module; wherein the switch box is configured to receive a plurality of removable optical interconnects.

Images