WO2018086101A1

WO2018086101A1 - Method for determining hotspot address, and device using same

Info

Publication number: WO2018086101A1
Application number: PCT/CN2016/105656
Authority: WO
Inventors: 陈溪; 陈挚睿; 李泉
Original assignee: 华为技术有限公司
Priority date: 2016-11-14
Filing date: 2016-11-14
Publication date: 2018-05-17
Also published as: CN109923846A; CN109923846B

Abstract

Provided are a method for determining a hotspot address and a device using same. The method is applied to a multi-core processor system and comprises: in a target time period, acquiring the number of times an ith segment of address is accessed in an access matching condition according to the access matching condition of the ith segment of address, wherein the access matching condition of the ith segment of address comprises an address field of the ith segment of address, the ith segment of address is one segment of address in N segments of addresses, wherein 1 ≤ i ≤ N, and N is a positive integer; and determining a hotspot address in the N segments of addresses according to the number of times each segment of address, in the N segments of addresses, is accessed in the corresponding access matching condition. Therefore, the method capable of determining an access hotspot address provided in the embodiments of the present invention can accurately and efficiently detect an access condition of multi-core shared variables and provide a basis for system performance optimization.

Description

Method and device for determining hotspot address

Technical field

Embodiments of the present invention relate to the field of information processing, and, more particularly, to a method of determining a hotspot address and an apparatus therefor.

Background technique

In the scenario where multi-core processors coexist, there may be cases where shared variables are shared among multi-core processors, and such shared storage variables include true sharing and pseudo-sharing, and pseudo-share refers to the actual shared variable. The above is a processor's own storage variable, and does not need to be shared with other processors, but because the variables are stored in the same cache block (English: CACHE LINE), the operation of these shared variables will be equivalent to shared. access.

The shared address field where the shared storage variable is located refers to the field with a relatively high number of accesses in a period of time, that is, the hotspot address. Only after determining the hotspot address can the system be better optimized.

One of the methods for determining a hotspot address in the prior art is: a method of checking by a manual code, which is a method for checking the hotspot address by the whole artificial party, which is inefficient, incomplete, and inaccurate; another way of determining the hotspot address is: The instruction tracking record of the chip simulation environment, while the simulation environment runs slower and less efficient, and the statistics on access to shared variables can only be inferred by the statistics of other modules, which is inaccurate.

Therefore, how to efficiently determine the access hotspot address is an urgent problem to be solved.

Summary of the invention

Embodiments of the present invention provide a method for determining a hotspot address, which can efficiently determine an access address.

The first method provides a method for determining a hotspot address, and the method is applied to a multi-core processor system, including: obtaining access in the i-th segment address according to an access matching condition of an i-th segment address in a target time period The number of accesses in the matching condition, wherein the access matching condition of the i-th segment address includes an address field of the i-th segment address, and the i-th segment address is an segment of the N-segment address, 1≤i≤N And N is a positive integer; determining a hotspot address in the N segment address according to the number of accesses of each of the N segments of addresses under the corresponding access matching condition.

That is to say, by recording the number of accesses of each of the N segments of addresses, it is possible to compare and determine the hotspot addresses in the N segments of addresses.

With reference to the first aspect, in a first possible implementation manner of the first aspect, the determining, according to the number of accesses of each of the N segments of addresses in a corresponding access matching condition, determining the N segments of addresses The hotspot address includes: determining, as a hotspot address, an address field in which each of the N segments of addresses has the most access times under the corresponding access matching condition.

With reference to the first aspect and the foregoing implementation manner, in a second possible implementation manner of the first aspect, the access matching condition of the i th segment address further includes at least one of: accessing the i th segment address Source processor identifier and destination processor identifier; access type of access to the i-th segment address.

With reference to the first aspect and the foregoing implementation manner, in a third possible implementation manner of the first aspect, the access matching condition of the i-th segment address includes a source processor identifier and a target processor of the i-th segment address And the method further includes: determining, according to the number of accesses of each of the N segments of addresses in the corresponding access matching condition, the source processor and the target processor with the most access times of the i th segment address.

With reference to the first aspect and the foregoing implementation manner, in a fourth possible implementation manner of the first aspect, the access matching condition of the i-th segment address includes an access type of the i-th segment address, and the method further includes And determining, according to the number of accesses of each of the N segments of addresses in the corresponding access matching condition, the access type with the most access times of the i th segment address.

It should be understood that determining the type of access with the most access times or determining the source and target processors with the most access times can facilitate subsequent system optimization.

Therefore, the embodiment of the present invention provides a method for determining an access hotspot address, which can detect the access situation of the multi-core shared variable accurately, efficiently, and at low cost, and provide a basis for system performance optimization.

In a second aspect, a network device is provided for performing the method of any of the above first aspect or any of the possible implementations of the first aspect. In particular, the apparatus comprises means for performing the method of any of the above-described first aspect or any of the possible implementations of the first aspect.

In a third aspect, a network device is provided, the network device comprising: a transceiver, a memory, a processor, and a bus system. Wherein the transceiver, the memory and the processor are connected by the bus system, the memory is for storing instructions for executing instructions stored by the memory to control the transceiver to receive and/or transmit signals, and The processor executes the instructions stored by the memory The execution causes the processor to perform the method of the first aspect or any possible implementation of the first aspect.

In a fourth aspect, a computer readable medium is provided for storing a computer program comprising instructions for performing the method of the first aspect or any of the possible implementations of the first aspect.

DRAWINGS

1 is a schematic flow chart of a method in accordance with an embodiment of the present invention.

2 is a schematic diagram of a method in accordance with another embodiment of the present invention.

FIG. 3 is a schematic structural diagram of a network device 300 according to an embodiment of the present invention.

Figure 4 is a schematic block diagram of an apparatus 400 in accordance with one embodiment of the present invention.

detailed description

1 is a schematic flowchart of a method according to an embodiment of the present invention. The method is applied to a multi-core processor system. The execution body of the method may be a physical machine. The method may be applied to a system on chip (English System on Chip, abbreviated: In SOC), as shown in FIG. 1, the method includes:

In step 110, the number of accesses under the access matching condition of the i-th segment address is obtained according to the access matching condition of the i-th segment address, wherein the access matching condition of the i-th segment address includes the i-th The address field of the segment address, the address of the i-th segment is an address in the N-segment address, 1 ≤ i ≤ N, and N is a positive integer.

Step 120: Determine a hotspot address in the N segment address according to the number of accesses of each segment of the N segment address under the corresponding access matching condition.

Specifically, the target time period may be a preset period of time, for example, may be 100 milliseconds, 1 second, etc., which is not limited in the embodiment of the present invention.

Wherein, in step 110, the N-segment address may be an N-segment address divided into a target address to be inspected, for example, the target address is [a1, a12], then the N-segment address may be the following three segments [a1, a3] ], [a4, a9], [a10, a12], the i-th address can be one of the three-segment addresses.

In step 110, the access matching condition of the i-th segment address includes at least the specific address field of the i-th segment address; further, the access matching condition may further include accessing the identifier of the source processor and the identifier of the destination processor to be accessed. To determine which processor the source processing to initiate the access request is, and which processor the destination processor to access is; further, the above access matching condition It can also include specific access types for access, such as read operation access types, write operation access types, and the like.

Specifically, in step 110, for determining the number of accesses under the access matching condition of the i-th segment address, the monitoring logic of the shared access integrated in the chip may be recorded for the shared access initiated by each access component, for example, For systems that support caching (English: CACHE) consistency, there will be a module similar to the snoop filter (English: SNOOP-FILTER), which can monitor all shared accesses and save the accessed records to the backup state. Directory, which determines the destination storage address for a particular visit.

In step 120, in several access operations of the operating system, the hotspot address in the N-segment address can be determined according to the number of accesses of each pair of addresses under the corresponding access matching condition.

For example, the number of accesses satisfying the access matching condition of the first segment address is 3, the number of accesses satisfying the access matching condition of the second segment address is 5, and the number of accesses satisfying the access matching condition of the third segment address is 1 time. Then the second segment of the address may be a hot address.

It should be understood that the hotspot address may be one address with the most number of accesses, or may be the address with the most number of accesses, which is not limited in the embodiment of the present invention.

Therefore, the embodiment of the present invention provides a method for determining an access hotspot address, which can accurately and efficiently detect the access situation of the multi-core shared variable, and provide a basis for system performance optimization.

Further, considering the saving of logical resources, the method only records the number of shared accesses of a certain address. Therefore, the system can dynamically configure the monitoring address, obtain the number of shared accesses of each address segment, and analyze the data to the precise shared access hotspot.

Optionally, as an embodiment of the present invention, determining a hotspot address in the N segment address according to the number of accesses of each segment of the N segment address under the corresponding access matching condition, including: The address field with the most access times under the corresponding access matching condition is determined as the hotspot address.

That is, in this alternative embodiment, the hotspot address is the address field that has the most access times to satisfy the access condition.

Optionally, as an embodiment of the present invention, the access matching condition of the i-th segment address further includes at least one of the following: accessing the source processor identifier and the destination processor identifier of the i-th segment address; accessing the i-th segment address Access type.

For example, the access matching condition of the first segment address [a1:a3] is: address [a1: a3], access type [read], source processor identifier, and destination processor identifier [processor 1, processor 2]; The number of visits is 2.

For example, the access matching condition of the first segment address [a1:a3] is: address [a1:a3], access type [read], The source processor ID and destination processor ID are [Processor 1, Processor 3]; the number of accesses is 5.

For example, the access matching condition of the first segment address [a1:a3] is: address [a1: a3], access type [write], source processor identifier, and destination processor identifier [processor 1, processor 3]; The number of visits is 3.

For example, the access matching condition of the second segment address [a4:a9] is: address [a4:a9], access type [read], source processor identifier, and destination processor identifier [processor 1, processor 3]; The number of visits is 6.

For example, the access matching condition of the third segment address [a10:a12] is: [a10:a12], access type [write], source processor identifier, and destination processor identifier [processor 2, processor 3]; access The number of times is 1.

Optionally, as an embodiment of the present invention, the access matching condition of the i-th segment address includes a source processor identifier and a target processor identifier of the i-th segment address, and the method further includes: according to the N-segment address The number of accesses of each segment of the address under the corresponding access matching condition determines the source processor and the target processor with the most access times of the i-th segment address.

Optionally, the access matching condition of the i-th segment address includes an access type of the i-th segment address, and the method further includes: accessing each segment of the N-segment address under a corresponding access matching condition. The number of times to determine the access type with the most access times of the i-th segment address.

2 is a schematic diagram of a method in accordance with another embodiment of the present invention. As shown in FIG. 2, the figure includes a processor core core1, a core core2, ..., a core core N, and each core accesses or accesses a memory. Memory is monitored by SNOOP-FILTER, and the component can monitor all the shares. Access, save the backup status directory of each CACHE LINE, determine the destination address of a visit, that is, the purpose of the access is to which CACHE or subordinate storage.

Specifically, the system configures the SNOOP FILTER parameter, specifically including access matching conditions [CORE-source, CORE-dest]+[A1:A2]+[access type <read-write>] and time period T1, and SNOOP FILTER records according to configuration parameters. The number of times that the matching condition is met within T1 in a time period.

Further, the following data can be finalized by modifying the configuration parameters:

Matching condition 1: shared access times N _a ;

Matching condition 2: number of shared accesses N _b ;

Matching condition x: number of shared accesses N _x ; where N _a , N _b , and N _x are all non-negative integers.

Further, by performing statistical analysis on the acquired data, it can be obtained:

The most visited address segment [Ax-Ay] is the hotspot address of the shared data;

[source-destination] with the most visits;

The most visited type of access.

After obtaining the hotspot address of the shared data, find the function that shares the data access by disassembling the analysis software code image, that is, finding the point to be optimized, optimizing, or, further, configuring the part by integrating the TRACE function inside the CPU. When the [Ax-Ay] occurs, the current PC pointer and stack information are recorded. The TRACE information is read by the software, and the function information related to the hotspot data can be obtained to find an optimization point.

1 and 2 illustrate in detail the flow of the method of the embodiment of the present invention, and the apparatus and apparatus of the embodiment of the present invention are described below with reference to FIGS. 3 and 4.

FIG. 3 is a schematic structural diagram of a network device 300 according to an embodiment of the present invention. It should be understood that the network device 300 is capable of performing the various steps in the method of FIG. 1 or FIG. 2, and to avoid repetition, it will not be described in detail herein. Network device 200 includes:

An obtaining unit 310, configured to acquire, according to an access matching condition of the i-th segment address, an access number under an access matching condition of the i-th segment address, where the i-th segment The access matching condition of the address includes an address field of the i-th segment address, and the i-th segment address is an segment of the N-segment address, 1≤i≤N, and N is a positive integer.

The determining unit 320 is configured to determine a hotspot address in the N-segment address according to the number of accesses of each of the N-segment addresses under the corresponding access matching condition.

Optionally, as an embodiment of the present invention, the determining unit 320 is specifically configured to: determine, as a hotspot address, an address field in which each of the N segments of addresses has the most access times under the corresponding access matching condition.

Optionally, as an embodiment of the present invention, the access matching condition of the i-th segment address further includes at least one of the following: a source processor identifier and a destination processor identifier that access the i-th segment address Knowledge; access to the access type of the i-th segment address.

Optionally, as an embodiment of the present invention, the access matching condition of the i-th segment address includes a source processor identifier and a target processor identifier of the i-th segment address, and the determining unit 320 is further configured to: The number of accesses of each of the N segments of addresses in the corresponding access matching condition is determined, and the source processor and the target processor with the most access times of the i th segment address are determined.

Optionally, as an embodiment of the present invention, the access matching condition of the ith segment address includes an access type of the ith segment address, and the determining unit 320 is further configured to: according to each segment of the N segment address The number of accesses of the address under the corresponding access matching condition determines the access type with the most access times of the i-th segment address.

Figure 4 is a schematic block diagram of an apparatus 400 in accordance with one embodiment of the present invention. FIG. 4 shows an apparatus 400 provided by an embodiment of the present invention. It should be understood that apparatus 400 is capable of performing the various steps of the method of FIG. 1 or FIG. 2, and to avoid repetition, details are not described herein. Apparatus 400 includes:

a memory 410, configured to store a program;

The transceiver 420 is configured to communicate with other devices;

a processor 430, configured to execute a program in the memory 410, the processor 430 is coupled to the memory 410 and the transceiver 420, respectively, for executing the instructions stored by the memory 410 to execute the instruction Perform the following steps:

The processor 430 is configured to obtain, by the transceiver 420, the number of accesses under the access matching condition of the i-th segment address according to an access matching condition of the i-th segment address in the target time period, where the i-th segment The access matching condition of the address includes an address field of the i-th segment address, the i-th segment address is an segment of the N-segment address, 1≤i≤N, and N is a positive integer; according to each of the N-segment addresses The number of accesses of the segment address under the corresponding access matching condition determines the hotspot address in the N segment address.

Those of ordinary skill in the art will appreciate that the devices or elements and algorithm steps of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the solution. A person skilled in the art can use different methods to implement the described functions for each particular application, but such implementation should not be considered to be beyond the scope of the embodiments of the invention.

A person skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the system, the device and the device or the unit described above can refer to the corresponding process in the foregoing method embodiment, and details are not described herein again.

In the several embodiments provided by the present application, it should be understood that the disclosed systems, devices, and methods may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of the device is only a logical function division, and the actual implementation may have another division manner, for example, multiple devices or components may be combined or Can be integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or device, and may be in an electrical, mechanical or other form.

In addition, each functional device in various embodiments of the present invention may be integrated in one processing device, or each device may exist physically separately, or two or more devices may be integrated in one device.

The functions, if implemented in the form of software functional devices and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention, which is essential or contributes to the prior art, or a part of the technical solution, may be embodied in the form of a software product, which is stored in a storage medium, including The instructions are used to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

The above is only a specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think of changes or substitutions within the technical scope of the present invention. It should be covered by the scope of the present invention. Therefore, the scope of the invention should be determined by the scope of the claims.

Claims

A method for determining a hotspot address, the method being applied to a multi-core processor system, comprising:

Obtaining, in the target time period, the number of accesses under the access matching condition of the i-th segment address according to the access matching condition of the i-th segment address, wherein the access matching condition of the i-th segment address includes the i-th segment Address field of the address, the address of the i-th segment is an address in the N-segment address, 1≤i≤N, and N is a positive integer;

Determining a hotspot address in the N-segment address according to the number of accesses of each of the N-segment addresses under the corresponding access matching condition.
The method according to claim 1, wherein the determining the hotspot address in the N segment address according to the number of accesses of each of the N segments of addresses in the corresponding access matching condition comprises:

The address field with the most access times of each of the N segments of addresses in the corresponding access matching condition is determined as a hotspot address.
The method according to claim 1 or 2, wherein the access matching condition of the i-th segment address further comprises at least one of the following:

Accessing the source processor identifier and the destination processor identifier of the i-th segment address;

Access the access type of the i-th segment address.
The method according to claim 3, wherein the access matching condition of the i-th segment address comprises a source processor identifier and a target processor identifier of the i-th segment address, the method further comprising:

Determining, according to the number of accesses of each of the N segments of addresses under the corresponding access matching condition, the source processor and the target processor having the most access times of the i th segment address.
The method according to claim 3 or 4, wherein the access matching condition of the i-th segment address comprises an access type of the i-th segment address, the method further comprising:

And determining, according to the number of accesses of each of the N segments of addresses in the corresponding access matching condition, the access type with the most access times of the i th segment address.
A network device, comprising:

An obtaining unit, configured to acquire, according to an access matching condition of the i-th segment address, an access number under an access matching condition of the i-th segment address, where the i-th segment address is The access matching condition includes an address field of the i-th segment address, the i-th segment The address is an address in the N segment address, 1≤i≤N, and N is a positive integer;

a determining unit, configured to determine a hotspot address in the N segment address according to the number of accesses of each of the N segments of addresses under a corresponding access matching condition.
The network device according to claim 6, wherein the determining unit is specifically configured to:

The address field with the most access times of each of the N segments of addresses in the corresponding access matching condition is determined as a hotspot address.
The network device according to claim 6 or 7, wherein the access matching condition of the i-th segment address further comprises at least one of the following:

Accessing the source processor identifier and the destination processor identifier of the i-th segment address;

Access the access type of the i-th segment address.
The network device according to claim 7 or 8, wherein the access matching condition of the i-th segment address comprises a source processor identifier and a target processor identifier of the i-th segment address, and the determining unit further uses to:

Determining, according to the number of accesses of each of the N segments of addresses under the corresponding access matching condition, the source processor and the target processor having the most access times of the i th segment address.
The network device according to claim 8 or 9, wherein the access matching condition of the i-th segment address includes an access type of the i-th segment address, and the determining unit is further configured to:

And determining, according to the number of accesses of each of the N segments of addresses in the corresponding access matching condition, the access type with the most access times of the i th segment address.