Computer cabinet and related computing system

Abstract

Aspects of the present disclosure relate to a computer cabinet for a computing system and a computing system having one or more computer cabinets. The computer cabinet may include a power plane, a computing plane, and a host plane for power conversion. The computer cabinets disclosed herein may be modular and scalable. A computer cabinet may be configured such that components thereof may be hot-swapped.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 377,942, filed on September 30, 2022, entitled “SYSTEM TRAY AND CABINET FOR A COMPUTING SYSTEM,” the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0003] The present disclosure relates generally to computing systems, and more particularly to computing systems having one or more computer cabinets. Background Art

[0004] Certain computing systems may be used and / or specifically configured for high performance computing and / or computing intensive applications, such as neural network training, neural network reasoning, machine learning, artificial intelligence, complex simulations, etc. In some applications, a computing system may be used to perform neural network training. For example, such neural network training may generate data for an autonomous driving system of a vehicle (e.g., a car), other autonomous vehicle functions, or advanced driver assistance system (ADAS) functions.

[0005] In high performance computing systems, high speed connectivity, desired power performance, and dense integration are generally required. In high performance computing systems, there may be a large number of parts and connections between parts. There are technical challenges associated with scaling certain high performance computing systems. Summary of the invention

[0006] The innovations described in the claims each have several aspects, no single one of which is solely responsible for its desirable attributes. Without limiting the scope of the claims, some of the prominent features of the disclosure will now be briefly described.

[0007] In one aspect, the technology described herein relates to a computer cabinet. The computer cabinet may include a first section, the first section including a first power plane, a first computing plane configured to receive power from the first power plane and perform calculations, and a first host plane in communication with the first computing plane. The computer cabinet may include a second section, the second section including a second power plane, a second computing plane configured to receive power from the second power plane and perform calculations, and a second host plane in communication with the second computing plane. The second section may operate independently of the first section. The computer cabinet may include a cabinet frame. The first section and the second section are positioned within the cabinet frame.

[0008] In one embodiment, the first section is stacked vertically with the second section.

[0009] In one embodiment, the first compute plane, the first host plane, the second power plane, and the second host plane are positioned between the first power plane and the second power plane.

[0010] In one embodiment, when the first compute plane is operating, at least a portion of the first power plane is hot-swappable.

[0011] In one embodiment, the first compute plane is hot-pluggable while the second compute plane is operating.

[0012] In one embodiment, the first host plane is hot-pluggable while the second compute plane is operating.

[0013] In one embodiment, the computing cabinet may include one or more redundant connections between the first power plane and the first computing plane.

[0014] In one embodiment, the computer cabinet may include an interface including a connection to an external power source and a coolant inlet. The connection to the external power source may be electrically connected to the first power plane and the second power plane. The coolant inlet is in fluid communication with the first computing plane and the second computing plane.

[0015] In one embodiment, the first computing plane includes a computing tray and a plurality of computing tiles positioned on the computing tray, each of the plurality of computing tiles including a plurality of dies and a cooling solution integrated with the plurality of dies.

[0016] In one embodiment, the first power plane includes a plurality of power trays configured to convert external power to power for the first compute plane.

[0017] In one embodiment, individual power trays of the plurality of power trays are hot-swappable while other power trays of the plurality of power trays are operating.

[0018] In one embodiment, a computer cabinet may include a connector extending from a first computing plane and configured to connect with a computing plane of an adjacent cabinet.

[0019] In one embodiment, the computer cabinet may include a blind cooling connector on a side of the cabinet frame and a blind power connector on a side of the cabinet frame, wherein the first computing plane is connected to the blind cooling connector and the blind power connector when inserted into the computer cabinet.

[0020] In one aspect, the technology described herein relates to a computing system. The computing system may include a first computing cabinet, the first computing cabinet including a first power plane, a first computing plane configured to receive power from the first power plane and perform calculations, and a first host plane in communication with the first computing plane. The computing system may include a second computing cabinet, the second computing cabinet including a second power plane, a second computing plane configured to receive power from the second power plane and perform calculations, and a second host plane in communication with the second computing plane. The first computing plane may be connected to the second computing plane via a connector that extends through a side of the first computing cabinet and a side of the second computing cabinet. The position of the first computing plane may be aligned with the position of the second computing plane.

[0021] In one embodiment, the connector connecting the first computing plane and the second computing plane is a blind connector.

[0022] In one embodiment, the first computer cabinet may further include a third computing plane. The second computer cabinet may further include a fourth computing plane. The third computing plane may be connected to the fourth computing plane via a second connector that extends through a side of the first computer cabinet and a side of the second computer cabinet. The position of the third computing plane may be aligned with the position of the fourth computing plane.

[0023] In one embodiment, the first computer cabinet further includes a third power plane configured to provide power to the third computing plane, and a third host plane in communication with the third computing plane. The second computer cabinet further includes a fourth power plane configured to provide power to the fourth computing plane, and a fourth host plane in communication with the fourth computing plane.

[0024] In one embodiment, the first power plane includes a plurality of power trays, and a single power tray of the plurality of power trays is hot-swappable.

[0025] In one embodiment, the first computing cabinet further includes one or more redundant connections between the first power plane and the first computing plane.

[0026] In one embodiment, at least one of the first power plane and the first host plane is hot-swappable.

[0027] In one embodiment, the first computer cabinet and the second computer cabinet are each independently operable.

[0028] In one embodiment, the computing system further includes a third computing cabinet, the third computing cabinet including a third power plane, a third computing plane configured to receive power from the third power plane and perform computing, and a third host plane communicating with the third computing plane. The first computing plane is connected to the third computing plane via a third connector, the third connector extending through the second side of the first computing cabinet and the side of the third computing cabinet. The first computing plane, the second computing plane, and the third computing plane are connected.

[0029] In one embodiment, the computing system further comprises a third computing cabinet connected to the second computing cabinet only by a connector extending from a second side of the second computing cabinet. The third computing cabinet comprises a third connector extending from a second side of the third computing cabinet for connecting to a fourth computing cabinet.

[0030] To summarize the present disclosure, certain aspects, advantages, and novel features of the innovations are described herein. It should be understood that not all such advantages may be achieved according to any particular embodiment. Thus, the innovations may be embodied or implemented in a manner that achieves or optimizes one advantage or a group of advantages taught herein without necessarily achieving other advantages taught or suggested herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The detailed description will now be described with reference to the following figures, which are provided by way of example and not limitation.

[0032] Figure 1A , Figure 1B and Figure 1C A computer cabinet for a computing system is illustrated according to one embodiment.

[0033] Figure 2A , Figure 2B and Figure 2C Illustrated is a front view of a connection between two computer cabinets using blind connectors according to one embodiment.

[0034] Figure 3 Illustrated is a perspective view of a computer cabinet according to one embodiment.

[0035] Figure 4 Illustrated is a perspective view of two computer cabinets of a computing system according to one embodiment.

[0036] Figure 5A and Figure 5B Illustrated is a connection between a first computing tray of a first computing cabinet and a second computing tray of a second computing cabinet using a blind connector according to one embodiment.

[0037] Figure 6 Illustrated is a view of a power tray according to one embodiment.

[0038] Figure 7 Illustrated are multiple power trays tied together for active cooling according to one embodiment.

[0039] Fig. 8A An array of power trays is illustrated according to one embodiment.

[0040] Figure 8B and Figure 8C Illustrated is an array of two power trays positioned in a computer cabinet according to one embodiment.

[0041] Fig. 9 A power bus integrated into a cabinet frame is illustrated according to one embodiment.

[0042] Fig. 10A and Fig. 10B A computing tray is illustrated according to one embodiment.

[0043] Fig.11 A computing tray is illustrated positioned vertically relative to a host computer according to one embodiment. DETAILED DESCRIPTION

[0044] The following detailed description of certain embodiments presents various descriptions of specific embodiments. However, the innovations described herein may be embodied in a variety of different ways, for example, as defined and covered by the claims. In this specification, reference is made to the accompanying drawings in which the same reference numerals and / or terms may indicate the same or functionally similar elements. It should be understood that the elements illustrated in the figures are not necessarily drawn to scale. In addition, it should be understood that certain embodiments may include more elements than illustrated in the drawings and / or a subset of the elements illustrated in the drawings. In addition, some embodiments may combine any suitable combination of features from two or more drawings.

[0045] As described above, certain computing systems may be used and / or specifically configured for high performance computing and / or computing intensive applications, such as neural network training, neural network reasoning, machine learning, artificial intelligence, complex simulations, etc. In some applications, a computing system may be used to perform neural network training. For example, such neural network training may generate data for an autonomous driving system of a vehicle (e.g., a car), other autonomous vehicle functions, or advanced driver assistance system (ADAS) functions.

[0046] Certain computing systems may include various levels of hierarchy to perform computing tasks. For example, a computing system may include chips, computing tiles each including multiple chips packaged together and integrated with a cooling solution, computing trays including arrays of connected computing tiles, power supplies that deliver power to various components, and computing cabinets each including one or more computing trays (one or more) and one or more power supplies (one or more).

[0047] The present disclosure relates to a novel computing system. The computing system described herein can be configured for high-performance computing applications. The computing system described herein can include hot-swappable components that can be removed and / or inserted into the computing system when the computing system is actively powered and operated. For example, hot-swappable components can include computing trays and power trays that can be removed and / or replaced from the computing system without shutting down the computing system. Each computer cabinet can have hot-swappable power. The computer cabinet can include two half-cabinet sections that work independently of each other. Half of the computing, power, and host can be hot-swapped without affecting the other half. The cabinet can include blind connections to enable such hot-swapping of components without interacting with the power connection, thereby ensuring safety during hot-swapping. As described herein, a blind connection can refer to a connection that can be coupled to a component without direct user access. The computing system described herein can continue to operate when various components fail.

[0048] A computing system with independent computer cabinets is disclosed. Each computer cabinet can have its own power conversion, host and computing platform. Each computer cabinet can function completely independently when receiving power and cooling.

[0049] The computer cabinets disclosed herein are modular and expandable. Each computer cabinet can be connected to an adjacent cabinet without any external parts. Blind connections are added to the side of each computer cabinet. The opposite side of the computer cabinet can be matched with another computer cabinet to continuously expand the computing system. The computer cabinet can be used alone or in combination with one or more other cabinets to expand as needed for computing tasks (e.g., training of different models) to form as large a computing plane as possible.

[0050] The computer cabinets disclosed herein may have power redundancy. Each computer cabinet may have built-in power redundancy to prevent a faulty component from affecting the functionality of other components of the computer cabinet.

[0051] Figure 1A , Figure 1B and Figure 1CA computer cabinet 100 for a computing system according to one embodiment is illustrated. The computer cabinet 100 can operate as a fully functional computing system, or can be connected to one or more other computer cabinets to increase the computing power of the computing system. The computer cabinet 100 itself can be fully functional when powered and cooled. The computer cabinet 100 receives power and cooling. The computer cabinet 100 distributes power and cooling to various components housed in the structure of the computer cabinet 100.

[0052] like Figure 1A As shown, the computer cabinet 100 can be divided into a first section 101a and a second section 101b, each section including a power plane 104, a computing plane 102, and a host plane 106. In some embodiments, the first section 101a and the second section 101b can be operated independently of each other. In this way, the components of the first section 101a can be hot-swapped without affecting the operation of the second section 101b.

[0053] Each power plane 104 receives high voltage input power and converts the input power to meet the specifications of various other computer cabinet 100 components (e.g., computing trays 112). Each power plane 104 may include an array of power trays 114. Each power tray 114 may operate in conjunction with other power trays 114 of the power plane 104 to meet the specifications of a computer cabinet 100 section, such as the first section 101a or the second section 101b. In this way, if a first power tray 114 fails, the other power trays 114 of the power plane 104 can compensate for the failure before the failed first power tray 114 is replaced. The power tray 114 can be removed and replaced from the power plane 104 without deactivating the power plane 104 and / or without disconnecting the power plane 104 from the high voltage input power.

[0054] Each computing plane 102 includes a computing tray 112, which includes one or more computing tiles. Each computing tile includes a plurality of chips or tube cores packaged together and integrated with one or more cooling solutions. In some such applications, the computing tile may include a system on a chip, which includes an array of tube cores. In some such applications, the cold plate may be integrated with the system on a chip. The computing plane 102 may be used for and / or specially configured for high-performance computing and / or computing-intensive applications, such as neural network training, neural network reasoning, machine learning, artificial intelligence, complex simulation, etc. In some applications, the computing plane 102 may be used to perform neural network training. For example, such neural network training may generate data for an autonomous driving system, other autonomous driving vehicle functions, or advanced driver assistance system (ADAS) functions of a vehicle (e.g., a car). The computing plane 102 may be operated alone and / or may be connected to a computing plane 102 in one or more other computing cabinets 100 to increase the computing power of the computing system. For example, multiple computing planes 102 may be connected to expand the computing plane as needed to provide a higher capacity computing system. With higher capacity, a computing system can (1) perform computing tasks of higher complexity (e.g., training more demanding models) and / or (2) run a greater number of computing tasks in parallel with each other.

[0055] Each host board 106 includes a host tray 116. The host tray 116 can implement the ingestion processing of the computing plane 102 of the same segment of the computer cabinet segment (such as the first segment 101a or the second segment 101b). The host tray 116 may include peripheral component express interconnect (PCIe) connectivity to the interface processor. The host tray 116 may provide video decoder support. In some embodiments, the host tray 116 can operate in an x86 Linux environment.

[0056] like Figure 1A As shown, the computer cabinet 100 includes a first computing tray 112 vertically positioned above a first host tray 116 and a second computing tray 112 vertically positioned above the second host tray 116. As shown, the first host tray 116 can be positioned between the first computing tray 112 and the second computing tray 112. Power trays 114 can be included at the top and bottom of the computer cabinet 100 and are vertically positioned relative to the computing trays 112 and the host trays 116.

[0057] Figure 1B Illustration from the front view Figure 1A A computer cabinet 100, and Figure 1C The computer cabinet 100 is shown in a rear view. Figure 1B and Figure 1CAs shown, the computer cabinet 100 may include a blind connector 120, a coolant interface 122, a power interface 121, a coolant distribution system 124, and a power bus 126. The power interface 121 may include a connection to an external power source. The external power source may be a high voltage power with a sufficient power rating to power the computer cabinet 100. The power tray 114 may receive power from the power interface 121. The coolant interface 122 may include a connection to an external cooling source. The external cooling source may be a connection to a fluid source (e.g., a coolant) that may be distributed throughout the computer cabinet 100 to help maintain a sufficiently low operating temperature.

[0058] The coolant distribution system 124 can deliver coolant to and from the interface 122, and distribute the coolant to components of the computer cabinet 100, such as the power tray 114, the compute tray 112, and the host tray 116. The coolant distribution system 124 can also output coolant from the computer cabinet 100. The coolant distribution system 124 can include one or more hoses, one or more manifolds, one or more coolant connections, etc., or any suitable combination thereof, to facilitate the flow of coolant within the computer cabinet 100.

[0059] The power bus 126 may include electrical conductors and / or electrical connection points. The power bus 126 may transmit the converted power from the power tray 114 to other cabinet components, such as to the compute tray 112 and the host tray 116. The power bus 126 may include redundant electrical connections. For example, the compute tray 112 may be connected to the power tray 114 via multiple electrical connections. In this way, if the power tray 114 fails or is removed, the compute tray 112 may receive power from one or more other power trays 114.

[0060] like Figure 1C As shown, the coolant distribution system 124 and the power bus 126 can be positioned at the rear of the computer cabinet. As will be described, various components of the computer cabinet 100 can include blind connectors that allow the components to be connected to the coolant distribution system 124 and the power bus 126 without requiring a user to make manual connections. For example, when the computing tray 112 is fully inserted into the computer cabinet 100, the blind cooling connector can be inserted into the coolant distribution system 124, and the blind power connector can be inserted into the power bus 126 at the rear of the computer cabinet 100.

[0061] The computer cabinet 100 includes a cabinet frame 108. The cabinet frame 108 provides structural support for various components, such as support rails for the compute tray 112, the power tray 114, and the host tray 116. The coolant distribution system 124, the power bus 126, and the interface 122 can be integrated into the cabinet frame 108. In this way, physical support for the components and connections to the coolant distribution system 124 and the power bus 126 can be established in a single action. For example, when the compute tray 112 is fully inserted into the computer cabinet 100, the compute tray 112 can be physically coupled to the cabinet frame 108 and connected to the coolant distribution system 124 and the power bus 126.

[0062] The blind connector 120 can connect a first computing tray 112 of a first computing cabinet 100 to a second computing tray 112 of a second computing cabinet 100, thereby allowing for a computing plane 102 of two computing cabinets 100. In some cases, the length of the connection between the computing trays 112 may result in a loss of computing power. In this way, the connected computing trays 112 of adjacent computing cabinets 100 can have increased computing power the closer the connected computing trays 112 are located. To facilitate a tighter connection, the blind connector 120 can enable the computing trays 112 of adjacent computing cabinets 100 to be connected blindly and / or without physically contacting the blind connector 120. The blind connector 120 will be described in more detail below.

[0063] Figure 2A , Figure 2B and Figure 2C Illustrated is a front view of a connection between two computer cabinets 100 using a blind connector 120 according to one embodiment. Figure 2A and Figure 2B The blind connector 120 is shown spanning the first cabinet frame 108a and the second cabinet frame 108b before the compute tray is positioned within the cabinet frames 108a, 108b. Figure 2A and Figure 2B As shown, there may be minimal space between the first cabinet frame 108a and the second cabinet frame 108b. In some embodiments, there may be no space between the first cabinet frame 108a and the second cabinet frame 108b. Thus, it may be difficult or impossible to access the blind connector 120 from the outside of the first cabinet frame 108a or the second cabinet frame 108b. In addition, although in Figure 2A or Figure 2B 108a and 108b may include components that occupy a majority of the space within the cabinet frames 108a and 108b, such as computer trays 112, power trays 114, and host trays 116. As such, the blind connectors 120 inside the first cabinet frame 108a or the second cabinet frame 108b may be difficult to access.

[0064] Figure 2B The illustrated cabinet frames 108a and 108b each have a computing tray 112 positioned therein. Blind connectors 120 may connect computing trays 112 between adjacent computing cabinets, such as Figure 2C As shown, there is no need to directly contact the blind connector 120. Although Figure 2C One computing tray 112 is shown in each cabinet frame 108a, 108b, but two or more computing trays 112 may be included in a single computer cabinet. Examples of computer trays 112 connected using blind connectors 120 are shown in FIG. Figure 5A and Figure 5B and discussed in more detail below.

[0065] Figure 3 Illustrated is a perspective view of a single computer cabinet according to one embodiment. Figure 3 Picture shows Figure 2A A perspective view of the second cabinet frame 108b is shown, wherein the blind connector 120 extends out of the second cabinet frame 108b. Figure 3 As shown, the blind connector 120 may include multiple sets of connectors. For example, Figure 3 Six groups of connectors for blind connectors 120 are shown extending from the cabinet frame 108b. Figure 3 In FIG. 1 , two groups of three correctors extend from the cabinet frame 108 b. Each group of connectors can be connected to a corresponding computing tile located on a computing tray 112. Figure 3 The blind connector 120 shown can connect to three compute tiles on the first compute tray 112 and three compute tiles on the second compute tray 112. The number of connector groups of the blind connector 120 can vary based on the number of compute tiles connected by the blind connector 120.

[0066] Figure 4 Illustrated is a perspective view of two computer cabinets according to one embodiment. Figure 4 Picture shows Figure 2A A perspective view of a first cabinet frame 108a and a second cabinet frame 108b, wherein a blind connector 120 extends from a side of the first cabinet frame 108a into the second cabinet frame 108b and extends out from an opposite side of the cabinet frame 108a. Figure 4 108a and 108b and are connected by blind connectors 120. Any suitable number of computer cabinets 100 may be connected in this manner. Such connections and modular and self-sufficient cabinets may enable a scalable computing system.

[0067] Figure 5A and Figure 5B FIG. 1 illustrates the connection of a first computing tray 112a and a second computing tray 112b using a blind connector 120 according to one embodiment. Figure 5A , the blind connector 120 is shown extending into the first cabinet frame 108a and the second cabinet frame 108b. The blind connector 120 includes a first connection interface 502a and a second connection interface 502b. The first connection interface 502a and the second connection interface 502b can each include a plurality of connectors that can be electrically coupled to corresponding connectors on the computing tray 112, such as Fig. 10B The inter-tray tile connector 1008. The first connection interface 502a and the second connection interface 502b may include an actuator that allows the first connection interface 502a and the second connection interface 502b to be switched between a connected position and a disconnected position.

[0068] When the first connection interface 502a and the second connection interface 502b are in the disconnected position, the blind connector 120 can be positioned outside the mounting path of the computer trays 112 in the first cabinet frame 108a and the second cabinet frame 108b so that the computer trays 112 can be inserted into the first cabinet frame 108a and the second cabinet frame 108b. When the first connection interface 502a and the second connection interface 502b are in the connected position, the blind connector 120 can be positioned in the mounting path of the computing trays 112 and / or coupled to the computing trays 112 in the first cabinet frame 108a and the second cabinet frame 108b.

[0069] refer to Figure 5B , illustrates that the blind connector 120 extends into the first cabinet frame 108a and the second cabinet frame 108b and is coupled to the first computing tray 112a and the second computing tray 112b. Figure 5B Some hardware of the first computing tray 112a and the second computing tray 112b are omitted, such as Fig. 10B The first computing tray 112a is first inserted into the first cabinet frame 108a, and the first connection interface 502a can be actuated downward to the disconnected position. When the first computing tray 112a is fully inserted into the first cabinet frame 108a, the first connection interface 502a can be actuated upward to the connected position to couple the first connection interface 502a to the first computing tray 112a.

[0070] Figure 6 A view of a power tray 114 is shown according to one embodiment. For example, Figure 6 The power tray 114 may implement Figure 1AThe power tray 114 may include a coolant inlet 602, a coolant outlet 604, and a power handle 606 positioned on a first end of the power tray 114. The power tray 114 may include a blind power connector 610 positioned on a second end of the power tray 114, wherein the second end is opposite the first end. The first end may be a front end, and the second end may be a rear end.

[0071] The coolant inlet 602 may connect one or more internal coolant manifolds of the power tray 114 to a coolant source, such as Figure 1C The internal coolant manifolds may distribute coolant throughout the power tray 114 to actively cool the power tray 114. The coolant outlets 604 may be connected to one or more internal coolant manifolds of the power tray 114 and discharge the coolant to a coolant destination, such as Figure 1C A coolant distribution system 124 is provided.

[0072] The blind power connector 610 can be electrically coupled to and physically coupled to a power source and / or deliver power to a power destination. For example, the blind power connector 610 can receive high voltage input power and provide the input power to the power tray 114. The power tray 114 can perform power conversion and deliver the converted power to the power bus 126 for use by components of the computer cabinet 100 (such as the compute tray 112 and / or the host tray 116). The power tray 114 may include various internal electronic components, such as power converters, capacitors, resistors, inductors, transistors, etc., or any suitable combination thereof. In some embodiments, the internal electrical components allow the power tray 114 to be inserted into an actively powered computer cabinet 100 without damaging the power tray 114 or other components of the computer cabinet 100.

[0073] Figure 7 FIG. 1 illustrates a plurality of power trays 114 connected together for active cooling according to one embodiment. Figure 7 As shown, the coolant can be taken from Figure 1C The coolant distribution system 124 ( Figure 7 14) is delivered to a coolant connector 702 located between the power trays 114. In some embodiments, a coolant inlet manifold may be located between the power trays 114 to deliver coolant from a coolant distribution system 124 located at the rear of the two power trays 114. In some embodiments, a coolant outlet manifold may be located between the power trays 114 to discharge coolant to a coolant distribution system 124 located at the rear of the two power trays 114.

[0074] like Figure 7As shown, the coolant connector 702 includes a coolant input connector and a coolant output connector. The coolant can be delivered from the coolant input connector to the coolant inlet 602 ( Figure 7 The discharged coolant can be discharged from the coolant outlet 604 ( Figure 7 ) is transmitted to the coolant output connector.

[0075] like Figure 7 As shown, each power tray 114 may have a blind power connector 610 positioned at the rear of the power tray 114. In some embodiments, in the same operation, the blind power connector 610 may be connected to the power bus 126 and the coolant connector 702 may be connected to the coolant distribution system 124. For example, when the power tray 114 is inserted into the cabinet frame 108, the blind power connector 610 may be connected to the power bus 126 and the coolant connector 702 may be connected to the coolant distribution system 124 at the same time.

[0076] Fig. 8A , Figure 8B and Figure 8C An array of power trays 800 is shown in accordance with one embodiment. The array of power trays 800 may include a plurality of power trays 114. For example, the array of power trays 800 may implement Figure 1A The power plane 104. Figure 8B As shown, the array of power trays 800 can be inserted into the cabinet frame 108 for use in the computer cabinet 100. Each power tray 114 of the array of power trays 800 can be operated individually. For example, a single power tray 114 can fail or be removed from the array of power trays 800 without causing the array of power trays 800 to fail.

[0077] The array 800 of power trays may include redundant power trays 114. For example, a subset of the power trays 114 of the array 800 of power trays may be operated to meet the power specifications of the computer cabinet 100. Figure 8C As shown, the cabinet frame 108 can accommodate more than one array of power trays 800. For example, a first array of power trays 800 can power a first section of the computer cabinet 100, and a second array of power trays 800 can power a second section of the computer cabinet 100.

[0078] Fig. 9A power bus 126 is illustrated as being integrated into the cabinet frame 108 according to one embodiment. As described above, the power bus 126 may include electrical conductors and electrical connection points. The power bus 126 may also transmit converted power from the power tray 114 to other cabinet components, such as the compute tray 112 and the host tray 116. The power bus 126 may include one or more redundant electrical connections. For example, the compute tray 112 may be connected to the power tray 114 via multiple electrical connections. In this way, if a single power tray 114 (or a group of power trays 114) fails or is removed, the compute tray 112 may receive power from other power trays 114.

[0079] The power bus 126 may include a plurality of connection points. The connection points may be configured to couple with power connections from various components, such as a blind power connector 610 (eg, Figure 6 ) and the blind computing connector 1010 of the computing tray 112 (as discussed in Fig. 10B discussed in ).

[0080] Fig. 10A and Fig. 10B An example computing tray 112 according to one embodiment is illustrated. The computing tray 112 may include an array of computing tiles 1002 connected together and supported by the computing tray 112. In some embodiments, each computing tile 1002 includes a system on a chip, which includes an array of tube cores integrated with a cooling solution (e.g., a cold plate). For example, the system on a chip may include 16, 25, 36, or 49 tube cores, which are arranged to perform computing functions in various applications. According to some applications, the system on a chip may be positioned between a cold plate and another cooling component to dissipate heat, remove heat, or otherwise reduce the temperature of the components of the computing tile 1002 during operation. In neural network training applications, the computing tile 1002 may be referred to as a training tile. Each computing tile 1002 of the computing tray 112 can be operated separately. In this way, if the computing tile 1002 of the computing tray 112 fails and / or is removed from the computing tray 112, the computing tray 112 can continue to operate. Any suitable number of computing tiles 1002 may be connected to each other on computing tray 112. For example, Fig. 10A Six compute tiles 1002 are illustrated that are connected to each other.

[0081] The computing tray 112 can have high computing capabilities. For example, the computing tray 112 can perform more than 50 petaflops (PFLOPS). In some applications, the computing tray can perform in the range of 50 PFLOPS to 200 PFLOPS.

[0082] The computing tray 112 may include an intra-tray signal delivery cable 1004 to facilitate communication between each computing tile 1002 and / or the connectors 1008 of the computing tiles 1002. The intra-tray signal delivery cable 1004 may include one or more redundant connections. For example, the computing tiles 1002 may be connected together via multiple intra-tray signal delivery cables 1004. In this way, if the intra-tray signal delivery cable 1004 fails and / or is removed, and / or the computing tile 1002 fails and / or is removed, the computing tray 112 may continue to operate.

[0083] In the computing tray 112, adjacent computing tiles 1002 are connected to each other through intra-tray signal delivery cables 1004. If a computing tile 1002 fails, other computing tiles 1002 on the computing tray 112 may still function. For example, adjacent computing tiles 1002 may route signals around the failed computing tile 1002 to functional computing tiles (one or more) to perform computing tasks and / or route signals around the failed computing tile 1002.

[0084] refer to Fig. 10B , the computing tile 1002 may include a connector 1008 around its edge. The connector 1008 of the computing tile 1002 may be connected to a blind connector to connect the computing tiles 1002 of two computing trays 112 to each other. For example, the connector 1008 may be connected to a connection interface such as Figure 5B The first connection interface 502a and / or the second connection interface 502b enable the computing tile 1002 in the first computing cabinet 100 and the second computing tile 1002 in the second computing cabinet 100 to be connected.

[0085] The compute tray 112 may include compute cooling connectors 1006. Some of the compute cooling connectors 1006 may receive coolant and provide coolant to the compute tray 112 to cool the compute tray 112 components, such as the compute tiles 1002. Other compute cooling connectors 1006 may exhaust coolant from the computer tray 112. For example, the computer cooling connectors 1006 may be connected to Figure 1C For example, some of the computer cooling connectors 1006 may receive coolant from the coolant distribution system 124, while other computer cooling connectors 1006 discharge coolant into the coolant distribution system 124. The compute cooling connectors 1006 may be positioned at the rear of the compute tray 112 such that when the compute tray 112 is fully inserted into the computer cabinet 100, the cooling connectors 1006 are connected to the coolant distribution system 124.

[0086] The computing tray 112 may include a blind computing connector 1010 configured to connect the computing tray 112 to a power source. For example, the blind computing connector 1010 may be inserted into Figure 1C 126 and receives power from the power tray 114. The blind compute connector 1010 can be positioned at the rear of the compute tray 112. In this way, when the compute tray 112 is fully inserted into the computer cabinet 100, the compute tray 112 can be safely plugged into a valid power source, such as the power bus 126.

[0087] like Fig. 10B As shown, the compute tray 112 can include a capacitor plate 1012. The capacitor plate 1012 can allow the compute tray 112 to be safely hot-swapped from a powered-on computer cabinet 100. For example, the capacitor plate 1012 can allow the compute tray 112 to be removed from active power and / or inserted into active power without damaging the compute tray 112 components, such as the compute tile 1002.

[0088] Fig.11 The diagram shows a compute tray 112 positioned above a host tray 116 according to one embodiment. As described above, the host tray 116 can implement ingest processing for the compute tray 112 alone or in combination with an additional compute tray 112 in the compute plane 102. The host tray 116 can include peripheral component express interconnect (PCIe) connectivity to an interface processor. The host tray 116 can provide video decoder support. In some embodiments, the host tray 116 can operate in an x86 Linux environment. The vertical stacking of the compute tray 112 and the host tray 116 can effectively utilize space, thereby enabling dense integration with a minimum connection length between the compute trays of adjacent cabinets.

[0089] In some applications, the modular design of the computer cabinet can allow a first section of the computer cabinet to continue to operate while the host trays 116 of the second section of the computer cabinet fail, are under repair, or are otherwise offline. Including computer trays 112 and host trays 116 that are paired with each other in the modular computer cabinet design can achieve such features.

[0090] In some embodiments, multiple host trays 116 can perform ingestion processing for the computing plane 102. In this way, a host tray 116 can fail and / or be removed from the computing system, and the computing plane 102 can continue to operate. In addition, the computing plane 102 can be partitioned into multiple operations, with one or more host trays 116 performing ingestion processing for each partitioned operation.

[0091] Unless the context clearly requires otherwise, throughout the specification and claims, the words "comprise", "comprising", "include", "including", etc. should be interpreted as inclusive, as opposed to exclusive or exhaustive; that is, in the sense of "including, but not limited to". The word "coupled" as generally used herein refers to two or more elements that can be directly connected or connected through one or more intermediate elements. Similarly, the word "connected" as generally used herein refers to two or more elements that can be directly connected or connected through one or more intermediate elements. In addition, when used in this application, the words "herein", "above", "below" and words of similar meaning shall refer to the entirety of this application and not to any particular part of this application. Where the context permits, words used in the above detailed description in the singular or plural may also include the plural or singular, respectively. The word "or" when referring to a list of two or more items covers all of the following interpretations of the word: any item in the list, all items in the list, and any combination of items in the list.

[0092] In addition, conditional language used herein, such as "can", "could", "might", "may", "eg", "for example", "such as", etc., unless otherwise specifically stated or otherwise understood in the context of use, is generally intended to convey that some embodiments include, while other embodiments do not include, certain features, elements, and / or states. Therefore, such conditional language is generally not intended to imply that one or more embodiments require features, elements, and / or states in any way.

[0093] The foregoing has been described with reference to specific embodiments. However, the above illustrative discussion is not intended to be exhaustive or to limit the invention to the precise form described. In view of the above teachings, many modifications and variations are possible. Therefore, others skilled in the art will be able to best utilize these techniques and various embodiments with various modifications to suit a variety of uses.

[0094] Although the present disclosure and examples have been described with reference to the accompanying drawings, various changes and modifications will become apparent to those skilled in the art. Such changes and modifications should be understood to be included within the scope of the present disclosure.

Claims

1. A computer cabinet, comprising: a first section including a first power plane, a first compute plane configured to receive power from the first power plane and perform computations, and a first host plane in communication with the first compute plane; a second segment, the second segment comprising a second power plane, a second compute plane configured to receive power from the second power plane and perform computations, and a second host plane in communication with the second compute plane; wherein the second segment is capable of operating independently of the first segment; as well as A cabinet frame, wherein the first section and the second section are positioned within the cabinet frame. 2 . The computer cabinet of claim 1 , wherein the first section and the second section are stacked vertically.

3. The computer cabinet of claim 2, wherein the first compute plane, the first host plane, the second power plane, and the second host plane are positioned between the first power plane and the second power plane.

4. The computer cabinet of claim 1, wherein at least a portion of the first power plane is hot-swappable when the first compute plane is operating.

5. The computer cabinet of claim 1, wherein the first computing plane is hot-swappable while the second computing plane is operating.

6. The computer cabinet of claim 1, wherein the first host plane is hot-swappable when the second compute plane is operating.

7. The computer cabinet of claim 1, further comprising one or more redundant connections between the first power plane and the first compute plane.

8. The computer cabinet of claim 1 , further comprising an interface comprising a connection to an external power source and a coolant inlet, wherein the connection to the external power source is electrically connected to the first power plane and the second power plane, and wherein the coolant inlet is in fluid communication with the first computing plane and the second computing plane.

9. The computer cabinet of claim 1, wherein the first computing plane comprises a computing tray and a plurality of computing tiles positioned on the computing tray, and each of the plurality of computing tiles comprises a plurality of die and a cooling solution integrated with the plurality of die.

10. The computer cabinet of claim 1, wherein the first power plane comprises a plurality of power trays configured to convert external power to power for the first computing plane.

11. The computer cabinet of claim 10, wherein a single one of the plurality of power trays is hot-swappable while other ones of the plurality of power trays are operating.

12. The computer cabinet of claim 1, further comprising a connector extending from the first computing plane and configured to connect to a computing plane of an adjacent cabinet.

13. The computer cabinet of claim 1 , further comprising a blind cooling connector on a side of the cabinet frame and a blind power connector on the side of the cabinet frame, wherein the first computing plane is connected to the blind cooling connector and the blind power connector when inserted into the computer cabinet.

14. A computing system, the computing system comprising: a first computing cabinet comprising a first power plane, a first computing plane configured to receive power from the first power plane and to perform computations, and a first host plane in communication with the first computing plane; and a second computing cabinet comprising a second power plane, a second computing plane configured to receive power from the second power plane and to perform computations, and a second host plane in communication with the second computing plane; Wherein the first computing plane is connected to the second computing plane via a connector, the connector extending through a side of the first computer cabinet and a side of the second computer cabinet, and wherein a position of the first computing plane is aligned with a position of the second computing plane.

15. The computing system of claim 14, wherein the connector connecting the first computing plane and the second computing plane is a blind connector.

16. The computing system of claim 14, wherein: The first computer cabinet also includes a third computing plane; The second computing cabinet further includes a fourth computing plane; and wherein the third computing plane is connected to the fourth computing plane via a second connector extending through the side of the first computer cabinet and the side of the second computer cabinet, and wherein a position of the third computing plane is aligned with a position of the fourth computing plane.

17. The computing system of claim 16, wherein: The first computing cabinet further includes a third power plane configured to provide power to the third computing plane, and a third host plane in communication with the third computing plane; and The second computing cabinet also includes a fourth power plane configured to provide power to the fourth computing plane, and a fourth host plane in communication with the fourth computing plane.

18. The computing system of claim 14, wherein the first power plane comprises a plurality of power trays, and a single power tray of the plurality of power trays is hot-swappable.

19. The computing system of claim 14, wherein the first computing cabinet further comprises one or more redundant connections between the first power plane and the first computing plane.

20. The computing system of claim 14, wherein at least one of the first power plane and the first host plane is hot-swappable.

21. The computing system of claim 14, wherein the first computer cabinet and the second computer cabinet are each capable of independent operation.

22. The computing system of claim 14, further comprising: a third computing cabinet comprising a third power plane, a third computing plane configured to receive power from the third power plane and to perform computations, and a third host plane in communication with the third computing plane; wherein the first computing plane is connected to the third computing plane via a third connector, the third connector extending through a second side of the first computing cabinet and a side of the third computing cabinet, and The first computing plane, the second computing plane and the third computing plane are connected.

23. The computing system of claim 14, further comprising a third computing cabinet connected to the second computing cabinet only via a connector extending from a second side of the second computing cabinet, and wherein the third computing cabinet comprises a third connector extending from a second side of the third computing cabinet for connection with a fourth computing cabinet.

Images