Orthogonal data transposition system and method during data transfers to/from a processing array

Information

  • Patent Grant
  • 10860320
  • Patent Number
    10,860,320
  • Date Filed
    Tuesday, October 2, 2018
    6 years ago
  • Date Issued
    Tuesday, December 8, 2020
    3 years ago
Abstract
A device and method for facilitating orthogonal data transposition during data transfers to/from a processing array and a storage memory since the data words processed by the processing array (using computational memory cells) are stored orthogonally to how the data words are stored in storage memory. Thus, when data words are transferred between storage memory and the processing array, a mechanism orthogonally transposes the data words.
Description
FIELD

The disclosure relates generally to a device and method for connecting a processing array to a storage memory and in particular to a device and method that 1) transposes the data words output from the processing array for storage in the storage memory; and 2) transposes the data words from the storage to the processing array since the storage memory and the processing array store data differently.


BACKGROUND

A storage memory in a computer system or another conventional storage memory may be connected to each other by a data bus, DBus(63:0), as shown in FIG. 1. As shown in FIG. 1A, the data words in the storage memory are stored in the same orientation as the data words in the other conventional storage memory.


In contrast, each bit line in a processing array essentially functions as a mini-processor and has a plurality of computational memory cells connected to each bit line. In a processing array with the plurality of computational memory cells, reading multiple computational memory cells along the same bit line simultaneously produces a logical function (e.g. logical AND) of the memory cell contents on the read bit line. Additional circuitry can be implemented around the bit line and its associated memory cells to enable more complex logical operations on the data stored in those memory cells. The processing array may have a plurality of sections wherein each section has a plurality of bit line sections and each bit line section has a plurality of computational memory cells whose read bit lines are connected together to produce the logical function. Because the bit line is the central processing element in the processing array, and because all bit lines within each section receive the same control signals and therefore perform the same computations on their respective data, data words are stored in the processing array along bit lines (with each bit on a separate word line)—either entirely on the same bit line within a section or along the same relative bit line across multiple sections.


For example, one way to add two 16-bit data words is to store the first LSB of each data word in memory cells on bit line 0 in section 0, the second LSB of each data word in memory cells on bit line 0 in section 1, etc. Then, the software algorithm performs the logic and shift operations necessary to add the two 16-bit data words together; shift operations are needed to shift the carry bit result from section 0 to section 1 after adding the first LSBs in section 0, to shift the carry bit result from section 1 to section 2 after adding the carry bit from section 0 to the second L2Bs in section 1, etc. In this way, if the processing array consists of 16 sections with “n” bit lines per section, then “n” simultaneous 16-bit ADD operations can be performed.


The way that data words are stored in the processing array—i.e. along the same bit line (with each bit on a separate word line)—is orthogonal to how data words are conventionally stored in a memory array used exclusively for storage (e.g. storage memory)—i.e. along the same word line (with each bit on a separate bit line). Consequently, when data words are transferred between storage memory and the processing array, a mechanism is needed to orthogonally transpose the data and it is to that end that this disclosure is directed.





BRIEF DESCRIPTION OF THE DRAWINGS


FIG. 1 depicts storage memory connected to other conventional storage memory. In this depiction, a 64-bit data bus facilitates data transfers between the two memory arrays.



FIG. 1A is an expansion of FIG. 1 that illustrates the data byte layout in storage memory and the other conventional storage memory when data is transferred between them.



FIG. 2 depicts storage memory connected to a processing array via an intermediate transposer that orthogonally transposes the data during data transfers between the two memory arrays. In this depiction, a 64-bit data bus facilitates data transfers between storage memory and the transposer, and an n-bit data bus facilitates data transfers between the transposer and the processing array.



FIGS. 3A-3C illustrate an example of the processing array shown in FIGS. 2, 4, 6A and 6B that may be used with the transposer.



FIG. 4 is an expansion of FIG. 2 that depicts the transposer constructed as a 64-row by n-column array of register bits. The diagram illustrates the data byte layout in storage memory, the transposer, and the processing array when data is transferred between them.



FIG. 5 depicts an exemplary implementation of the transposer from FIG. 2 or 4 constructed as a two-way shift register. The diagram illustrates how data is column-shifted through the transposer during data transfers between it and storage memory, and how data is row-shifted through the transposer during data transfers between it and the processing array.



FIG. 6A depicts storage memory connected to a processing array via an intermediate transposer that orthogonally transposes the data during data transfers between the two memory arrays, and via an intermediate “buffer” that buffers data transfers between the transposer and the processing array. In this depiction, a 64-bit data bus facilitates data transfers between storage memory and the transposer, a 64-bit data bus facilitates data transfers between the transposer and the buffer, and an n-bit data bus facilitates data transfers between the buffer and the processing array.



FIG. 6B is an expansion of FIG. 6A that depicts the transposer constructed as a 64-row by 64-column array of register bits, and that depicts the buffer constructed as a 64-row by n-column array of storage elements. The diagram illustrates the data byte layout in storage memory, the transposer, the buffer, and the processing array when data is transferred between them.



FIG. 6C depicts the transposer from FIG. 6B constructed as a two-way shift register. The diagram illustrates how data is column-shifted through the transposer during data transfers between it and storage memory, and how data is row-shifted through the transposer during data transfers between it and the buffer.



FIG. 6D depicts the detailed circuit implementation of the transposer in FIG. 6C.



FIG. 7 depicts an alternate form of the transposer depicted in FIGS. 3a & 3b, constructed as a 32-row by 64-column array of register bits, and as a two-way shift register.



FIG. 8 depicts an alternate form of the transposer depicted in FIGS. 5 and 6C, constructed as a 16-row by 64-column array of register bits, and as a two-way shift register.



FIG. 9 depicts an alternate form of the transposer depicted in FIGS. 5 and 6C, constructed as an 8-row by 64-column array of register bits, and as a two-way shift register.





DETAILED DESCRIPTION OF ONE OR MORE EMBODIMENTS

The disclosure is particularly applicable to a processing array device using SRAM memory cells in which data words are stored along word lines in storage memory and other conventional storage memory, and data words are stored along bit lines in the processing array and it is in this context that the disclosure will be described. It will be appreciated, however, that the apparatus and method has greater utility, such as to being used with other processing array devices. The device and method may orthogonally transpose the data transferred between the storage memory and the processing array device. The device and method may use a transposer or the combination of a transposer and a buffer to perform the orthogonal transposing of the data. In the embodiments described below, the data being transposed may be a certain number of bit units of data that are transferred between the storage memory and the processing array device. Examples using a sixty four bit unit of data, a thirty two bit unit of data, a sixteen bit unit of data and an eight bit unit of data are provided below although the device and method may operate with other different sized units of data. For purposes of the disclosure, all of the above different sized bit units of data that may be transferred by the device and method will be known as a unit of data.


FIRST EMBODIMENT


FIG. 2 depicts a first embodiment of a processing unit 20 that includes a storage memory 22 connected to a processing array 30 (constructed from a plurality of computational memory cells) via a data path 26, that may be an intermediate transposer 26 in one implementation, that orthogonally transposes the data during data transfers between the storage memory array 22 and the processing array 30 since data words are stored along word lines in storage memory 22 and other conventional storage memory and data words are stored along bit lines in the processing array 30. In this depiction, a 64-bit data bus 24, DBus(63:0) for example shown in FIG. 2, facilitates data transfers between storage memory 22 and the transposer 26, and an n-bit data bus 28, TBus(n−1:0) for example shown in FIG. 2, facilitates data transfers between the transposer 26 and the processing array 30. The elements of the processing unit 20 shown in FIG. 2 may be housed in a single semiconductor chip/integrated circuit (with a single die or multiple dies) or multiple integrated circuits. Alternatively, the processing array 30 and transposer 26 may be part of the integrated processing unit 20 (a single integrated circuit or semiconductor chip with one or multiple dies) while the storage memory 22 may be separate.


One way to orthogonally transpose data during data transfers between storage memory 22 and the processing array 30 is to implement a two-dimensional storage array block between them that facilitates the orthogonal data transposition. This intermediate storage block is henceforth referred to as the “transposer” 26. When a transposer is utilized in this manner, data transfers between storage memory 22 and the processing array 30 may be accomplished in two steps:


1. A data transfer between storage memory 22 and the transposer 26.


2. A data transfer between the transposer 26 and the processing array 30.


The execution order of the two steps depends on the direction of data transfer, storage memory→processing array or processing array→storage memory.


An example of the processing array 30 that may be used with the processing unit 20 and transposer 26 is shown in FIGS. 3A-3C although it is understood that the transposer 26 may be used with other processing arrays 30 is a similar manner that are within the scope of this disclosure. FIG. 3A illustrates an example of a processing array 30 with computational memory cells in an array that may be incorporated into a semiconductor memory or computer system and may include transposer circuitry. The processing array 30 may include an array of computational memory cells (cell 00, . . . , cell 0n and cell m0, . . . , cell mn). In one embodiment, the array of computational memory cells may be rectangular as shown in FIG. 3A and may have a plurality of columns and a plurality of rows wherein the computational memory cells in a particular column may also be connected to the same read bit line (RBL0, . . . , RBLn). The processing array 30 may further include a wordline (WL) generator and read/write logic control circuit 32 that may be connected to and generate signals for the read word line (RE) and write word line (WE) for each memory cell (such as RE0, . . . , REn and WE0, . . . , WEn) to control the read and write operations is well known and one or more read/write circuitry 34 that are connected to the read and write bit lines of the computational memory cells. The wordline (WL) generator and read/write logic control circuit 32 may also generate one or more control signals that control each read/write circuitry 34.


In the embodiment shown in FIG. 3A, the processing array may have read/write circuitry 34 for each set of bit line signals of the computational memory cells (e.g., for each column of the computational memory cells whose read bit lines are connected to each other). For example, BL0 read/write logic 340 may be coupled to the read and write bit lines (WBLb0, WBL0 and RBL0) for the computational memory cells in column 0 of the array and BLn read/write logic 34n may be coupled to the read and write bit lines (WBLbn, WBLn and RBLn) for the computational memory cells in column n of the array as shown in FIG. 3A.


During a read operation, the wordline (WL) generator and read/write logic control circuit 32 may activate one or more word lines that activate one or more computational memory cells so that the read bit lines of those one or more computational memory cells may be read out. Further details of the read operation are not provided here since the read operation is well known.


Each computational memory cell in the processing array may be a static random access memory (SRAM) cell based computational memory cell that is able to perform a computation as described above. It is noted that the processing array 30 may be constructed using other different types of memory cells. The details of an exemplary computational memory cell that may be used as part of the processing array 30 may be found in co-pending U.S. patent application Ser. No. 15/709,399, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations”, U.S. patent application Ser. No. 15/709,401, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations”, U.S. patent application Ser. No. 15/709,379, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, U.S. patent application Ser. No. 15/709,382, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, and U.S. patent application Ser. No. 15/709,385, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells” and U.S. Provisional Patent Application No. 62/430,767, filed Dec. 6, 2016 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations” and U.S. Provisional Patent Application No. 62/430,762, filed Dec. 6, 2016 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, all of which are incorporated by reference herein.



FIGS. 3B and 3C illustrate the processing array 30 with computational memory cells having sections having the same elements as shown in FIG. 3A. The array 30 in FIG. 3B has one section (Section 0) with “n” bit lines (bit line 0 (BL0), . . . , bit line n (BLn)) in different bit line sections (b1-sect), where each bit line connects to “m” computational memory cells (cell 00, . . . , cell m0 for bit line 0, for example). In the example in FIG. 3B, the m cells may be the plurality of computational memory cells that are part of each column of the array 30. FIG. 3C illustrates the processing array 30 with computational memory cells having multiple sections. In the example in FIG. 3C, the processing array device 30 comprises “k” sections with “n” bit lines each, where each bit line within each section connects to “m” computational memory cells. Note that the other elements of the processing array 30 are present in FIG. 3C, but not shown for clarity. In FIG. 3C, the BL-Sect(0,0) block shown corresponds to the BL-Sect(0,0) shown in FIG. 3B with the plurality of computational memory cells and the read/write logic 340 and each other block shown in FIG. 3C corresponds to a separate portion of the processing array. As shown in FIG. 3C, the set of control signals, generated by the wordline generator and read/write logic controller 32, for each section may include one or more read enable control signals (for example S[0]_RE[m:0] for section 0), one or more write enable control signals (for example S[0]_WE[m:0] for section 0) and one or more read/write control signals (for example S[0]_RW_Ctrl[p:0] for section 0). As shown in FIG. 3C, the array 30 may have a plurality of sections (0, . . . , k in the example in FIG. 3C) and each section may have multiple bit line sections (0, . . . , n per section, in the example in FIG. 3C).


Returning to FIG. 2, in the first embodiment, the transposer 26 is constructed as a two-way shift register array with “x” rows and “y” columns. The number of rows “x” is equal to the width of the data bus “DBus” that connects the transposer 26 to storage memory 22; e.g. x=64 if DBus=64 bits. The number of columns “y” is equal to the number of columns (i.e. bit lines) in a processing array section—e.g. y=n if a processing array section has “n” bit lines, and establishes the width of the data bus “TBus” that connects the transposer to the processing array—e.g. if y=n then TBus=n bits. The number of columns in the transposer is equal to the number of columns in a processing array section so that all columns associated with a particular row in a particular processing array section can be written to or read from simultaneously, thereby eliminating the need for any sort of column-addressability within a processing array section.


An example of the transposer 26 constructed as a 64-row by n-column array of register bits, where “n” is equal to the number of columns (bit lines) in a section of the processing array 30 is shown in FIG. 4. FIG. 4 illustrates that the initial data byte layout in storage memory 22, or after “n” units of 64-bit column data have been transferred from the transposer to storage memory, the data byte layout in the transposer after “n” units of 64-bit column data have been transferred from storage memory to the transposer, or after 64 units of n-bit row data have been transferred from the processing array to the transposer; and the data byte layout in the processing array after 64 units of n-bit row data have been transferred from the transposer to the processing array.


Operation of First Embodiment


The processing unit 20 with the storage memory 22, the transposer 26 and the processing array 30 may be operated to transfer data from the storage memory 22 to the processing array 30 and to transfer data from the processing array 30 to the storage memory 22 using different processes, each of which is now described in more detail.


During Storage Memory 22 to Processing Array 30 Data Transfers




  • 1. When data is transferred from storage memory 22 to the transposer 26 on the 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:
    • The 64-bit unit of transfer data on DBus(63:0) is stored in column “n−1”, rows 63:0 of the transposer.
    • The 64-bit units of previously-loaded data in columns 1 through “n−1” of the transposer are column-shifted to columns 0 through “n−2”.
    • The 64-bit unit of previously-loaded data in column 0, rows 63:0 of the transposer is discarded.



This procedure repeats until all “n” columns of the transposer have been loaded with storage memory data. The column shifting is the first of the two ways in which the transposer can be shifted.

  • 2. When data is subsequently transferred from the transposer 26 to the processing array 30 on an n-bit data bus “TBus(n−1:0)”, the following occurs simultaneously:
    • The n-bit unit of data in row 0, columns n−1:0 of the transposer is output to TBus(n−1:0).
    • The n-bit units of data in rows 1 through 63 of the transposer are row-shifted to rows 0 through 62.
    • A logic “0” is stored in row 63, columns n−1:0 of the transposer. This is arbitrary—the state of the transposer doesn't matter after storage memory data has been transferred to the processing array.


This procedure repeats until all 64 rows of transposer data have been transferred to the processing array 30. The row shifting is the second of the two ways in which the transposer can be shifted.


During Processing Array 30 to Storage Memory 22 Data Transfers:




  • 1. When data is transferred from the processing array to the transposer on a n-bit data bus “TBus(n−1:0)”, the following occurs simultaneously:
    • The n-bit unit of transfer data on TBus(n−1:0) is stored in row 63, columns n−1:0 of the transposer.
    • The n-bit units of previously-loaded data in rows 1 through 63 of the transposer are row-shifted to rows 0 through 62.
    • The n-bit unit of previously-loaded data in row 0, columns n−1:0 of the transposer is discarded.



This procedure repeats until all 64 rows of the transposer have been loaded with processing array data.

  • 2. When data is subsequently transferred from the transposer to storage memory on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:
    • The 64-bit unit of data in column 0, rows 63:0 of the transposer is output to DBus(63:0).
    • The 64-bit units of data in columns 1 through “n−1” of the transposer are column-shifted to columns 0 through “n−2”.
    • A logic “0” is stored in column “n−1”, rows 63:0 of the transposer. This is arbitrary—the state of the transposer doesn't matter after the processing array data has been transferred to storage memory.


This procedure repeats until all “n” columns of transposer data have been transferred to storage memory.


Transposer Exemplary Implementation



FIG. 5 depicts an exemplary implementation of the transposer 26 from FIG. 2 or 4 constructed as a two-way shift register. The diagram illustrates how data is column-shifted through the transposer during data transfers between it and storage memory, and how data is row-shifted through the transposer during data transfers between it and the processing array. More specifically, FIG. 5 shows the two data buses of FIG. 2 and:

    • The data byte layout in the transposer after “n” units of 64-bit column data have been transferred from storage memory to the transposer, or after 64 units of n-bit row data have been transferred from the processing array to the transposer.
    • How column-shifts are used when writing data into the transposer from DBus(63:0), reading data from the transposer onto DBus(63:0), and/or shifting data through the transposer during data transfers between it and storage memory.
    • How row-shifts are used when writing data into the transposer from TBus(n−1:0), reading data from the transposer onto TBus(n−1:0), and/or shifting data through the transposer during data transfers between it and the processing array.


In the first embodiment, the transposer 26 has the same number of columns as a processing array section. In that case, the column pitch of the transposer should match that of the processing array to allow for a simple data connection between them. However, that may be difficult to implement in some transposer and processing array designs without wasting die area, due to the differences in the circuit design of the two blocks.


A second embodiment described below adds a buffer between the transposer and the processing array with the same number of columns as the processing array and allows for the transposer to have fewer columns than the processing array with no particular column pitch requirements. The second embodiment still eliminates the need for column-addressability in the processing array.


SECOND EMBODIMENT

A second embodiment implements a second way to orthogonally transpose data during data transfers between storage memory 22 and the processing array 30 by using the transposer 26 and a buffer 29 between them as shown in FIG. 6A. The transposer 26 facilitates the orthogonal data transposition (as in the first embodiment), and the buffer 29 buffers data transfers between the transposer 26 and the processing array 30. In this depiction, a 64-bit data bus, such as Dbus(63:0) for example shown in FIG. 6A, facilitates data transfers between storage memory 22 and the transposer 26, a 64-bit data bus, TBus(63:0) for example shown in FIG. 6A, facilitates data transfers between the transposer 26 and the buffer 29, and an n-bit data bus, PBus(n−1:0) for example shown in FIG. 6A, facilitates data transfers between the buffer 29 and the processing array 30.


In the second embodiment, the transposer 26 may have fewer columns than the processing array section 30 while the buffer 29 has the same number of columns as a processing array section 30 as shown in FIG. 6B, thereby eliminating the need for column-addressability in the processing array (as in the first embodiment). The buffer 29 is a simpler circuit than both the transposer 26 and the processing array 30, and therefore easier to implement with the same column pitch as the processing array without wasting die area. In addition, the buffer 29 is a conventional storage circuit (unlike the processing array) so that it is easier to support the column-addressability needed therein to facilitate data transfers between it and the (fewer-column) transposer.


When the transposer 26 and buffer 29 are utilized in this manner, data transfers between storage memory 22 and the processing array 30 are accomplished in three steps:


1. The data transfer between storage memory 22 and the transposer 26.


2. The data transfer between the transposer 26 and the buffer 29.


3. The data transfer between the buffer 29 and the processing array 30.


The execution order of the three steps depends on the direction of data transfer, storage memory→processing array or processing array→storage memory.


In the second embodiment, as shown in FIG. 6C, the transposer 26 is constructed as a two-way shift register array with 64 rows (to match the width of the data bus that connects it storage memory, as in the first embodiment) and 64 columns. The number of columns “64” is a whole fraction of the number of columns “n” in the buffer, and establishes the width of the data bus “TBus” that connects the transposer to the buffer. As shown in FIG. 6C, the buffer 29 is constructed as an array of storage elements with 64 rows to match the number of rows in the transposer 26, and “n” columns to match the number of columns in a processing array section 30 and establish the width of the data bus “PBus” that connects the buffer to the processing array.


During Storage Memory→Processing Array Data Transfers


1. When data is transferred from storage memory to the transposer on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:

    • The 64-bit unit of transfer data on DBus(63:0) is stored in column 63, rows 63:0 of the transposer.
    • The 64-bit units of previously-loaded data in columns 1 through 63 of the transposer are column-shifted to columns 0 through 62.
    • The 64-bit unit of previously-loaded data in column 0, rows 63:0 of the transposer is discarded.


This procedure repeats until all 64 columns of the transposer have been loaded with storage memory data.


2. When data is subsequently transferred from the transposer to the buffer on a 64-bit data bus “TBus(63:0)”, the following occurs simultaneously:

    • The 64-bit unit of data in row 0, columns 63:0 of the transposer is output to TBus(63:0).
    • The 64-bit units of data in rows 1 through 63 of the transposer are row-shifted to rows 0 through 62.
    • The 64-bit unit of data in row 0, columns 63:0 of the transposer is stored in row 63, columns 63:0 in the transposer. This facilitates a “row wrap” feature in the transposer.


This procedure repeats until all 64 rows of transposer data have been transferred to the buffer.


The “row wrap” feature means that after all 64 rows of transposer data have been transferred to the buffer, the transposer contains the same data as it did before the transfer started, as if the transposer data had been copied to the buffer. Such an implementation allows for the same transposer data to be copied to multiple 64-column groups in the buffer and, ultimately, in the processing array, without having to reload the transposer from storage memory each time. This is a desirable feature in some use cases.


3. Data is subsequently transferred from the buffer to the processing array row by row, on an n-bit data bus “PBus(n−1:0)”. This is accomplished via conventional means, and beyond the scope of this disclosure.


Steps 1˜3 may repeat until storage memory data has been transferred to all “n” columns of the processing array. Step 1 of the next iteration may overlap with step 3 of the previous iteration—i.e. storage memory to transposer transfers may overlap with buffer to processing array transfers.


During Processing Array→Storage Memory Data Transfers


1. Data is initially transferred from the processing array to the buffer row by row, on an n-bit data bus “PBus(n−1:0)”. This is accomplished via conventional means, and beyond the scope of this disclosure.


2. When data is subsequently transferred from the buffer to the transposer on a 64-bit data bus “TBus(63:0)”, the following occurs simultaneously:

    • The 64-bit unit of transfer data on TBus(63:0) is stored in row 63, columns 63:0 of the transposer.
    • The 64-bit units of previously-loaded data in rows 1 through 63 of the transposer are row-shifted to rows 0 through 62.
    • The 64-bit unit of previously-loaded data in row 0, columns 63:0 of the transposer is discarded.


This procedure repeats until all 64 rows of the transposer have been loaded with processing array data.


3. When data is subsequently transferred from the transposer to storage memory on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:

    • The 64-bit unit of data in column 0, rows 63:0 of the transposer is output to DBus(63:0).
    • The 64-bit units of data in columns 1 through 63 of the transposer are column-shifted to columns 0 through 62.
    • A logic “0” is stored in column 63, rows 63:0 of the transposer. This is arbitrary—it doesn't matter what data resides in the transposer after it has been transferred to storage memory.


This procedure repeats until all 64 columns of transposer data have been sent to storage memory.


Steps 1˜3 may repeat until all “n” columns of processing array data have been transferred to storage memory. Step 1 of the next iteration may overlap with step 3 of the previous iteration—i.e. processing array to buffer transfers may overlap with transposer to storage memory transfers.


Second Embodiment Transposer and Buffer Details



FIG. 6B depicts the transposer 26 constructed as a 64-row by 64-column array of register bits. FIG. 6B illustrates: the initial data byte layout in storage memory, or after “n” units of 64-bit column data have been transferred from the transposer to storage memory; the data byte layout in the transposer after 64 units of 64-bit column data have been transferred from storage memory to the transposer, or after 64 units of 64-bit row data have been transferred from the buffer to the transposer; the data byte layout in the buffer after (n/64)*64 units of 64-bit row data have been transferred from the transposer to the buffer, or after 64 units of n-bit row data have been transferred from the processing array to the buffer; and the data byte layout in the processing array after 64 units of n-bit row data have been transferred from the buffer to the processing array.



FIG. 6C depicts the transposer implemented as a two-way shift register and illustrates:

    • The data byte layout in the transposer after 64 units of 64-bit column data have been transferred from storage memory to the transposer, or after 64 units of 64-bit row data have been transferred from the buffer to the transposer.
    • The data byte layout in the buffer after 64 units of 64-bit row data have been transferred from the transposer to the buffer.
    • How column-shifts are used when writing data into the transposer from DBus(63:0), reading data from the transposer onto DBus(63:0), and/or shifting data through the transposer during data transfers between it and storage memory.
    • How row-shifts are used when writing data into the transposer from TBus(63:0), reading data from the transposer onto TBus(63:0), and/or shifting data through the transposer during data transfers between it and the buffer.



FIG. 6D depicts an example of an implementation of the transposer of the second embodiment with detailed circuit of the transposer in FIG. 6C. In this implementation, multiplexers (Muxes) as shown in FIG. 6D are used to select the data input to each register bit in the transposer 26, to facilitate its column-shift and row-shift capabilities. The mux input that is selected as the data input to each register bit depends on whether:

    • A “column write” is occurring—i.e. a data transfer from storage memory to the transposer.
    • A “column read” is occurring—i.e. a data transfer from the transposer to storage memory.
    • A “row write” is occurring—i.e. a data transfer from the buffer to the transposer.
    • A “row read” is occurring—i.e. a data transfer from the transposer to the buffer.


For example, for row 63, column 63, a 4:1 mux 600 is utilized to select one of four data input sources:

    • DBus(63) is selected during column write.
    • A logic “0” is selected during column read.
    • TBus(63) is selected during row write.
    • Row 0, column 63 is selected during row read.


For rows 62:0, column 63, a 3:1 mux 602 for each row and column pair is utilized to select one of three data input sources:

    • For row “x”: DBus(x) is selected during column write.
    • For row “x”: a logic “0” is selected during column read.
    • For row “x”: row “x+1”, column 63 is selected during row read and row write.


For row 63, columns 62:0, a 3:1 mux 604 for each row and column pair is utilized to select one of three data input sources:

    • For column “y”: row 63, column “y+1” is selected during column write and column read.
    • For column “y”: TBus(m) is selected during row write.
    • For column “y”: row 0, column “y” is selected during row read.


For rows 62:0, columns 62:0, a 2:1 mux for each column and row pair is utilized to select one of two data input sources:

    • For row “x”, column “y”: row “x”, column “y+1” is selected during column write and column read.
    • For row “x”, column “y”: row “x+1”, column “y” is selected during row write and row read.


In the first and second embodiments, all 64 bits of data captured in the transposer 26 during any single data transfer from storage memory 22 to the transposer 26 are ultimately stored along the same bit line in the processing array 30. That is desirable if the processing array 30 is used to process 64-bit data words. But if the processing array 30 is used to process, say, 32-bit data words, then it is desirable to be able to store each 32-bit data word captured in the transposer during a sequence of data transfers from storage memory to the transposer in a different bit line in the processing array. That is not possible with the transposers described in the first and second embodiments, because each pair of 32-bit data words comprising the 64 bits of data captured in the transposer during any single data transfer from storage memory to the transposer are ultimately stored along the same bit line in the processing array. However, the below disclosed third, fourth, and fifth embodiments disclose implementations of transposers to transpose 32-bit, 16-bit, and 8-bit data words onto separate bit lines in the processing array when the data bus that connects storage memory to the transposer is 64 bits regardless of the data word size.


THIRD EMBODIMENT


FIG. 7 depicts an alternate form of the transposer 26 and buffer 29 constructed as a 32-row by 64-column array of register bits and as a two-way shift register. In the third embodiment, as in the second embodiment, a transposer 26 (with fewer columns that the processing array) and a buffer 29 (with the same number of columns as the processing array) are implemented between storage memory 22 and the processing array 30 as shown in FIG. 6A. The difference is that the transposer 26 in the third embodiment is constructed to transpose 32-bit data words onto separate bit lines in the processing array 30 when the data bus that connects storage memory 22 to the transposer 26 remains 64 bits. Specifically, the transposer is constructed as a two-way shift register array with 32 rows (instead of 64 rows) and 64 columns. In addition, a portion of the buffer 29 is unused since the buffer 29 may be used for different embodiments of the transposer shown in FIGS. 6C and 7-9.


During Storage Memory→Processing Array Data Transfers


1. When data is transferred from storage memory to the transposer on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:

    • The 32-bit unit of transfer data on DBus(63:32) is stored in column 63, rows 31:0 of the transposer.
    • The 32-bit unit of transfer data on DBus(31:0) is stored in column 62, rows 31:0 of the transposer.
    • The 32-bit units of previously-loaded data in columns 3, 5, 7, . . . 63 of the transposer are column-shifted to columns 1, 3, 5, . . . 61.
    • The 32-bit units of previously-loaded data in columns 2, 4, 6, . . . 62 of the transposer are column-shifted to columns 0, 2, 3, . . . 60.
    • The 32-bit unit of previously-loaded data in column 1, rows 31:0 of the transposer is discarded.
    • The 32-bit unit of previously-loaded data in column 0, rows 31:0 of the transposer is discarded.


This procedure repeats until all 64 columns of the transposer have been loaded with storage memory data.


2. When data is subsequently transferred from the transposer to the buffer on a 64-bit data bus “TBus(63:0)”, the following occurs simultaneously:

    • The 64-bit unit of data in row 0, columns 63:0 of the transposer is output to TBus(63:0).
    • The 64-bit units of data in rows 1 through 31 of the transposer are row-shifted to rows 0 through 30.
    • The 64-bit unit of data in row 0, columns 63:0 of the transposer is stored in row 31, columns 63:0 in the transposer. This facilitates the “row wrap” feature in the transposer, as in the second embodiment.


This procedure repeats until all 32 rows of transposer data have been transferred to the buffer.


Note that the buffer may still have 64 rows, as in the second embodiment, but only 32 rows are utilized.


3. Data is subsequently transferred from the buffer to the processing array row by row, on an n-bit data bus “PBus(n−1:0)”. This is accomplished via conventional means, and beyond the scope of this disclosure.


Steps 1˜3 may repeat until storage memory data has been transferred to all “n” columns of the processing array.


During Processing Array→Storage Memory Data Transfers


1. Data is initially transferred from the processing array to the buffer row by row, on an n-bit data bus “PBus(n−1:0)”. This is accomplished via conventional means, and beyond the scope of this disclosure.


2. When data is subsequently transferred from the buffer to the transposer on a 64-bit data bus “TBus(63:0)”, the following occurs simultaneously:

    • The 64-bit unit of transfer data on TBus(63:0) is stored in row 31, columns 63:0 of the transposer.
    • The 64-bit units of previously-loaded data in rows 1 through 31 of the transposer are row-shifted to rows 0 through 30.
    • The 64-bit unit of previously-loaded data in row 0, columns 63:0 of the transposer is discarded.


This procedure repeats until all 32 rows of the transposer have been loaded with processing array data.


3. When data is subsequently transferred from the transposer to storage memory on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:

    • The 32-bit unit of data in column 1, rows 31:0 of the transposer is output to DBus(63:32).
    • The 32-bit unit of data in column 0, rows 31:0 of the transposer is output to DBus(31:0).
    • The 32-bit units of data in columns 3, 5, 7, . . . 63 of the transposer are column-shifted to columns 1, 3, 5, . . . 61.
    • The 32-bit units of data in columns 2, 4, 6, . . . 62 of the transposer are column-shifted to columns 0, 2, 4, . . . 60.
    • A logic “0” is stored in column 63, rows 31:0 of the transposer.
    • A logic “0” is stored in column 62, rows 31:0 of the transposer.


This procedure repeats until all 64 columns of transposer data have been sent to storage memory.


Steps 1˜3 may repeat until all “n” columns of processing array data have been transferred to storage memory.



FIG. 7 depicts the transposer constructed as a 32-row by 64-column array of register bits, and as a two-way shift register. It illustrates:

    • The data byte layout in the transposer after 32 units of 64-bit column data have been transferred from storage memory to the transposer, or after 32 units of 64-bit row data have been transferred from the buffer to the transposer.
    • The data byte layout in the buffer after 32 units of 64-bit row data have been transferred from the transposer to the buffer.
    • How column-shifts are used when writing data into the transposer from DBus(63:0), reading data from the transposer onto DBus(63:0), and/or shifting data through the transposer during data transfers between it and storage memory.
    • How row-shifts are used when writing data into the transposer from TBus(63:0), reading data from the transposer onto TBus(63:0), and/or shifting data through the transposer during data transfers between it and the buffer.


FOURTH EMBODIMENT


FIG. 8 depicts an alternate form of the transposer depicted in FIGS. 5 and 6C, constructed as a 16-row by 64-column array of register bits, and as a two-way shift register. In the fourth embodiment, as in the second embodiment, a transposer (with fewer columns that the processing array) and a buffer (with the same number of columns as the processing array) are implemented between storage memory and the processing array. The difference is, the transposer in the fourth embodiment is constructed to transpose 16-bit data words onto separate bit lines in the processing array when the data bus that connects storage memory to the transposer remains 64 bits. Specifically, the transposer is constructed as a two-way shift register array with 16 rows (instead of 64 rows) and 64 columns. In addition, a larger portion of the buffer 29 is unused than with the third embodiment since this embodiment is 16 bits instead of 32 bits.


During Storage Memory→Processing Array Data Transfers


1. When data is transferred from storage memory to the transposer on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:

    • The 16-bit unit of transfer data on DBus(63:48) is stored in column 63, rows 15:0 of the transposer.
    • The 16-bit unit of transfer data on DBus(47:32) is stored in column 62, rows 15:0 of the transposer.
    • The 16-bit unit of transfer data on DBus(31:16) is stored in column 61, rows 15:0 of the transposer.
    • The 16-bit unit of transfer data on DBus(15:0) is stored in column 60, rows 15:0 of the transposer.
    • The 16-bit units of previously-loaded data in columns 7, 11, 15, . . . 63 of the transposer are column-shifted to columns 3, 7, 11, . . . 59.
    • The 16-bit units of previously-loaded data in columns 6, 10, 14, . . . 62 of the transposer are column-shifted to columns 2, 6, 10, . . . 58.
    • The 16-bit units of previously-loaded data in columns 5, 9, 13, . . . 61 of the transposer are column-shifted to columns 1, 5, 9, . . . 57.
    • The 16-bit units of previously-loaded data in columns 4, 8, 12, . . . 60 of the transposer are column-shifted to columns 0, 4, 9, . . . 56.
    • The 16-bit unit of previously-loaded data in column 3, rows 15:0 of the transposer is discarded.
    • The 16-bit unit of previously-loaded data in column 2, rows 15:0 of the transposer is discarded.
    • The 16-bit unit of previously-loaded data in column 1, rows 15:0 of the transposer is discarded.
    • The 16-bit unit of previously-loaded data in column 0, rows 15:0 of the transposer is discarded.


This procedure repeats until all 64 columns of the transposer have been loaded with storage memory data.


2. When data is subsequently transferred from the transposer to the buffer on a 64-bit data bus “TBus(63:0)”, the following occurs simultaneously:

    • The 64-bit unit of data in row 0, columns 63:0 of the transposer is output to TBus(63:0).
    • The 64-bit units of data in rows 1 through 15 of the transposer are row-shifted to rows 0 through 14.
    • The 64-bit unit of data in row 0, columns 63:0 of the transposer is stored in row 15, columns 63:0 in the transposer. This facilitates the “row wrap” feature in the transposer, as in the second embodiment.


This procedure repeats until all 16 rows of transposer data have been transferred to the buffer.


Note that the buffer may still have 64 rows, as in the second embodiment, but only 16 rows are utilized.


3. Data is subsequently transferred from the buffer to the processing array row by row, on an n-bit data bus “PBus(n−1:0)”. This is accomplished via conventional means, and beyond the scope of this disclosure.


Steps 1˜3 may repeat until storage memory data has been transferred to all “n” columns of the processing array.


During Processing Array→Storage Memory Data Transfers


1. Data is initially transferred from the processing array to the buffer row by row, on an n-bit data bus “PBus(n−1:0)”. This is accomplished via conventional means, and beyond the scope of this disclosure.


2. When data is subsequently transferred from the buffer to the transposer on a 64-bit data bus “TBus(63:0)”, the following occurs simultaneously:

    • The 64-bit unit of transfer data on TBus(63:0) is stored in row 15, columns 63:0 of the transposer.
    • The 64-bit units of previously-loaded data in rows 1 through 15 of the transposer are row-shifted to rows 0 through 14.
    • The 64-bit unit of previously-loaded data in row 0, columns 63:0 of the transposer is discarded.


This procedure repeats until all 16 rows of the transposer have been loaded with processing array data.


3. When data is subsequently transferred from the transposer to storage memory on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:

    • The 16-bit unit of data in column 3, rows 15:0 of the transposer is output to DBus(63:48).
    • The 16-bit unit of data in column 2, rows 15:0 of the transposer is output to DBus(47:32).
    • The 16-bit unit of data in column 1, rows 15:0 of the transposer is output to DBus(31:16).
    • The 16-bit unit of data in column 0, rows 15:0 of the transposer is output to DBus(15:0).
    • The 16-bit units of data in columns 7, 11, 15, . . . 63 of the transposer are column-shifted to columns 3, 7, 11, . . . 59.
    • The 16-bit units of data in columns 6, 10, 14, . . . 62 of the transposer are column-shifted to columns 2, 6, 10, . . . 58.
    • The 16-bit units of data in columns 5, 9, 13, . . . 61 of the transposer are column-shifted to columns 1, 5, 9, . . . 57.
    • The 16-bit units of data in columns 4, 8, 12, . . . 60 of the transposer are column-shifted to columns 0, 4, 9, . . . 56.
    • A logic “0” is stored in column 63, rows 15:0 of the transposer.
    • A logic “0” is stored in column 62, rows 15:0 of the transposer.
    • A logic “0” is stored in column 61, rows 15:0 of the transposer.
    • A logic “0” is stored in column 60, rows 15:0 of the transposer.


This procedure repeats until all 64 columns of transposer data have been sent to storage memory.


Steps 1˜3 may repeat until all “n” columns of processing array data have been transferred to storage memory.


FIFTH EMBODIMENT


FIG. 9 depicts an alternate form of the transposer depicted in FIGS. 5 and 6C, constructed as an 8-row by 64-column array of register bits, and as a two-way shift register. In the fifth embodiment, as in the second embodiment, a transposer (with fewer columns that the processing array) and a buffer (with the same number of columns as the processing array) are implemented between storage memory and the processing array. The difference is, the transposer in the fifth embodiment is constructed to transpose 8-bit data words onto separate bit lines in the processing array when the data bus that connects storage memory to the transposer remains 64 bits. Specifically, the transposer is constructed as a two-way shift register array with 8 rows (instead of 64 rows) and 64 columns. In addition, a still larger portion of the buffer 29 is unused than with the third or fourth embodiments since this embodiment is 8 bits instead of 32 bits or 16 bits.


During Storage Memory→Processing Array Data Transfers




  • 1. When data is transferred from storage memory to the transposer on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:
    • The 8-bit unit of transfer data on DBus(63:56) is stored in column 63, rows 7:0 of the transposer.
    • The 8-bit unit of transfer data on DBus(55:48) is stored in column 62, rows 7:0 of the transposer.
    • The 8-bit unit of transfer data on DBus(47:40) is stored in column 61, rows 7:0 of the transposer.
    • The 8-bit unit of transfer data on DBus(39:32) is stored in column 60, rows 7:0 of the transposer.
    • The 8-bit unit of transfer data on DBus(31:24) is stored in column 59, rows 7:0 of the transposer.
    • The 8-bit unit of transfer data on DBus(23:16) is stored in column 58, rows 7:0 of the transposer.
    • The 8-bit unit of transfer data on DBus(15:8) is stored in column 57, rows 7:0 of the transposer.
    • The 8-bit unit of transfer data on DBus(7:0) is stored in column 56, rows 7:0 of the transposer.
    • The 8-bit units of previously-loaded data in columns 15, 23, 31, . . . 63 of the transposer are column-shifted to columns 7, 15, 23, . . . 55.
    • The 8-bit units of previously-loaded data in columns 14, 22, 30, . . . 62 of the transposer are column-shifted to columns 6, 14, 22, . . . 54.
    • The 8-bit units of previously-loaded data in columns 13, 21, 29, . . . 61 of the transposer are column-shifted to columns 5, 13, 21, . . . 53.
    • The 8-bit units of previously-loaded data in columns 12, 20, 28, . . . 60 of the transposer are column-shifted to columns 4, 12, 20, . . . 52.
    • The 8-bit units of previously-loaded data in columns 11, 19, 27, . . . 59 of the transposer are column-shifted to columns 3, 11, 19, . . . 51.
    • The 8-bit units of previously-loaded data in columns 10, 18, 26, . . . 58 of the transposer are column-shifted to columns 2, 10, 18, . . . 50.
    • The 8-bit units of previously-loaded data in columns 9, 17, 25, . . . 57 of the transposer are column-shifted to columns 1, 9, 17, . . . 49.
    • The 8-bit units of previously-loaded data in columns 8, 16, 24, . . . 56 of the transposer are column-shifted to columns 0, 8, 16, . . . 48.
    • The 8-bit unit of previously-loaded data in column 7, rows 7:0 of the transposer is discarded.
    • The 8-bit unit of previously-loaded data in column 6, rows 7:0 of the transposer is discarded.
    • The 8-bit unit of previously-loaded data in column 5, rows 7:0 of the transposer is discarded.
    • The 8-bit unit of previously-loaded data in column 4, rows 7:0 of the transposer is discarded.
    • The 8-bit unit of previously-loaded data in column 3, rows 7:0 of the transposer is discarded.
    • The 8-bit unit of previously-loaded data in column 2, rows 7:0 of the transposer is discarded.
    • The 8-bit unit of previously-loaded data in column 1, rows 7:0 of the transposer is discarded.
    • The 8-bit unit of previously-loaded data in column 0, rows 7:0 of the transposer is discarded.


      This procedure repeats until all 64 columns of the transposer have been loaded with storage memory data.

  • 2. When data is subsequently transferred from the transposer to the buffer on a 64-bit data bus “TBus(63:0)”, the following occurs simultaneously:
    • The 64-bit unit of data in row 0, columns 63:0 of the transposer is output to TBus(63:0).
    • The 64-bit units of data in rows 1 through 7 of the transposer are row-shifted to rows 0 through 6.
    • The 64-bit unit of data in row 0, columns 63:0 of the transposer is stored in row 7, columns 63:0 in the transposer. This facilitates the “row wrap” feature in the transposer, as in the second embodiment.


      This procedure repeats until all 8 rows of transposer data have been transferred to the buffer.


      Note that the buffer may still have 64 rows, as in the second embodiment, but only 8 rows are utilized.

  • 3. Data is subsequently transferred from the buffer to the processing array row by row, on an n-bit data bus “PBus(n−1:0)”. This is accomplished via conventional means, and beyond the scope of this disclosure.



Steps 1˜3 may repeat until storage memory data has been transferred to all “n” columns of the processing array.


During Processing Array→Storage Memory Data Transfers




  • 1. Data is initially transferred from the processing array to the buffer row by row, on an n-bit data bus “PBus(n−1:0)”. This is accomplished via conventional means, and beyond the scope of this disclosure.

  • 2. When data is subsequently transferred from the buffer to the transposer on a 64-bit data bus “TBus(63:0)”, the following occurs simultaneously:
    • The 64-bit unit of transfer data on TBus(63:0) is stored in row 7, columns 63:0 of the transposer.
    • The 64-bit units of previously-loaded data in rows 1 through 7 of the transposer are row-shifted to rows 0 through 6.
    • The 64-bit unit of previously-loaded data in row 0, columns 63:0 of the transposer is discarded.



This procedure repeats until all 8 rows of the transposer have been loaded with processing array data.

  • 3. When data is subsequently transferred from the transposer to storage memory on a 64-bit data bus “DBus(63:0)”, the following occurs simultaneously:
    • The 8-bit unit of data in column 7, rows 7:0 of the transposer is output to DBus(63:56).
    • The 8-bit unit of data in column 6, rows 7:0 of the transposer is output to DBus(55:48).
    • The 8-bit unit of data in column 5, rows 7:0 of the transposer is output to DBus(47:40).
    • The 8-bit unit of data in column 4, rows 7:0 of the transposer is output to DBus(39:32).
    • The 8-bit unit of data in column 3, rows 7:0 of the transposer is output to DBus(31:24).
    • The 8-bit unit of data in column 2, rows 7:0 of the transposer is output to DBus(23:16).
    • The 8-bit unit of data in column 1, rows 7:0 of the transposer is output to DBus(15:8).
    • The 8-bit unit of data in column 0, rows 7:0 of the transposer is output to DBus(7:0).
    • The 8-bit units of data in columns 15, 23, 31, . . . 63 of the transposer are column-shifted to columns 7, 15, 23, . . . 55.
    • The 8-bit units of data in columns 14, 22, 30, . . . 62 of the transposer are column-shifted to columns 6, 14, 22, . . . 54.
    • The 8-bit units of data in columns 13, 21, 29, . . . 61 of the transposer are column-shifted to columns 5, 13, 21, . . . 53.
    • The 8-bit units of data in columns 12, 20, 28, . . . 60 of the transposer are column-shifted to columns 4, 12, 20, . . . 52.
    • The 8-bit units of data in columns 11, 19, 27, . . . 59 of the transposer are column-shifted to columns 3, 11, 19, . . . 51.
    • The 8-bit units of data in columns 10, 18, 26, . . . 58 of the transposer are column-shifted to columns 2, 10, 18, . . . 50.
    • The 8-bit units of data in columns 9, 17, 25, . . . 57 of the transposer are column-shifted to columns 1, 9, 17, . . . 49.
    • The 8-bit units of data in columns 8, 16, 24, . . . 56 of the transposer are column-shifted to columns 0, 8, 16, . . . 48.
    • A logic “0” is stored in column 63, rows 7:0 of the transposer.
    • A logic “0” is stored in column 62, rows 7:0 of the transposer.
    • A logic “0” is stored in column 61, rows 7:0 of the transposer.
    • A logic “0” is stored in column 60, rows 7:0 of the transposer.
    • A logic “0” is stored in column 59, rows 7:0 of the transposer.
    • A logic “0” is stored in column 58, rows 7:0 of the transposer.
    • A logic “0” is stored in column 57, rows 7:0 of the transposer.
    • A logic “0” is stored in column 56, rows 7:0 of the transposer.
    • This procedure repeats until all 64 columns of transposer data have been sent to storage memory.


      Steps 1˜3 may repeat until all “n” columns of processing array data have been transferred to storage memory.


The multiple embodiments of the transposers 26 (e.g. embodiments 2, 3, 4, and 5) may be implemented in a single design in which a processing array is implemented. In this case only one transposer is enabled/selected during any particular storage memory <−> processing array data transfer.


The foregoing description, for purpose of explanation, has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the disclosure to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the disclosure and its practical applications, to thereby enable others skilled in the art to best utilize the disclosure and various embodiments with various modifications as are suited to the particular use contemplated.


The system and method disclosed herein may be implemented via one or more components, systems, servers, appliances, other subcomponents, or distributed between such elements. When implemented as a system, such systems may include an/or involve, inter alia, components such as software modules, general-purpose CPU, RAM, etc. found in general-purpose computers. In implementations where the innovations reside on a server, such a server may include or involve components such as CPU, RAM, etc., such as those found in general-purpose computers.


Additionally, the system and method herein may be achieved via implementations with disparate or entirely different software, hardware and/or firmware components, beyond that set forth above. With regard to such other components (e.g., software, processing components, etc.) and/or computer-readable media associated with or embodying the present inventions, for example, aspects of the innovations herein may be implemented consistent with numerous general purpose or special purpose computing systems or configurations. Various exemplary computing systems, environments, and/or configurations that may be suitable for use with the innovations herein may include, but are not limited to: software or other components within or embodied on personal computers, servers or server computing devices such as routing/connectivity components, hand-held or laptop devices, multiprocessor systems, microprocessor-based systems, set top boxes, consumer electronic devices, network PCs, other existing computer platforms, distributed computing environments that include one or more of the above systems or devices, etc.


In some instances, aspects of the system and method may be achieved via or performed by logic and/or logic instructions including program modules, executed in association with such components or circuitry, for example. In general, program modules may include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular instructions herein. The inventions may also be practiced in the context of distributed software, computer, or circuit settings where circuitry is connected via communication buses, circuitry or links. In distributed settings, control/instructions may occur from both local and remote computer storage media including memory storage devices.


The software, circuitry and components herein may also include and/or utilize one or more type of computer readable media. Computer readable media can be any available media that is resident on, associable with, or can be accessed by such circuits and/or computing components. By way of example, and not limitation, computer readable media may comprise computer storage media and communication media. Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and can accessed by computing component. Communication media may comprise computer readable instructions, data structures, program modules and/or other components. Further, communication media may include wired media such as a wired network or direct-wired connection, however no media of any such type herein includes transitory media. Combinations of the any of the above are also included within the scope of computer readable media.


In the present description, the terms component, module, device, etc. may refer to any type of logical or functional software elements, circuits, blocks and/or processes that may be implemented in a variety of ways. For example, the functions of various circuits and/or blocks can be combined with one another into any other number of modules. Each module may even be implemented as a software program stored on a tangible memory (e.g., random access memory, read only memory, CD-ROM memory, hard disk drive, etc.) to be read by a central processing unit to implement the functions of the innovations herein. Or, the modules can comprise programming instructions transmitted to a general purpose computer or to processing/graphics hardware via a transmission carrier wave. Also, the modules can be implemented as hardware logic circuitry implementing the functions encompassed by the innovations herein. Finally, the modules can be implemented using special purpose instructions (SIMD instructions), field programmable logic arrays or any mix thereof which provides the desired level performance and cost.


As disclosed herein, features consistent with the disclosure may be implemented via computer-hardware, software and/or firmware. For example, the systems and methods disclosed herein may be embodied in various forms including, for example, a data processor, such as a computer that also includes a database, digital electronic circuitry, firmware, software, or in combinations of them. Further, while some of the disclosed implementations describe specific hardware components, systems and methods consistent with the innovations herein may be implemented with any combination of hardware, software and/or firmware. Moreover, the above-noted features and other aspects and principles of the innovations herein may be implemented in various environments. Such environments and related applications may be specially constructed for performing the various routines, processes and/or operations according to the invention or they may include a general-purpose computer or computing platform selectively activated or reconfigured by code to provide the necessary functionality. The processes disclosed herein are not inherently related to any particular computer, network, architecture, environment, or other apparatus, and may be implemented by a suitable combination of hardware, software, and/or firmware. For example, various general-purpose machines may be used with programs written in accordance with teachings of the invention, or it may be more convenient to construct a specialized apparatus or system to perform the required methods and techniques.


Aspects of the method and system described herein, such as the logic, may also be implemented as functionality programmed into any of a variety of circuitry, including programmable logic devices (“PLDs”), such as field programmable gate arrays (“FPGAs”), programmable array logic (“PAL”) devices, electrically programmable logic and memory devices and standard cell-based devices, as well as application specific integrated circuits. Some other possibilities for implementing aspects include: memory devices, microcontrollers with memory (such as EEPROM), embedded microprocessors, firmware, software, etc. Furthermore, aspects may be embodied in microprocessors having software-based circuit emulation, discrete logic (sequential and combinatorial), custom devices, fuzzy (neural) logic, quantum devices, and hybrids of any of the above device types. The underlying device technologies may be provided in a variety of component types, e.g., metal-oxide semiconductor field-effect transistor (“MOSFET”) technologies like complementary metal-oxide semiconductor (“CMOS”), bipolar technologies like emitter-coupled logic (“ECL”), polymer technologies (e.g., silicon-conjugated polymer and metal-conjugated polymer-metal structures), mixed analog and digital, and so on.


It should also be noted that the various logic and/or functions disclosed herein may be enabled using any number of combinations of hardware, firmware, and/or as data and/or instructions embodied in various machine-readable or computer-readable media, in terms of their behavioral, register transfer, logic component, and/or other characteristics. Computer-readable media in which such formatted data and/or instructions may be embodied include, but are not limited to, non-volatile storage media in various forms (e.g., optical, magnetic or semiconductor storage media) though again does not include transitory media. Unless the context clearly requires otherwise, throughout the description, the words “comprise,” “comprising,” and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is to say, in a sense of “including, but not limited to.” Words using the singular or plural number also include the plural or singular number respectively. Additionally, the words “herein,” “hereunder,” “above,” “below,” and words of similar import refer to this application as a whole and not to any particular portions of this application. When the word “or” is used in reference to a list of two or more items, that word covers all of the following interpretations of the word: any of the items in the list, all of the items in the list and any combination of the items in the list.


Although certain presently preferred implementations of the invention have been specifically described herein, it will be apparent to those skilled in the art to which the invention pertains that variations and modifications of the various implementations shown and described herein may be made without departing from the spirit and scope of the invention. Accordingly, it is intended that the invention be limited only to the extent required by the applicable rules of law.


While the foregoing has been with reference to a particular embodiment of the disclosure, it will be appreciated by those skilled in the art that changes in this embodiment may be made without departing from the principles and spirit of the disclosure, the scope of which is defined by the appended claims.

Claims
  • 1. An apparatus, comprising: a storage memory having an array of memory cells, wherein a unit of data is stored in a plurality of memory cells connected to a same word line with each memory cell connected to a different bit line;a processing array device, connected to the storage memory, comprising a plurality of memory cells arranged in an array having a plurality of columns and a plurality of rows, each memory cell having a storage element wherein the array has a plurality of sections and each section has a plurality of bit line sections and a plurality of bit lines with one bit line per bit line section, wherein the memory cells in each bit line section are all connected to a single read bit line to perform a computation using the unit of data communicated between the processing array device and the storage memory;a data path between the storage memory and the processing array device having a transposer that orthogonally transposes the unit of data transferred between the storage memory and the processing array and wherein the transfer of the unit of data between the storage memory and the processing array device is performed in two steps, wherein the transposer shifts the unit of data from being stored along the same word line in the storage memory to the unit of data being stored along the same bit line in the processing array device; andwherein the processing array device stores the unit of data in memory cells connected to the single read bit line in a particular section of the processing array device or connected to a same relative bit line in a plurality of sections of the processing array device and each memory cell is connected to a different word line.
  • 2. The apparatus of claim 1, wherein the data transfer from the storage memory to the processing array device further comprises a first data transfer between the storage memory and the transposer and a second data transfer between the transposer and the processing array device.
  • 3. The apparatus of claim 1, wherein the processing array device and the transposer are integrated into a single processing unit.
  • 4. The apparatus of claim 1, wherein the transposer is a two-dimensional storage array block.
  • 5. The apparatus of claim 1 further comprising a data path between the storage memory and the processing array device having a transposer that orthogonally transposes the unit of data transferred between the storage memory and the processing array and a two-dimensional buffer that buffers the unit of data transfers between the transposer and the processing array device and wherein a transfer of the unit of data between the storage memory and the processing array device is performed in three steps.
  • 6. The apparatus of claim 5, wherein the data transfer from the storage memory to the processing array device comprising a first data transfer between the storage memory and the transposer, a second data transfer between the transposer and the buffer and a third data transfer between the buffer and the processing array device.
  • 7. The apparatus of claim 6, wherein the transposer is a two-way shift register array with “x” rows and “y” columns, wherein a number of “x” rows is equal to a width of a data bus that connects the transposer to the storage memory and wherein a number of “y” columns is a whole fraction of a number of columns in the processing array device section.
  • 8. The apparatus of claim 7, wherein the width of the data bus that connects the transposer to the storage memory is sixty four bits, wherein the buffer has an array of storage elements with a number of rows that match the width of the data bus that connects the transposer and storage memory and a number of columns that match the “n” columns in the processing array device section, wherein a width of the data bus that connects the transposer to the buffer is equal to the numbers of columns “y” in the transposer and wherein a width of the data bus that connects the buffer to the processing array device is equal to the number of columns “n” in the buffer and in a section of the processing array device.
  • 9. The apparatus of claim 7, wherein the transposer is constructed to transpose a thirty two bit unit of data from storage memory onto separate bit lines in the processing array device.
  • 10. The apparatus of claim 7, wherein the transposer is constructed to transpose a sixteen bit unit of data from storage memory onto separate bit lines in the processing array device.
  • 11. The apparatus of claim 7, wherein the transposer is constructed to transpose an eight bit unit of data from storage memory onto separate bit lines in the processing array device.
  • 12. The apparatus of claim 5, wherein the transposer further comprises one or more transposers wherein each transposer is one of a transposer that transposes a sixty-four bit unit of data, a transposer that transposes a thirty two bit unit of data, a transposer that transposes a sixteen bit unit of data and a transposer that transposes an eight bit unit of data.
  • 13. An apparatus, comprising: a storage memory having an array of memory cells, wherein a unit of data is stored in a plurality of memory cells connected to a same word line with each memory cell connected to a different bit line;a processing array device, connected to the storage memory, comprising a plurality of memory cells arranged in an array having a plurality of columns and a plurality of rows, each memory cell having a storage element wherein the array has a plurality of sections and each section has a plurality of bit line sections and a plurality of bit lines with one bit line per bit line section, wherein the memory cells in each bit line section are all connected to a single read bit line to perform a computation using the unit of data communicated between the processing array device and the storage memory;a data path between the storage memory and the processing array device having a transposer that orthogonally transposes the unit of data transferred between the storage memory and the processing array and wherein the transfer of the unit of data between the storage memory and the processing array device is performed in two steps, wherein the transposer is a two-way shift register array with “x” rows and “y” columns, wherein a number of “x” rows is equal to a width of a data bus that connects the transposer to the storage memory and wherein a number of “y” columns is equal to a number of bit lines in the processing array device; andwherein the processing array device stores the unit of data in memory cells connected to the single read bit line in a particular section of the processing array device or connected to a same relative bit line in a plurality of sections of the processing array device and each memory cell is connected to a different word line.
  • 14. The apparatus of claim 13, wherein the width of the data bus that connects the transposer to the storage memory is sixty four bits and wherein a width of a data bus that connects the transposer to the processing array device is equal to the “y” columns in the transposer and in the processing array device section.
  • 15. A method, comprising: providing a storage memory having an array of memory cells, wherein a unit of data is stored in a plurality of memory cells connected to a same word line with each memory cell connected to a different bit line;connecting a processing array device to the storage memory, the processing array device comprising a plurality of memory cells arranged in an array having a plurality of columns and a plurality of rows, each memory cell having a storage element wherein the array has a plurality of sections and each section has a plurality of bit line sections and a plurality of bit lines with one bit line per bit line section, wherein the memory cells in each bit line section are all connected to a single read bit line to perform a computation using a unit of data; andcommunicating the unit of data between the processing array device and the storage memory, wherein the processing array device stores the unit of data in memory cells connected to the single read bit line in a particular section of the processing array device or connected to a same relative bit line in a plurality of sections of the processing array device and each memory cell is connected to a different word line, wherein communicating the unit of data between the processing array device and the storage memory further comprises orthogonally transposing, using a transposer, the unit of data transferred between the storage memory and the processing array and wherein the transfer of the unit of data between the storage memory and the processing array device is performed in two steps and wherein orthogonally transposing the unit of data further comprises shifting the unit of data from being stored along the same word line in the storage memory to the unit of data being stored along the same bit line in the processing array device.
  • 16. The method of claim 15, wherein communicating the unit of data word between the processing array device and the storage memory further comprises transferring, for a transfer from storage memory to the processing array device, the unit of data from the storage memory to transposer and transferring, for a transfer from storage memory to the processing array device, from the transposer to the processing array device.
  • 17. The method of claim 16, wherein transferring, for the transfer from the storage memory to the transposer, the unit of data further comprises transferring, a sixty four bit unit of data on a 64 bit data bus between the storage memory and the transposer into column n−1 and rows 63:0 of the transposer, simultaneously column shifting, in the transposer, sixty four bit units of data stored in column 1:n−1 to column 0:n−2 and simultaneously discarding the sixty four bit unit of data stored in column 0 rows 63:0 of the transposer until all columns of the transposer are loaded with data from the storage memory.
  • 18. The method of claim 16, wherein transferring, for the transfer from the transposer to the processing array device further comprises transferring, an n bit unit of data from row 0 columns n−1:0 of the transposer on an n bit data bus between the transposer and the processing array device, simultaneously row shifting, in the transposer, n bit units of data in row 1:63 in the transposer to rows 0:62 in the transposer and simultaneously storing a logic “0” into row 63, columns n−1:0 of the transposer until all rows of the transposer are transferred to the processing array device.
  • 19. The method of claim 15, wherein communicating the unit of data between the processing array device and the storage memory further comprises transferring, for a transfer from the processing array device to the storage memory, the unit of data from the processing array device to the transposer and transferring, for a transfer from the processing array device to the storage memory, from the transposer to the storage memory.
  • 20. The method of claim 19, wherein transferring, for the transfer from the processing array device to the transposer, further comprises transferring, an n bit unit of data on a n-bit data bus between the processing array device and the transposer into row 63, columns n−1:0 of the transposer, simultaneously row shifting, in the transposer, the n bit units of data stored in row 1:63 of the transposer to rows 0:62 of the transposer and simultaneously discarding the n-bit unit of data stored in row 0, columns n−1:0 of the transposer until all rows of the transposer are loaded with data from the processing array device.
  • 21. The method of claim 19, wherein transferring, for the transfer from the transposer to the storage memory further comprises transferring, a sixty four bit unit of data in column 0, rows 63:0 of the transposer on a 64 bit data bus between the transposer and the storage memory, simultaneously column shifting, in the transposer, sixty four bit units of data in columns 1:n−1 of the transposer to columns 0:n−2 of the transposer and simultaneously storing a logic “0” into a set of rows of the transposer until all n columns of the transposer are transferred to the storage memory.
  • 22. A method, comprising: providing a storage memory having an array of memory cells, wherein a unit of data is stored in a plurality of memory cells connected to a same word line with each memory cell connected to a different bit line;connecting a processing array device to the storage memory, the processing array device comprising a plurality of memory cells arranged in an array having a plurality of columns and a plurality of rows, each memory cell having a storage element wherein the array has a plurality of sections and each section has a plurality of bit line sections and a plurality of bit lines with one bit line per bit line section, wherein the memory cells in each bit line section are all connected to a single read bit line to perform a computation using a unit of data; andcommunicating the unit of data between the processing array device and the storage memory, wherein the processing array device stores the unit of data in memory cells connected to the single bit line in a particular section of the processing array device or connected to a same relative bit line in a plurality of sections of the processing array device and each memory cell is connected to a different word line, wherein communicating the unit of data between the processing array device and the storage memory further comprises orthogonally transposing, using a transposer, the unit of data transferred between the storage memory and the processing array and wherein the transfer of the unit of data between the storage memory and the processing array device is performed in two steps, wherein the transposer is a two-way shift register array with “x” rows and “y” columns, wherein a number of “x” rows is equal to a width of a data bus that connects the transposer to the storage memory and wherein a number of “y” columns is equal to a number of bit lines in the processing array device.
PRIORITY CLAIM/RELATED APPLICATIONS

This application is a continuation in part of and claims priority under 35 USC 120 to U.S. patent application Ser. No. 15/709,399, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations”, U.S. patent application Ser. No. 15/709,401, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations”, U.S. patent application Ser. No. 15/709,379, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, U.S. patent application Ser. No. 15/709,382, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, and U.S. patent application Ser. No. 15/709,385, filed Sep. 19, 2017 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells” that in turn claim priority under 35 USC 119(e) and 120 and claim the benefit of U.S. Provisional Patent Application No. 62/430,767, filed Dec. 6, 2016 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells For Xor And Xnor Computations” and U.S. Provisional Patent Application No. 62/430,762, filed Dec. 6, 2016 and entitled “Computational Dual Port Sram Cell And Processing Array Device Using The Dual Port Sram Cells”, the entirety of all of which are incorporated herein by reference.

US Referenced Citations (396)
Number Name Date Kind
3451694 Hass Jun 1969 A
3747952 Graebe Jul 1973 A
3795412 John Mar 1974 A
4227717 Bouvier Oct 1980 A
4308505 Messerschmitt Dec 1981 A
4587496 Wolaver May 1986 A
4594564 Yarborough, Jr. Jun 1986 A
4677394 Vollmer Jun 1987 A
4716322 D'Arrigo et al. Dec 1987 A
4741006 Yamaguchi et al. Apr 1988 A
4856035 Lewis Aug 1989 A
5008636 Markinson Apr 1991 A
5302916 Pritchett Apr 1994 A
5375089 Lo Dec 1994 A
5382922 Gersbach Jan 1995 A
5400274 Jones Mar 1995 A
5473574 Clemen et al. Dec 1995 A
5530383 May Jun 1996 A
5535159 Nii Jul 1996 A
5563834 Longway et al. Oct 1996 A
5587672 Ranganathan et al. Dec 1996 A
5608354 Hori Mar 1997 A
5661419 Bhagwan Aug 1997 A
5696468 Nise Dec 1997 A
5736872 Sharma et al. Apr 1998 A
5744979 Goetting Apr 1998 A
5744991 Jefferson et al. Apr 1998 A
5748044 Xue May 1998 A
5768559 Iino et al. Jun 1998 A
5805912 Johnson et al. Sep 1998 A
5883853 Zheng et al. Mar 1999 A
5937204 Schinnerer Aug 1999 A
5942949 Wilson et al. Aug 1999 A
5963059 Partovi et al. Oct 1999 A
5969576 Trodden Oct 1999 A
5969986 Wong Oct 1999 A
5977801 Boerstler Nov 1999 A
5999458 Nishimura et al. Dec 1999 A
6005794 Sheffield et al. Dec 1999 A
6044034 Katakura Mar 2000 A
6058063 Jang May 2000 A
6072741 Taylor Jun 2000 A
6100721 Durec et al. Aug 2000 A
6100736 Wu et al. Aug 2000 A
6114920 Moon et al. Sep 2000 A
6115320 Mick et al. Sep 2000 A
6133770 Hasegawa Oct 2000 A
6167487 Camacho Dec 2000 A
6175282 Yasuda Jan 2001 B1
6226217 Riedlinger et al. May 2001 B1
6262937 Arcoleo et al. Jul 2001 B1
6263452 Jewett et al. Jul 2001 B1
6265902 Klemmer et al. Jul 2001 B1
6286077 Choi et al. Sep 2001 B1
6310880 Waller Oct 2001 B1
6366524 Abedifard Apr 2002 B1
6377127 Fukaishi et al. Apr 2002 B1
6381684 Hronik et al. Apr 2002 B1
6385122 Chang May 2002 B1
6407642 Dosho et al. Jun 2002 B2
6418077 Naven Jul 2002 B1
6441691 Jones et al. Aug 2002 B1
6448757 Hill Sep 2002 B2
6473334 Bailey et al. Oct 2002 B1
6483361 Chiu Nov 2002 B1
6504417 Cecchi et al. Jan 2003 B1
6538475 Johansen et al. Mar 2003 B1
6567338 Mick May 2003 B1
6594194 Gold Jul 2003 B2
6642747 Chiu Nov 2003 B1
6661267 Walker et al. Dec 2003 B2
6665222 Wright et al. Dec 2003 B2
6683502 Groen et al. Jan 2004 B1
6683930 Dalmia Jan 2004 B1
6732247 Berg et al. May 2004 B2
6744277 Chang et al. Jun 2004 B1
6757854 Zhao Jun 2004 B1
6789209 Suzuki et al. Sep 2004 B1
6816019 Delbo' et al. Nov 2004 B2
6836419 Loughmiller Dec 2004 B2
6838951 Nieri et al. Jan 2005 B1
6842396 Kono Jan 2005 B2
6853696 Moser et al. Feb 2005 B1
6854059 Gardner Feb 2005 B2
6856202 Lesso Feb 2005 B2
6859107 Moon et al. Feb 2005 B1
6882237 Singh et al. Apr 2005 B2
6897696 Chang et al. May 2005 B2
6933789 Molnar et al. Aug 2005 B2
6938142 Pawlowski Aug 2005 B2
6940328 Lin Sep 2005 B2
6954091 Wurzer Oct 2005 B2
6975554 Lapidus et al. Dec 2005 B1
6998922 Jensen et al. Feb 2006 B2
7002404 Gaggl et al. Feb 2006 B2
7002416 Pettersen et al. Feb 2006 B2
7003065 Homol et al. Feb 2006 B2
7017090 Endou et al. Mar 2006 B2
7019569 Fan-Jiang Mar 2006 B2
7042271 Chung et al. May 2006 B2
7042792 Lee et al. May 2006 B2
7042793 Masuo May 2006 B2
7046093 McDonagh et al. May 2006 B1
7047146 Chuang et al. May 2006 B2
7053666 Tak et al. May 2006 B2
7095287 Maxim et al. Aug 2006 B2
7099643 Lin Aug 2006 B2
7141961 Hirayama et al. Nov 2006 B2
7142477 Tran et al. Nov 2006 B1
7152009 Bokui et al. Dec 2006 B2
7180816 Park Feb 2007 B2
7200713 Cabot et al. Apr 2007 B2
7218157 Van De Beek et al. May 2007 B2
7233214 Kim et al. Jun 2007 B2
7246215 Lu et al. Jul 2007 B2
7263152 Miller et al. Aug 2007 B2
7269402 Uozumi et al. Sep 2007 B2
7282999 Da Dalt et al. Oct 2007 B2
7312629 Chuang et al. Dec 2007 B2
7313040 Chuang et al. Dec 2007 B2
7330080 Stoiber et al. Feb 2008 B1
7340577 Van Dyke et al. Mar 2008 B1
7349515 Chew et al. Mar 2008 B1
7352249 Balboni et al. Apr 2008 B2
7355482 Meltzer Apr 2008 B2
7355907 Chen et al. Apr 2008 B2
7369000 Wu et al. May 2008 B2
7375593 Self May 2008 B2
7389457 Chen et al. Jun 2008 B2
7439816 Lombaard Oct 2008 B1
7463101 Tung Dec 2008 B2
7464282 Abdollahi-Alibeik et al. Dec 2008 B1
7487315 Hur et al. Feb 2009 B2
7489164 Madurawe Feb 2009 B2
7512033 Hur et al. Mar 2009 B2
7516385 Chen et al. Apr 2009 B2
7538623 Jensen et al. May 2009 B2
7545223 Watanabe Jun 2009 B2
7565480 Ware et al. Jul 2009 B2
7577225 Azadet et al. Aug 2009 B2
7592847 Liu et al. Sep 2009 B2
7595657 Chuang et al. Sep 2009 B2
7622996 Liu Nov 2009 B2
7630230 Wong Dec 2009 B2
7633322 Zhuang et al. Dec 2009 B1
7635988 Madurawe Dec 2009 B2
7646215 Chuang et al. Jan 2010 B2
7646648 Arsovski Jan 2010 B2
7659783 Tai Feb 2010 B2
7660149 Liaw Feb 2010 B2
7663415 Chatterjee et al. Feb 2010 B2
7667678 Guttag Feb 2010 B2
7675331 Jung et al. Mar 2010 B2
7719329 Smith May 2010 B1
7719330 Lin et al. May 2010 B2
7728675 Kennedy et al. Jun 2010 B1
7737743 Gao et al. Jun 2010 B1
7746181 Moyal Jun 2010 B1
7746182 Ramaswamy et al. Jun 2010 B2
7750683 Huang et al. Jul 2010 B2
7760032 Ardehali Jul 2010 B2
7760040 Zhang et al. Jul 2010 B2
7760532 Shirley et al. Jul 2010 B2
7782655 Shau Aug 2010 B2
7812644 Cha et al. Oct 2010 B2
7830212 Lee et al. Nov 2010 B2
7839177 Soh Nov 2010 B1
7843239 Sohn et al. Nov 2010 B2
7843721 Chou Nov 2010 B1
7848725 Zolfaghari et al. Dec 2010 B2
7859919 De La Cruz, II et al. Dec 2010 B2
7876163 Hachigo Jan 2011 B2
7916554 Pawlowski Mar 2011 B2
7920409 Clark Apr 2011 B1
7920665 Lombaard Apr 2011 B1
7924599 Evans, Jr. et al. Apr 2011 B1
7940088 Sampath et al. May 2011 B1
7944256 Masuda May 2011 B2
7956695 Ding et al. Jun 2011 B1
7965108 Liu et al. Jun 2011 B2
8004920 Ito et al. Aug 2011 B2
8008956 Shin et al. Aug 2011 B1
8044724 Rao et al. Oct 2011 B2
8063707 Wang Nov 2011 B2
8087690 Kim Jan 2012 B2
8089819 Noda Jan 2012 B2
8117567 Arsovski Feb 2012 B2
8174332 Lombaard et al. May 2012 B1
8218707 Mai Jul 2012 B2
8242820 Kim Aug 2012 B2
8258831 Banai Sep 2012 B1
8284593 Russell Oct 2012 B2
8294502 Lewis et al. Oct 2012 B2
8400200 Kim et al. Mar 2013 B1
8488408 Shu et al. Jul 2013 B1
8493774 Kung Jul 2013 B2
8526256 Gosh Sep 2013 B2
8542050 Chuang et al. Sep 2013 B2
8575982 Shu et al. Nov 2013 B1
8593860 Shu et al. Nov 2013 B2
8625334 Liaw Jan 2014 B2
8643418 Ma et al. Feb 2014 B2
8692621 Snowden et al. Apr 2014 B2
8693236 Shu Apr 2014 B2
8817550 Oh Aug 2014 B1
8837207 Jou Sep 2014 B1
8885439 Shu et al. Nov 2014 B1
8971096 Jung et al. Mar 2015 B2
8995162 Sang Mar 2015 B2
9018992 Shu et al. Apr 2015 B1
9030893 Jung May 2015 B2
9053768 Shu et al. Jun 2015 B2
9059691 Lin Jun 2015 B2
9070477 Clark Jun 2015 B1
9083356 Cheng Jul 2015 B1
9093135 Khailany Jul 2015 B2
9094025 Cheng Jul 2015 B1
9135986 Shu Sep 2015 B2
9142285 Hwang et al. Sep 2015 B2
9159391 Shu et al. Oct 2015 B1
9171634 Zheng Oct 2015 B2
9177646 Arsovski Nov 2015 B2
9196324 Haig et al. Nov 2015 B2
9240229 Oh et al. Jan 2016 B1
9311971 Oh Apr 2016 B1
9318174 Chuang et al. Apr 2016 B1
9356611 Shu et al. May 2016 B1
9384822 Shu et al. Jul 2016 B2
9385032 Shu Jul 2016 B2
9396790 Chhabra Jul 2016 B1
9396795 Jeloka et al. Jul 2016 B1
9401200 Chan Jul 2016 B1
9412440 Shu et al. Aug 2016 B1
9413295 Chang Aug 2016 B1
9431079 Shu et al. Aug 2016 B1
9443575 Yabuuchi Sep 2016 B2
9484076 Shu et al. Nov 2016 B1
9494647 Chuang et al. Nov 2016 B1
9552872 Jung Jan 2017 B2
9608651 Cheng Mar 2017 B1
9613670 Chuang et al. Apr 2017 B2
9613684 Shu et al. Apr 2017 B2
9679631 Haig et al. Jun 2017 B2
9685210 Ghosh et al. Jun 2017 B1
9692429 Chang et al. Jun 2017 B1
9697890 Wang Jul 2017 B1
9722618 Cheng Aug 2017 B1
9729159 Cheng Aug 2017 B1
9789840 Farooq Oct 2017 B2
9804856 Oh et al. Oct 2017 B2
9847111 Shu et al. Dec 2017 B2
9853633 Cheng et al. Dec 2017 B1
9853634 Chang Dec 2017 B2
9859902 Chang Jan 2018 B2
9916889 Duong Mar 2018 B1
9935635 Kim et al. Apr 2018 B2
9966118 Shu et al. May 2018 B2
10065594 Fukawatase Sep 2018 B2
10153042 Ehrman Dec 2018 B2
10192592 Shu et al. Jan 2019 B2
10249312 Kim et al. Apr 2019 B2
10249362 Shu Apr 2019 B2
10388364 Ishizu et al. Aug 2019 B2
10425070 Chen et al. Sep 2019 B2
10521229 Shu et al. Dec 2019 B2
10535381 Shu et al. Jan 2020 B2
10659058 Cheng et al. May 2020 B1
10673440 Camarota Jun 2020 B1
10770133 Haig et al. Sep 2020 B1
10777262 Haig et al. Sep 2020 B1
20010052822 Kim et al. Dec 2001 A1
20020006072 Kunikiyo Jan 2002 A1
20020060938 Song May 2002 A1
20020136074 Hanzawa et al. Sep 2002 A1
20020154565 Noh et al. Oct 2002 A1
20020168935 Han Nov 2002 A1
20030016689 Hoof Jan 2003 A1
20030107913 Nii Jun 2003 A1
20030185329 Dickmann Oct 2003 A1
20040053510 Little Mar 2004 A1
20040062138 Partsch et al. Apr 2004 A1
20040090413 Yoo May 2004 A1
20040160250 Kim et al. Aug 2004 A1
20040169565 Gaggl et al. Sep 2004 A1
20040199803 Suzuki et al. Oct 2004 A1
20040240301 Rao Dec 2004 A1
20040264279 Wordeman Dec 2004 A1
20040264286 Ware et al. Dec 2004 A1
20050024912 Chen et al. Feb 2005 A1
20050026329 Kim et al. Feb 2005 A1
20050036394 Shiraishi Feb 2005 A1
20050186930 Rofougaran et al. Aug 2005 A1
20050226079 Zhu et al. Oct 2005 A1
20050226357 Yoshimura Oct 2005 A1
20050253658 Maeda et al. Nov 2005 A1
20050285862 Noda Dec 2005 A1
20060039227 Lai et al. Feb 2006 A1
20060055434 Tak et al. Mar 2006 A1
20060119443 Azam et al. Jun 2006 A1
20060139105 Maxim et al. Jun 2006 A1
20060143428 Noda Jun 2006 A1
20060248305 Fang Nov 2006 A1
20070001721 Chen et al. Jan 2007 A1
20070047283 Miyanishi Mar 2007 A1
20070058407 Dosaka Mar 2007 A1
20070109030 Park May 2007 A1
20070115739 Huang May 2007 A1
20070139997 Suzuki Jun 2007 A1
20070171713 Hunter Jul 2007 A1
20070189101 Lambrache et al. Aug 2007 A1
20070229129 Nakagawa Oct 2007 A1
20080010429 Rao Jan 2008 A1
20080049484 Sasaki Feb 2008 A1
20080068096 Feng et al. Mar 2008 A1
20080079467 Hou et al. Apr 2008 A1
20080080230 Liaw Apr 2008 A1
20080117707 Manickavasakam May 2008 A1
20080129402 Han et al. Jun 2008 A1
20080155362 Chang et al. Jun 2008 A1
20080175039 Thomas Jul 2008 A1
20080181029 Joshi et al. Jul 2008 A1
20080265957 Luong et al. Oct 2008 A1
20080273361 Dudeck et al. Nov 2008 A1
20090027947 Takeda Jan 2009 A1
20090089646 Hirose Apr 2009 A1
20090141566 Arsovski Jun 2009 A1
20090154257 Fukaishi et al. Jun 2009 A1
20090231943 Kunce et al. Sep 2009 A1
20090256642 Lesso Oct 2009 A1
20090296869 Chao et al. Dec 2009 A1
20090319871 Shirai Dec 2009 A1
20100085086 Nedovic et al. Apr 2010 A1
20100157715 Pyeon Jun 2010 A1
20100169675 Kajihara Jul 2010 A1
20100172190 Lavi Jul 2010 A1
20100177571 Shori et al. Jul 2010 A1
20100214815 Tam Aug 2010 A1
20100232202 Lu Sep 2010 A1
20100260001 Kasprak et al. Oct 2010 A1
20100271138 Thakur et al. Oct 2010 A1
20100322022 Shinozaki et al. Dec 2010 A1
20110018597 Lee et al. Jan 2011 A1
20110063898 Ong Mar 2011 A1
20110153932 Ware et al. Jun 2011 A1
20110211401 Chan et al. Sep 2011 A1
20110267914 Ish Iku Ra Nov 2011 A1
20110280307 Macinnis et al. Nov 2011 A1
20110292743 Zimmerman Dec 2011 A1
20110299353 Ito et al. Dec 2011 A1
20120049911 Ura Mar 2012 A1
20120133114 Choi May 2012 A1
20120153999 Kim Jun 2012 A1
20120242382 Tsuchiya et al. Sep 2012 A1
20120243347 Sampigethaya Sep 2012 A1
20120250440 Wu Oct 2012 A1
20120281459 Teman et al. Nov 2012 A1
20120327704 Chan Dec 2012 A1
20130039131 Haig et al. Feb 2013 A1
20130083591 Wuu Apr 2013 A1
20130170289 Grover et al. Jul 2013 A1
20140056093 Tran et al. Feb 2014 A1
20140125390 Ma May 2014 A1
20140136778 Khailany et al. May 2014 A1
20140185366 Chandwani et al. Jul 2014 A1
20140269019 Kolar Sep 2014 A1
20150003148 Iyer et al. Jan 2015 A1
20150029782 Jung Jan 2015 A1
20150063052 Manning Mar 2015 A1
20150187763 Kim et al. Jul 2015 A1
20150213858 Tao Jul 2015 A1
20150248927 Fujiwara Sep 2015 A1
20150279453 Fujiwara Oct 2015 A1
20150302917 Grover Oct 2015 A1
20150310901 Jung Oct 2015 A1
20150357028 Huang et al. Dec 2015 A1
20160005458 Shu et al. Jan 2016 A1
20160027500 Chuang et al. Jan 2016 A1
20160064068 Mojumder Mar 2016 A1
20160141023 Jung May 2016 A1
20160225436 Wang Aug 2016 A1
20160225437 Kumar Aug 2016 A1
20160247559 Atallah et al. Aug 2016 A1
20160284392 Block et al. Sep 2016 A1
20160329092 Akerib Nov 2016 A1
20170194046 Yeung, Jr. et al. Jul 2017 A1
20170345505 Noel et al. Nov 2017 A1
20180122456 Li May 2018 A1
20180123603 Chang May 2018 A1
20180157621 Shu et al. Jun 2018 A1
20180158517 Shu et al. Jun 2018 A1
20180158518 Shu et al. Jun 2018 A1
20180158519 Shu et al. Jun 2018 A1
20180158520 Shu Jun 2018 A1
20200117398 Haig et al. Apr 2020 A1
20200160905 Charles et al. May 2020 A1
20200301707 Shu et al. Sep 2020 A1
Foreign Referenced Citations (3)
Number Date Country
104752431 Jul 2015 CN
10133281 Jan 2002 DE
2005-346922 Dec 2005 JP
Non-Patent Literature Citations (1)
Entry
US 10,564,982 B1, 02/2020, Oh et al. (withdrawn)
Provisional Applications (2)
Number Date Country
62430767 Dec 2016 US
62430762 Dec 2016 US
Continuations (2)
Number Date Country
Parent 15709379 Sep 2017 US
Child 16150176 US
Parent 16150176 US
Child 16150176 US
Continuation in Parts (6)
Number Date Country
Parent 15709399 Sep 2017 US
Child 16150176 US
Parent 15709401 Sep 2017 US
Child 15709399 US
Parent 16150176 US
Child 15709399 US
Parent 15709382 Sep 2017 US
Child 16150176 US
Parent 16150176 US
Child 16150176 US
Parent 15709385 Sep 2017 US
Child 16150176 US