CN112783309A

CN112783309A - Server

Info

Publication number: CN112783309A
Application number: CN202110056404.2A
Authority: CN
Inventors: 王健; 关焜予
Original assignee: Suzhou Inspur Intelligent Technology Co Ltd
Current assignee: Suzhou Inspur Intelligent Technology Co Ltd
Priority date: 2021-01-15
Filing date: 2021-01-15
Publication date: 2021-05-11
Anticipated expiration: 2041-01-15
Also published as: CN112783309B

Abstract

The invention discloses a server, comprising: the power supply switching board is connected with the power supply switching board; the power supply comprises a power supply adapter plate and a GPU plate, wherein the GPU plate comprises a first power supply copper bar and a plurality of GPU plate adapter plates, each GPU plate adapter plate is connected with one power supply adapter plate, so that a power supply supplies power to the GPU plate through the power supply adapter plates and the GPU plate adapter plates; the mainboard comprises a second power supply copper bar, and the second power supply copper bar is connected with the first power supply copper bar so that the GPU board supplies power to the mainboard through the first power supply copper bar and the second power supply copper bar. The scheme provided by the invention changes the system power supply current path into a system power supply current path which is directly input into the GPU board by the PSU to supply power to the GPU and then is supplied with power by the GPU board- > power supply copper bar- > mainboard. Adopt PSU directly to supply power for the GPU board like this, the mainboard is provided with the power to the rethread power supply copper bar, can make the energy of power supply copper bar transmission reduce a lot, even if GPU power increases, also can not make the consumption increase of copper bar, consequently need not to increase the size and the quantity of copper bar.

Description

Server

Technical Field

The invention relates to the field of power supply, in particular to a server.

Background

In the prior art, the power supply is directly transmitted into the mainboard through the PSU, and then enters the GPU through the power supply copper bar. When the GPU type is upgraded, the power consumption of the GPU increases compared to the previous generation, and the current architecture can only support up to 8 GPUs of 300W.

After the GPU is upgraded, its power consumption may increase to 450W. If the current power supply architecture is continuously used, the required power supply copper bar needs to be further enlarged, and occupies a larger area. At the same time, the current density is also very high, with the risk of burning the board. Or adopt 4 power supply copper bars with the interconnection to increase to 6, but it is bigger to occupy the space of mainboard and GPU board like this, and the mainboard space can be more nervous, and the mainboard can not necessarily be finished the design under current quick-witted case size. Simultaneously, the power supply post quantity increases, and can be higher to structural requirement, needs the error littleer just can guarantee that the connection of power supply copper bar does not have the problem, and this problem can lead to the yield of production to descend.

Disclosure of Invention

In view of the above, in order to overcome at least one aspect of the above problems, an embodiment of the present invention provides a server, including:

the power supply comprises a plurality of power supplies, a plurality of control circuits and a power supply adapter plate, wherein each power supply is connected with one power supply adapter plate;

the power supply comprises a power supply adapter plate, a GPU plate and a power supply, wherein the GPU plate comprises a first power supply copper bar and a plurality of GPU plate adapter plates, each GPU plate adapter plate is connected with one power supply adapter plate, so that the power supply supplies power to the GPU plate through the power supply adapter plate and the GPU plate adapter plate;

the mainboard comprises a second power supply copper bar, wherein the second power supply copper bar is connected with the first power supply copper bar, so that the GPU board passes through the first power supply copper bar and the second power supply copper bar to supply power to the mainboard.

In some embodiments, a chassis is also included;

the plurality of power supplies and the mainboard are arranged at a first height in the case, and the GPU board is arranged at a second height in the case.

In some embodiments, the motherboard further comprises a heat sink.

In some embodiments, the GPU board is disposed above the plurality of power supplies and is not vertically coincident with the heat sink.

In some embodiments, the main plate is in a "convex" configuration, the main plate including a protruding portion and a remaining portion.

In some embodiments, the plurality of power sources are respectively located on both sides of the protruding portion of the main board;

the power adapter board is located between the power supply and the rest of the main board.

In some embodiments, the second copper bar of power supply is disposed on a protruding portion of the main board.

In some embodiments, the power adapter board includes a first slot and a second slot;

the first slot is connected with the power supply, and the second slot is connected with the GPU board adapter plate.

In some embodiments, the first slot and the second slot are located at two ends of the power adapter board, and the height of the second slot is greater than that of the first slot.

In some embodiments, several of the plurality of power supplies function as redundant power supplies.

The invention has one of the following beneficial technical effects: the scheme provided by the invention enables a system power supply current path to enter a mainboard for power supply from an original PSU (power supply), and then the power supply current path is supplied by the mainboard- > power supply copper bar- > GPU board, so that the power supply is optimized to be supplied to a GPU by directly entering the GPU board from the PSU, and then the power supply current path is supplied by the GPU board- > power supply copper bar- > mainboard. Adopt PSU directly to supply power for the GPU board like this, the mainboard is provided with the power to the rethread power supply copper bar, can make the energy of power supply copper bar transmission reduce a lot, even if GPU power increases, also can not make the consumption increase of copper bar, consequently need not to increase the size and the quantity of copper bar.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other embodiments can be obtained by using the drawings without creative efforts.

FIG. 1 is a side view of a server provided by an embodiment of the present invention;

fig. 2 is a top view of a motherboard of a server according to an embodiment of the present invention;

FIG. 3 is a top view of a motherboard of a prior art server;

FIG. 4 is a top view of a GPU board of a server provided by an embodiment of the present invention;

fig. 5 is a side view of a power adapter plate provided in an embodiment of the invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the following embodiments of the present invention are described in further detail with reference to the accompanying drawings.

It should be noted that all expressions using "first" and "second" in the embodiments of the present invention are used for distinguishing two entities with the same name but different names or different parameters, and it should be noted that "first" and "second" are merely for convenience of description and should not be construed as limitations of the embodiments of the present invention, and they are not described in any more detail in the following embodiments.

According to an aspect of the present invention, an embodiment of the present invention provides a server, as shown in fig. 1, which may include:

The scheme provided by the invention enables a system power supply current path to enter a mainboard for power supply from an original PSU (power supply), and then the power supply current path is supplied by the mainboard- > power supply copper bar- > GPU board, so that the power supply is optimized to be supplied to a GPU by directly entering the GPU board from the PSU, and then the power supply current path is supplied by the GPU board- > power supply copper bar- > mainboard. Adopt PSU directly to supply power for the GPU board like this, the mainboard is provided with the power to the rethread power supply copper bar, can make the energy of power supply copper bar transmission reduce a lot, even if GPU power increases, also can not make the consumption increase of copper bar, consequently need not to increase the size and the quantity of copper bar. And then realize under the condition that system power increases, need not change the structure of power supply copper bar, reduce the problem of burning board risk, reinforcing system power supply reliability.

In some embodiments, a chassis is also included;

Specifically, as shown in fig. 1, the plurality of power supplies and the motherboard are disposed at the bottom of the chassis (i.e., in the space of the first height), and the height of both is up to 1U. The GPU board and the GPU thereon are located at the upper part of the chassis (i.e. in the space of the second height), and the height occupied by the GPU board and the GPU board may be 3U, so that the height of the entire server may be 4U.

In some embodiments, the motherboard further comprises a heat sink.

In some embodiments, the GPU board is disposed above the plurality of power sources and is vertically misaligned with, i.e., does not obscure, the heat sink.

Specifically, as shown in fig. 1, a heat sink may be disposed above the motherboard, and the heat sink may be a CPU heat sink, and in order to avoid affecting the heat dissipation effect, the GPU board may be disposed above the power supply and may not shield the heat sink, so as to avoid a conflict with the CPU heat sink of the motherboard.

In some embodiments, the main board is in a convex structure.

Specifically, as shown in fig. 2, the main board has a zigzag structure, the number of power supplies may be 4, and the power supplies are evenly distributed on two sides of the protruding portion of the main board, that is, two sides of the main board are respectively provided with 2 power supplies. And because the PSU is at the height of 1U in the lowest layer and the GPU board is at the height of 2U, a power adapter board can be used to input power from the PSU to the GPU board, the power adapter board shown in fig. 2 can be located between the power supply and the rest of the motherboard, and similarly, the power adapter boards are equally distributed on both sides of the protruding portion of the motherboard.

As shown in fig. 3, the original power supply scheme is that the PSU directly supplies power to the motherboard, so a PSU connector is required to be arranged at the end of the motherboard. The PSU directly supplies power to the GPU in the optimized scheme, so that the PSU connector used for being connected with the PSU on the mainboard can be removed, and the position corresponding to the PSU connector of the mainboard retracts inwards to accommodate the power adapter plate. And the power supply copper bar is still reserved, and the GPU board supplies power to the mainboard.

It should be noted that, because the power supply of the motherboard is supplied to the motherboard by the GPU through the power supply copper bar, the motherboard does not need to collect the power supplies of the PSUs on the two sides, so that the space of a part of board cards can be saved, and no other problem is caused by retracting the space near the PSUs.

Specifically, as shown in fig. 2, in order to supply power to the motherboard through the GPU board, a power supply copper bar may be reserved, that is, the GPU board supplies power to the second power supply copper bar on the motherboard through the first copper bar thereon, and then supplies power to the motherboard.

In some embodiments, a protruding portion of the left side of the motherboard may also provide the PCH and BMC.

Specifically, as shown in fig. 5, the power adapter board may include a first slot and a second slot, the first slot may be connected to the PSU, and since the PSU is at the height of 1U of the lowermost layer and the GPU board is at the height of 2U, the height of the second slot is greater than that of the first slot.

Specifically, the power supply may employ 2+2 redundancy to supply power to the GPU board and the motherboard. And compared with the prior art, the size of the GPU board is increased, the GPU board adapter plate is added, the GPU board adapter plate is connected with the power supply adapter plate, current is collected, and the requirement of a 2+2 redundant power supply is met.

According to the scheme provided by the invention, the PSU is directly used for supplying power to the mainboard instead of directly used for supplying power to the GPU, so that the energy transmitted by the power supply copper bar can be greatly reduced, and even if the power of the GPU is increased, the power consumption of the copper bar can not be increased, and the size and the number of the copper bars do not need to be increased. For example, when the power consumption of the GPU board is estimated to be 8 GPUs of 450W, plus other chips on the GPU board, the power consumption is estimated to be 100W, and the common load power consumption is 3700W; the main loads of the main board end are CPU and memory, the evaluation is carried out according to 2 CPUs of 350W and 32 memory banks of 20W, PCH is 20W, 20W is shared by BMC and CPLD, and 1480W is calculated by adding reserved PCIE and USB equipment 100W, which is far less than the power consumption of a GPU board. If power supply of the hard disk and the fan is additionally provided, the power consumption is still much less than that of the GPU board according to the total 2800W calculated by 24 hard disks of 20W and 12 fans of 70W. Therefore, the power consumed on the power supply copper bar is smaller than that directly supplied to the mainboard through the PSU and then supplied to the GPU, and meanwhile, the power adapter plate does not have excessive power loss.

The foregoing is an exemplary embodiment of the present disclosure, but it should be noted that various changes and modifications could be made herein without departing from the scope of the present disclosure as defined by the appended claims. The functions, steps and/or actions of the method claims in accordance with the disclosed embodiments described herein need not be performed in any particular order. Furthermore, although elements of the disclosed embodiments of the invention may be described or claimed in the singular, the plural is contemplated unless limitation to the singular is explicitly stated.

It should be understood that, as used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly supports the exception. It should also be understood that "and/or" as used herein is meant to include any and all possible combinations of one or more of the associated listed items.

The numbers of the embodiments disclosed in the embodiments of the present invention are merely for description, and do not represent the merits of the embodiments.

Those of ordinary skill in the art will understand that: the discussion of any embodiment above is meant to be exemplary only, and is not intended to intimate that the scope of the disclosure, including the claims, of embodiments of the invention is limited to these examples; within the idea of an embodiment of the invention, also technical features in the above embodiment or in different embodiments may be combined and there are many other variations of the different aspects of the embodiments of the invention as described above, which are not provided in detail for the sake of brevity. Therefore, any omissions, modifications, substitutions, improvements, and the like that may be made without departing from the spirit and principles of the embodiments of the present invention are intended to be included within the scope of the embodiments of the present invention.

Claims

1. A server, comprising:

2. The server of claim 1, further comprising a chassis;

3. The server of claim 2, wherein the motherboard further comprises a heat sink.

4. The server of claim 3, wherein the GPU board is disposed above the plurality of power supplies and is vertically misaligned with the heat sink.

5. The server according to claim 1, wherein the motherboard has a dogbone structure, the motherboard including a protruding portion and a remaining portion.

6. The server according to claim 5, wherein the plurality of power supplies are respectively located on both sides of the protruding portion of the main board;

7. The server according to claim 5, wherein the second copper supply line is provided on a protruding portion of the main board.

8. The server of claim 1, wherein the power adapter board comprises a first slot and a second slot;

9. The server of claim 8, wherein the first slot and the second slot are located at opposite ends of the power adapter board, and wherein the second slot has a height greater than a height of the first slot.

10. The server of claim 1, wherein a number of the plurality of power supplies function as redundant power supplies.