Method and system for controlling data response by means of attributes of transaction identification code

文档序号:48901 发布日期:2021-09-28 浏览:39次 中文

阅读说明:本技术 借助于交易辨识码的属性来控制数据响应的方法以及系统 (Method and system for controlling data response by means of attributes of transaction identification code ) 是由 赖奇劭 于 2020-03-25 设计创作,主要内容包括:本申请提供一种借助于一交易辨识码的至少一属性来控制数据响应的方法以及系统。该方法包含:将该至少一属性连同该交易辨识码自一或多个主装置中的任一主装置一并传送至一从装置;以及依据该至少一属性来判断是否利用多个数据路径中的一特定数据路径将对应于该交易辨识码的数据自该从装置中的一存储装置响应至所述任一主装置。具体来说,该特定数据路径是该多个数据路径中具有最大传输效率的数据路径。(A method and system for controlling data response by means of at least one attribute of a transaction identifier is provided. The method comprises the following steps: transmitting the at least one attribute and the transaction identification code from any one of one or more master devices to a slave device; and determining whether to respond the data corresponding to the transaction identification code from a storage device in the slave device to any one of the master devices by using a specific data path in a plurality of data paths according to the at least one attribute. Specifically, the specific data path is a data path having the greatest transmission efficiency among the plurality of data paths.)

1. A method for controlling data response by at least one attribute of a transaction identifier, the method comprising:

transmitting the at least one attribute and the transaction identification code from any one of one or more master devices to a slave device; and

determining whether to utilize a specific data path of the plurality of data paths to respond data corresponding to the transaction identification code from a storage device of the slave device to any of the master devices according to the at least one attribute;

wherein the specific data path is a data path having the greatest transmission efficiency among the plurality of data paths.

2. The method of claim 1, wherein the at least one attribute indicates whether the transaction identifier is unique throughout a system that includes the one or more master devices and the slave device.

3. The method of claim 2, wherein the slave device includes a read reorder buffer for responding out-of-order data from the storage device to said any master device in an ordered manner; and the specific data path includes a bypass path for bypassing the read reorder buffer from forwarding the data corresponding to the transaction identifier from the memory device directly to the one of the masters.

4. The method of claim 3, further comprising:

directly transmitting the transaction identification code to the storage device to avoid storing the transaction identification code in the read reorder buffer in response to the storage device performing access control with a fixed burst length, the transaction identification code having a length less than or equal to the fixed burst length, and the at least one attribute indicating that the transaction identification code is unique in the entire system.

5. The method of claim 3, further comprising:

in response to the memory device performing access control with a fixed burst length and the transaction identifier being longer than the fixed burst length, truncating the transaction identifier to generate a plurality of burst truncation instructions having the fixed burst length, and storing the transaction identifier in the read reorder buffer;

obtaining the data corresponding to the transaction identification code from the storage device by using the plurality of burst truncation instructions; and

the transaction identifier stored in the read reorder buffer is responded to the any master device along with the data corresponding to the transaction identifier using the specific data path.

6. The method of claim 2, wherein any of the master device and the slave device are coupled to each other through a transmission interface circuit, the transmission interface circuit comprising the plurality of data paths, and the particular data path being dedicated to respond to data corresponding to a unique transaction identification code.

7. The method of claim 1, wherein the at least one attribute is transmitted using an extension field or a sideband channel of the transaction identifier.

8. A system for controlling data response by at least one attribute of a transaction identifier, comprising:

a slave device including a memory device for storing data;

one or more master devices coupled to the slave devices through a transmission interface circuit for obtaining data stored in the storage devices;

wherein:

any master device of the one or more master devices transmits the at least one attribute and the transaction identification code to the slave device; and

according to the at least one attribute, the slave device judges whether to utilize a specific data path in a plurality of data paths to respond the data corresponding to the transaction identification code from the storage device to any master device;

wherein the at least one attribute indicates whether the transaction identifier is unique throughout a system including the one or more master devices and the slave device, and the particular data path is a data path of the plurality of data paths having a maximum transmission efficiency.

9. The system of claim 8,

the slave device includes:

a read reorder buffer, coupled to the memory device, for responding out-of-order data from the memory device to any master device in an ordered manner from the slave device; and

the specific data path includes:

a bypass path for bypassing the read reorder buffer from forwarding the data corresponding to the transaction identifier from the memory device directly to any of the masters.

10. The system of claim 8, wherein the at least one attribute is transmitted using an extension field or a sideband channel of the transaction identifier.

Technical Field

The present invention relates to data response control, and more particularly, to a method and system for controlling data response by using at least one attribute (attribute) of a transaction identifier.

Background

Many transport interfaces in a system may be configured with read re-order buffers (READ RE-order buffers) to allow the system to reorder multiple concurrently issued instructions, which may be unrelated to each other, to achieve optimal response efficiency. However, although a particular instruction is mutually exclusive from other instructions (meaning that reordering can be performed between the instructions), the response data of the particular instruction may be blocked by the response data of the other instructions in some cases, thereby greatly increasing the delay time of data response.

Therefore, a novel data response control mechanism and related system are needed to allow the response data of a specific instruction that is not related to other instructions to be blocked by the response data of other instructions, so as to achieve the best data response efficiency.

Disclosure of Invention

An object of the present invention is to provide a method and system for controlling data response by means of at least one attribute (attribute) of a transaction identifier, so as to allow the response data of a specific instruction unrelated to other instructions not to be blocked by the response data of other instructions, thereby achieving the best data response efficiency.

At least one embodiment of the present invention provides a method for controlling data response by means of at least one attribute of a transaction identifier. The method comprises the following steps: transmitting the at least one attribute together with the transaction identifier from any one of one or more master devices (master devices) to a slave device (slave device); and determining whether to respond the data corresponding to the transaction identification code from a storage device in the slave device to any one of the master devices by using a specific data path in a plurality of data paths according to the at least one attribute. For example, the specific data path may be a data path having the greatest transmission efficiency among the plurality of data paths.

At least one embodiment of the present invention provides a system for controlling data response by means of at least one attribute of a transaction identifier. The system includes a slave device and one or more master devices, wherein the one or more master devices are coupled to the slave device through a transmission interface circuit. The slave device may include a memory device for storing data, and the one or more master devices may access the data stored in the memory device through the transmission interface circuit. Specifically, any master device of the one or more master devices may transmit the at least one attribute along with the transaction identifier to the slave device; and according to the at least one attribute, the slave device can judge whether to utilize a specific data path in a plurality of data paths to respond the data corresponding to the transaction identification code from the storage device to any master device. For example, the specific data path is a data path having the greatest transmission efficiency among the plurality of data paths.

The method and system provided by the embodiment of the invention can transmit the attribute of the transaction identifier when transmitting the transaction identifier so as to judge whether a bypass path is utilized to optimize the efficiency of data response. Compared with the related art, the embodiment of the invention does not greatly increase the additional cost. The present invention can therefore solve the problems of the related art without side effects or with less side effects.

Drawings

FIG. 1 is a diagram illustrating some implementation details of a read reorder buffer in a read operation, in accordance with one embodiment of the present invention.

FIG. 2 is a diagram illustrating a system for controlling data response by at least one attribute of a transaction identifier according to an embodiment of the invention.

Fig. 3 is a flowchart of a method for controlling data response by at least one attribute of a tag identification according to an embodiment of the invention.

Fig. 4 is some implementation details of a slave device in the system of fig. 2 in a situation where a tag identification is unique in the system, according to an embodiment of the present invention.

Fig. 5 is some implementation details of a slave device in the system of fig. 2 in a situation where a tag identification is not unique in the system, according to an embodiment of the present invention.

Fig. 6 is a flowchart of a method for controlling data response by at least one attribute of a tag identification according to an embodiment of the invention.

Fig. 7 is some implementation details of a slave device in the system of fig. 2 in a situation where a tag identification is unique in the system, according to an embodiment of the present invention.

Fig. 8 is some implementation details of a slave device in the system of fig. 2 in a situation where a tag identification is not unique in the system, according to an embodiment of the present invention.

[ notation ] to show

100 system

120_1,120_ N,120_ N master device

140 slave device

10,141 atomizer

20,142 read reorder buffer

30,143 storage device

142A,142B storage area

160 transmission interface circuit

162_1,162_ N host interface

164 slave interface

Master _ Tag _ ID (0), Master _ Tag _ ID (n), Master _ Tag _ ID (N), Tag identification code

Master_ID(0),Master_ID(n),Master_ID(N),Is_unique_ID(0),Is_unique_ID(n),

Is _ unique _ ID (N) attribute

302-324,402-424 step

200 tag ID mapping and read reorder buffer controller

220 control circuit

240,280 selection circuit

260 image circuit

Tag _ ID Tag identification code

Group _ ID Group ID

W _ pointer, write pointer

R _ pointer, reading pointer

cmd _ index, RRB _ index, index

Detailed Description

FIG. 1 shows some implementation details of a read re-order buffer (RRB) 20 in a read operation, which may be coupled between an atomizer (atomizer)10 and a memory device 30, according to an embodiment of the present invention. As shown in FIG. 1, the arrow labeled "read command request" may indicate the direction of a read command request in the read operation, and the arrow labeled "read data response" may indicate the direction of a read data response in the read operation. In the case where the memory device 30 receives a fixed burst length (burst length) for access control, the atomizer 10 reads a transaction (transaction) Identifier (ID), such as a tag (tag) identifier, truncates (chop) the transaction (transaction) to generate a plurality of burst truncation instructions having the fixed burst length, and gives each of the plurality of burst truncation instructions an index, such as rrb _ index, and stores the tag identifier with the index rrb _ index in the read reorder buffer. The storage device 30 may read the burst chop command with the index rrb _ index and respond to the valid data with the index rrb _ index to allow the read reorder buffer 20 to store the valid response with the index rrb _ index, wherein the storage device 30 may respond to the data in an out-of-order (out-of-order) manner to optimize the overall read performance (e.g., minimize the read latency and increase the usage of dynamic out-of-order access), and examples of the storage device 30 may include a Random Access Memory (RAM) such as Static RAM (SRAM) and Dynamic RAM (DRAM). In some embodiments, the read reorder buffer 20 may be implemented using a First In First Out (FIFO) buffer, and the atomizer 10 may fetch a tag, such as a flag, of the bottom entry of the read reorder buffer 20 and an index of this bottom entry (labeled "tag, index of the RRB bottom entry" in fig. 1), and fetch a valid response containing the tag identification code from the top of the read reorder buffer 20 (labeled "fetch valid response from the top of RRB (containing tag identification code)" in fig. 1), to allow the atomizer 10 to respond to the data with the tag identification code in order (in-order).

In some embodiments, the memory device 30 may be access controlled with different burst lengths, so the atomizer 10 may be replaced with a wrapper (wrapper) to transfer the complete instruction to the memory device 30 without intercepting the tag identifier, but the invention is not limited thereto. For ease of understanding, the following embodiments may be described in the context of memory device 30 performing access control with a fixed burst length.

Fig. 2 is a diagram of a system 100 for controlling data response by means of at least one attribute (attribute) of a transaction identifier according to an embodiment of the invention. As shown in fig. 2, the system 100 may include a slave device 140 and one or more master devices such as N master devices 120_1, …, and 120_ N, where N is a positive integer. The slave device 140 may include an atomizer 141, a read reorder buffer 142, and a storage device 143, wherein the atomizer 10, the read reorder buffer 20, and the storage device 30 may be exemplified by the atomizer 141, the read reorder buffer 142, and the storage device 143 shown in fig. 1, respectively. In this embodiment, the storage device 143 can be used to store data. In the present embodiment, the one or more masters, such as masters 120_1, … and 120_ N, may be coupled to the slave device 140 via a transmission interface circuit 160 (e.g., a Network-on-Chip (NoC) or bus/interconnect (interconnect) circuit) to retrieve data stored in the storage device 143. For example, the master device 120_1 may be coupled to the transport interface circuit 160 through the master interface 162_1, the master device 120_ N may be coupled to the transport interface circuit 160 through the master interface 162_ N, and the slave device 140 may be coupled to the transport interface circuit 160 through the slave interface 164, but the invention is not limited thereto. In this embodiment, any one of the masters 120_1, … and 120_ N (e.g., master 120_ N, where N is any integer between 1 and N) may transmit the at least one attribute along with the transaction id to the slave device 140. For example, the Master device 120_1 may send a transaction identification code such as Tag identification code Master _ Tag _ ID (0) by using the Master core circuit therein, and add the attribute Is _ unique _ ID (0) of the Tag identification code Master _ Tag _ ID (0) by using the core wrapper therein via an extension field, and the Master interface 162_1 may add the attribute Master _ ID (0) of the Tag identification code Master _ Tag _ ID (0) via an extension field; in this way, the Master device 120_ N may send a transaction identifier such as Tag identifier Master _ Tag _ ID (N) by using the Master core circuit therein, and add the attribute Is _ unique _ ID (N) of the Tag identifier Master _ Tag _ ID (0) by using the core wrapper therein via the extension field, and the Master interface 162_ N may add the attribute Master _ ID (0) of the Tag identifier Master _ Tag _ ID (0) via the extension field. The Tag identification code Master _ Tag _ id (n) may represent a transaction identification code issued from a Master core circuit in the Master device 120_ n, the attribute Is _ unique _ id (n) may indicate whether the Tag identification code Master _ Tag _ id (n) Is unique in the entire system 100, and the attribute Master _ id (n) indicates that the Master _ Tag _ id (n) Is issued from the Master device 120_ n.

In detail, if any two transaction identifiers are the same, it indicates that the two transaction identifiers must respond to the corresponding data in an ordered manner (e.g., respond according to a predetermined order); if the transaction identifiers are different, the corresponding data representing the transaction identifiers may be reordered (e.g., not necessarily responding in the predetermined order described above) to optimize the efficiency of the data response. When an instruction has a need to quickly obtain response data, a system developer may assign the instruction a unique identifier indicating that the response data corresponding to the identifier may be reordered to minimize latency.

In order to enable the slave device 140 to determine whether the Tag identification code Master _ Tag _ id (n) Is a unique identification code in the system 100 without a lookup table (lookup table), the Master device 120_ n may transmit the attribute Is _ unique _ id (n) to the slave device 140 together with the Tag identification code Master _ Tag _ id (n) from the Master device 120_ n. According to the attribute Is _ unique _ id (n), the slave device 140 may determine whether to utilize a specific data path of the plurality of data paths to respond the data corresponding to the Tag identification code Master _ Tag _ id (n) from the storage device 143 to the Master device 120_ n, wherein the specific data path Is a data path with the highest transmission efficiency among the plurality of data paths. For example, when the attribute Is _ unique _ id (n) indicates that the Tag identifier Master _ Tag _ id (n) Is not unique in the system 100 (e.g., Is _ unique _ id (n) Is "0"), the read reorder buffer 142 may receive data corresponding to the Tag identifier Master _ Tag _ id (n) from the storage device 143 to allow the data to be responded to from the slave device 140 to the Master device 120_ n in an ordered manner (e.g., according to the predetermined order). For another example, when the attribute Is _ unique _ id (n) indicates that the Tag identifier Master _ Tag _ id (n) Is unique in the system 100 (e.g., Is _ unique _ id (n) Is "1"), the slave device 140 may utilize a bypass path to directly respond data corresponding to the Tag identifier Master _ Tag _ id (n) from the storage device 143 to the Master device 120_ n without passing through the read reorder buffer 142. This means that the data corresponding to the unique transaction identifier is not unresponsive because other data in the read reorder buffer 142 is not yet ready, thereby ensuring data response efficiency for the unique transaction identifier.

Fig. 3 is a flowchart of a method for controlling data response by means of at least one attribute of a Tag identification code Tag _ ID according to an embodiment of the present invention, fig. 4 is some implementation details of the slave device 140 shown in fig. 2 in a case where the Tag identification code Tag _ ID is unique in the system 100, and fig. 5 is some implementation details of the slave device 140 shown in fig. 2 in a case where the Tag identification code Tag _ ID is not unique in the system 100, according to an embodiment of the present invention, wherein the length of the Tag identification code Tag _ ID in this embodiment is smaller than or equal to the burst length of the memory device 140 during access control. It should be noted that the workflow shown in fig. 3 is for illustrative purposes only and is not a limitation of the present invention, and one or more steps may be added, deleted or modified in the workflow. Moreover, if the same results were obtained, the steps need not be performed in the exact order shown in FIG. 3. In addition, the Tag identification code Tag _ ID may be an example of the Tag identification codes { Master _ Tag _ ID (n), Master _ ID (n), Is _ unique _ ID (n) } shown in fig. 2, wherein the lowest bit of the Tag identification code Tag _ ID (denoted by Tag _ ID [0]) may be an example of Is _ unique _ ID (n), and the other bits of the Tag identification code Tag _ ID (denoted by Tag _ ID [ k:1 ]) may be examples of the Tag _ ID (n) and the Master _ ID (n) }. In addition, the read ROB 142 may be integrated with other related control circuits in a tag ID mapping and read ROB controller 200, and particularly, the read ROB 142 may include storage areas 142A and 142B, but the invention is not limited thereto.

In step 302, the slave device 140 can read the instruction with the Tag identification code Tag _ ID.

In step 304, the slave device 140 determines whether the Tag identification code Tag _ ID is unique in the system 100, for example, whether the Tag identification code Tag _ ID is unique in the system 100 according to Tag _ ID [0 ]. If yes, go to step 312; otherwise, step 306 is entered.

In step 306, the slave device 140 may check whether the read ROB 142 (e.g., the storage areas 142A and/or 142B) is full. If so, indicating that the command cannot be received at this time, go to step 308; otherwise, step 310 is entered.

In step 308, since the slave device 140 (e.g., the read reorder buffer 142 therein) is not ready to receive data corresponding to the instruction and the response, the flow returns to step 306 for reconfirmation until the read reorder buffer 142 has space.

In step 310, since Tag _ ID [0] indicates that Tag ID Tag _ ID is not unique in system 100 (e.g., Tag ID [0] is 1 ' b0), slave device 140 may allocate one or more entries of read reorder buffer 142 to store Tag ID Tag _ ID, and a control circuit 220 in Tag ID mapping and read reorder buffer controller 200, after adding padding bits, sends the index cmd _ index and attribute Tag ID [0] (e.g., 1 ' b0) of the one or more entries to storage device 143, as shown in fig. 4, by reading ID { cmd _ index,1 ' b0 }. Additionally, each of the one or more entries may include a valid bit, Tag _ ID [ k:1], Tag _ ID [0], data valid counter, write pointer W _ pointer, read pointer R _ pointer, although the invention is not limited thereto.

In step 312, since Tag _ ID [0] indicates that Tag identification code Tag _ ID is unique in system 100 (e.g., Tag _ ID [0] is 1 'b 1), slave device 140 may bypass read reorder buffer 142, and control circuitry 220 within slave device 140 may send Tag identification code Tag _ ID directly to storage device 143 with read identification code { Tag _ ID [ k:1], 1' b1} after adding padding bits, as shown in FIG. 5.

In step 314, the storage device 143 may respond to the data according to the received command (e.g., the read identification code).

In step 316, the storage device 143 determines whether the tag identification code corresponding to the data is unique according to the lowest bit of the response identification code. If yes, for example, if the response ID is { Tag _ ID [ k:1], 1' b1} as shown in FIG. 5, go to step 322; otherwise, for example, in response to the identification code being { cmd _ index, 1' b0} as shown in FIG. 4, proceed to step 318.

In step 318, the slave device 140 may store the response data in the storage area 142B of the read ROB 142 with the read ROB write index { cmd _ index, W _ pointer } (which may correspond to the index RRB _ index of the storage area 142B), as shown in FIG. 4.

In step 320, the slave device 140 may output valid data from the top entry of the read reorder buffer 142 (e.g., storage area 142A and/or 142B) and the Tag identification Tag _ ID by Tag identification mapping and reading a selection circuit 240 (labeled "MUX" for clarity) in the reorder buffer 200, for example, reading the valid data from the storage area 142B according to the read index { cmd _ index, R _ pointer } (which may correspond to the index RRB _ index of the storage area 142B), as shown in fig. 4.

In step 322, the slave device 140 may bypass the read reorder buffer 140 and remove the padding bits from the tag identification. That is, the selection circuit 240 uses a bypass path to directly output the Tag identification code Tag _ ID and the response data from the storage device 143, avoiding the data from being transmitted through the read reorder buffer 142 (i.e., the data does not enter the read reorder buffer 142), as shown in fig. 5.

In step 324, the slave device 140 may respond data to the master device (e.g., master device 120_ n).

As described above, since the memory device 143 performs access control with the fixed burst length and the length of the Tag identification code Tag _ ID is smaller than or equal to the fixed burst length, if the Tag identification code Tag _ ID is unique in the system 100 (Tag _ ID [0] is 1' b1), the Tag identification code mapping and read reordering buffer controller 200 can directly transmit the Tag identification code Tag _ ID to the memory device 143 without storing the Tag identification code in the read reordering buffer 142 (e.g., the storage area 142A), and can directly respond data corresponding to the Tag identification code Tag _ ID from the memory device 143 to the host device 120_ n in response to the data without transmitting through the read reordering buffer 142. Thus, when the Tag identification code Tag _ ID is unique in the system 100, the slave device 140 can respond to the data with optimized efficiency without being blocked for waiting to read the data in the reorder buffer 142.

Fig. 6 is a flowchart of a method for controlling data response by means of at least one attribute of the Tag identification code Tag _ ID according to an embodiment of the present invention. In this embodiment, the length of the Tag identification code Tag _ ID is greater than the length of the Tag identification code used when the access control is performed on the storage device 140. It should be noted that the workflow shown in fig. 6 is for illustrative purposes only and is not a limitation of the present invention, and one or more steps may be added, deleted or modified in the workflow. Moreover, if the same results were obtained, the steps need not be performed in the exact order shown in FIG. 6. Fig. 7 is some implementation details of the slave device 140 shown in fig. 2 in a case where the Tag identification code Tag _ ID is unique in the system 100 according to an embodiment of the present invention. In addition, when the length of the Tag identification code Tag _ ID is greater than the length of the Tag identification code used when the storage device 140 performs access control, the implementation details of the slave device 140 in the case where the Tag identification code Tag _ ID is not unique in the system 100 can be referred to fig. 4.

In step 402, the slave device 140 may read the instruction with the Tag identification code Tag _ ID.

In step 406, the slave device 140 may check whether the read ROB 142 (e.g., the storage areas 142A and/or 142B) is full. If yes, indicating that the command cannot be received at this time, go to step 408; otherwise, step 410 is entered.

In step 408, since the slave device 140 (e.g., the read reorder buffer 142 therein) is not ready to receive data corresponding to the instruction and the response, the flow returns to step 406 for reconfirmation until the read reorder buffer 142 has space.

In step 410, whether the Tag identification code Tag _ ID is unique in the system 100 or not, the slave device 140 may allocate the one or more entries of the read reorder buffer 142 to store the Tag identification code Tag _ ID, and the control circuit 220 may send the index cmd _ index and the attribute Tag _ ID [0] (e.g., 1 'b 0 or 1' b1) of the one or more entries to the storage device 143 using the read identification code { cmd _ index, Tag _ ID [0] }. For example, the control circuit 220 sends { cmd _ index, 1' b0}, as shown in FIG. 4; as another example, the control circuit 220 sends { cmd _ index, 1' b1}, as shown in FIG. 7.

In step 414, the storage device 143 may respond to the data according to the received command (e.g., the read identifier).

In step 416, the storage device 143 determines whether the tag identification code corresponding to the data is unique according to the lowest bit of the response identification code. If yes, for example, as shown in FIG. 7, enter step 422 in response to the identification code being { cmd _ index, 1' b1 }; otherwise, for example, in response to the identification code being { cmd _ index, 1' b0} as shown in FIG. 4, step 418 is entered.

In step 418, the slave device 140 may store the response data in the storage area 142B of the read reorder buffer 142 with the read reorder buffer write index { cmd _ index, W _ pointer } (which may correspond to the index RRB _ index of the storage area 142B), as shown in fig. 4.

In step 420, the slave device 140 may output valid data and the Tag identification code Tag _ ID from the top entry of the read reorder buffer 142 (e.g., the storage area 142A and/or 142B) via the Tag identification code mapping and read select circuit 240 in the reorder buffer controller 200, for example, read valid data from the storage area 142B according to the read index { cmd _ index, R _ pointer } (which may correspond to the index RRB _ index of the storage area 142B), as shown in fig. 4.

In step 422, the slave device 140 may use the index cmd _ index to retrieve the Tag identification Tag _ ID from the read reorder buffer 142 (e.g., the storage area 142A therein), and directly respond to the data corresponding to the Tag identification Tag _ ID and the Tag identification Tag _ ID to avoid transferring data through the read reorder buffer 142 (i.e., the data does not enter the read reorder buffer 142 (e.g., the storage area 142B therein)), as shown in fig. 5.

In step 424, the slave device 140 may respond with data to the master device (e.g., master device 120_ n).

As described above, since the access control is performed by the memory device 143 with the fixed burst length and the length of the Tag identification code Tag _ ID is greater than the Tag identification code used by the fixed burst length, if the Tag identification code Tag _ ID is unique in the system 100 (Tag _ ID [0] is 1' b1), the atomizer 141 can intercept the Tag identification code Tag _ ID to generate a plurality of burst interception instructions (e.g., fixed-size read instructions) with the fixed burst length, so as to allow the Tag identification code mapping and read reorder buffer controller 200 to store the Tag identification code Tag _ ID in the read reorder buffer 142 (e.g., storage area 142A), and use the plurality of burst interception instructions to obtain the data corresponding to the Tag identification code Tag _ ID from the memory device 143, and use the specific data path (e.g., bypass path) to reorder the Tag identification code Tag _ ID stored in the storage area 142A in the reorder read reorder buffer 142 Along with data corresponding to the Tag identification code Tag _ ID to the master device 120_ n. Thus, when the Tag identification code Tag _ ID is unique in the system 100, the slave device 140 can respond to the data with optimized efficiency without being blocked from waiting for the data in the reorder buffer 142 to be read in response to the data.

In addition to the attribute Is _ unique _ id (n), the slave device 140 may control the data response by referring to the attribute Master _ id (n). For example, some tag identifiers are not unique in the system 100, but if the tag identifiers come from different masters (the attribute Master _ id (n) of the tag identifiers are different from each other), the response sequence of the data corresponding to the tag identifiers can be re-ordered. Fig. 8 is some implementation details of the slave device 140 shown in fig. 2 in a case where the Tag identification code Tag _ ID is not unique in the system 100 according to an embodiment of the present invention.

For example, in step 410, the slave device 140 may allocate the one or more entries of the read reorder buffer 142 and Group the entries according to the attribute Master _ ID (n) (e.g., map the attribute Master _ ID (n) to the Group ID by a mapping circuit 260 in the Tag ID mapping and read reorder buffer controller 200, as shown in fig. 8) to store the Tag ID Tag _ ID and the Group ID Group _ ID using the index cmd _ index, and the control circuit 220 may send the index cmd _ index and the attribute Tag _ ID [0] (e.g., 1 'b 0 or 1' b1) of the one or more entries to the storage device 143 using the read ID Tag _ index and Group ID Group _ ID { 0 }. In addition, in step 420, the slave device 140 may select a corresponding Group of the plurality of groups in the storage area 142B according to the Group identification code Group _ ID by mapping and reading the Tag identification code and the selection circuit 280 (labeled as "MUX" for simplicity) in the reorder buffer controller 200 through the Tag identification code, and then output the valid data and the Tag identification code Tag _ ID from the top entry of the read reorder buffer 142 (e.g., the corresponding Group in the storage areas 142A and/or 142B) by using the selection circuit 240, as shown in fig. 8.

As described above, since the read reorder buffer 142 is grouped into groups corresponding to the masters 120_1, … and 120_ N, the transaction identifiers of the data response operations from different masters can be performed independently, so that the slave device 140 can reorder the corresponding response data when receiving the read commands from different masters to achieve the optimal performance.

In addition, in some embodiments, the transmission interface circuit shown in FIG. 2 may include multiple data paths, and the multiple data paths may include a dedicated data path (which may be exemplary of the specific data paths described above) dedicated to transmitting data corresponding to a unique transaction identifier.

In addition, in the above embodiment, the at least one attribute (e.g., attribute Is _ unique _ ID (n)) Is transmitted using an extension field (e.g., Tag _ ID [0]) of the transaction ID. In other embodiments, the at least one attribute may be transmitted using a sideband (sideband) channel. It is within the scope of the present invention that all information regarding the attributes of the transaction identifier (e.g., whether the transaction identifier is unique in the system 100) be transmitted to the slave device 140 so that the slave device 140 can determine whether to provide a dedicated channel to respond to the data.

To summarize, the master device 120_ n can transmit the command to the slave device 140 along with the attribute information of the command when transmitting the command, so as to allow the slave device 140 to provide a corresponding data response path according to the attribute information, thereby achieving an optimized performance. For example, when a portion of the attribute information indicates that the tag identification of the instruction is unique in the system, the data corresponding to the instruction may be responded to via a dedicated signal path; for another example, when the portion of the attribute information indicates that the tag identifier of the command is not unique in the system, the slave device 140 may determine which master device the command is from based on another portion of the command to allow data requested by different master devices to respond in an out-of-order manner. Therefore, the invention can fully exert the advantage of responding data in an out-of-order mode and improve the overall efficiency of the system.

The above-mentioned embodiments are merely preferred embodiments of the present invention, and all equivalent changes and modifications made by the claims of the present application should be covered by the present invention.

19页详细技术资料下载
上一篇:一种医用注射器针头装配设备
下一篇:垂直定向正交连接系统的封装

网友询问留言

已有0条留言

还没有人留言评论。精彩留言会获得点赞!

精彩留言,会给你点赞!