Logic cells for programmable gate arrays

By designing the logic unit architecture of the K-input lookup table and carry circuit, the complexity and low power efficiency of the existing FPGA logic units when implementing the 3:2 compressor is solved, and efficient accumulator and counter functions are realized. Reduces cost and complexity.

CN117157880BActive Publication Date: 2024-11-08MICROCHIP TECHNOLOGY INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202180096762.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2021-11-18
Filing Date
2021-11-20
Publication Date
2024-11-08
Estimated Expiration
2041-11-20

AI Technical Summary

Technical Problem

Existing FPGA logic units require multiple logic units or add programmable routing network connections when implementing 3:2 compressors, resulting in low power efficiency and a large number of compressors are required in binary neural networks to implement group counting functions, which adds complexity Sex and cost.

Method used

A logic unit architecture is designed, including a K-input lookup table (LUT) circuit and carry circuit. The primary output and sum output are realized through multiplexers and exclusive OR gates. The carry output can be optionally configured to reduce the number of functions. The dependency of a programming routing network implements the ability of a single logical unit to implement 3:2 and 4:2 compressors.

Benefits of technology

It improves the efficiency of LUT, improves power efficiency in implementing accumulators and counters, reduces the connection requirements for programmable routing networks, and reduces the implementation costs. It is suitable for the implementation of a variety of logic functions and compressors.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117157880B_ABST
    Figure CN117157880B_ABST
Patent Text Reader

Abstract

The present invention discloses a logic unit for a programmable logic integrated circuit device. The logic unit includes a K-input lookup table (LUT) circuit having a primary output Y and at least one additional output (F), where Y is an arbitrary function of K inputs. A carry circuit receives the output of the LUT and a carry input CI. The carry circuit generates a sum output S and a carry output CO. The carry circuit can be configured to provide S=CI and select CO from a set {0,1,F}. The carry circuit can alternatively be configured to provide S=EXOR(Y,CI) and select CO from a set {0,1,F}. The carry circuit can alternatively be configured to provide S=EXOR(Y,CI) and CO=CI in the case of Y=q, or to select CO from a set {0,1,F} in the case of Y≠q, where q is a predetermined value (e.g., such as 0 or 1).
Need to check novelty before this filing date? Find Prior Art

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Provisional Patent Application No. 63 / 191,774 filed on May 21, 2021 and U.S. Non-Provisional Patent Application No. 17 / 529,522 filed on November 18, 2021, the entireties of which are incorporated herein by reference. Background Art

[0003] The present invention relates to logic circuits in user programmable integrated circuits such as field programmable gate arrays (FPGAs). More particularly, the present invention relates to logic cell architectures for such programmable integrated circuits to provide improved support for compressors.

[0004] An FPGA is an integrated circuit consisting of logic cells connected by a programmable routing network. Typically, any logic cell output provided to the programmable routing network can be transmitted through the network to any logic cell input connected to the network. Common types of logic cells include K-input lookup tables (LUTs), multiplexers, and carry chain logic. Such a logic cell will have at least K inputs from the programmable routing network and a carry input from the carry output of the previous cell in the logic cell chain.

[0005] An important characteristic of an FPGA logic cell is how many of its inputs and outputs are provided by or to the programmable routing network. The versatility of a logic cell can increase as the number of inputs and outputs available from or to the programmable routing network increases, but the complexity and cost of the FPGA also increases. From a cost metric, a logic cell architecture that requires fewer inputs and outputs connected to the programmable routing network to achieve the desired functionality may be superior to a different architecture that requires more inputs and outputs connected to the programmable routing network to achieve the same functionality.

[0006] Another important characteristic of an FPGA logic cell is the number of inputs, K, to the LUT. The value of K can range from 2 to 10 or more, but the most common values ​​in implementation are 3, 4, and 6. Some existing FPGAs use 3-input LUTs (K=3) or 4-input LUTs (K=4), which are generally best for low-cost, low-power FPGAs. Other existing FPGAs use a splittable 6-input LUT (K=6). These FPGAs can implement a wider variety of functions, but consume more area and more power. A more general definition of K that applies to any logic cell is to define it as the maximum number such that for every function of K inputs, the logic cell can be configured to calculate it.

[0007] FPGA logic cells can be used to form compressors. Compressors are a family of logic circuits that take multiple binary values ​​as input and produce their sums represented by fewer bits. For example, the well-known full adder may also be referred to as a 3:2 compressor because it takes three one-bit inputs and produces a single-bit sum and a single-bit carry output. (The two bits corresponding to the sum and the carry output collectively indicate the count of the number of 1s in the input). In some cases, compressors may be organized in a chain similar to a carry chain, where each compressor receives one or more additional inputs from the previous compressor in the chain, and generates the same number of additional outputs to the next compressor in the chain. An example is a 4:2 compressor.

[0008] Some prior art logic cell architectures require two logic cells to implement each 3:2 compressor: one to produce each of the two outputs. Other prior art logic cell architectures can implement a 3:2 compressor with a single logic cell, but at the expense of requiring more connections to the programmable routing network. To improve power efficiency, a single logic cell architecture that can implement a 3:2 compressor that requires fewer connections to the programmable routing network would be beneficial.

[0009] Some FPGA applications require a lot of compression. For example, a binary neural network is a network that quantizes weights and activations using binary values ​​rather than full precision values. Binary neural networks often require the implementation of a function called "population counting" or "pop counting". This function represents the number of Boolean inputs in a set of M Boolean inputs as log 2 (M)-bit binary output value. Pop counters are usually implemented using many compressors. Summary of the invention

[0010] According to one aspect of the present invention, a logic cell for a programmable logic integrated circuit device is provided. The logic cell includes a K-input lookup table (LUT) circuit, the K-input LUT circuit including: i) a first (K-1)-input LUT and a second (K-1)-input LUT, both of which share the second to K-th inputs to the K-input LUT, each of the first (K-1)-input LUT and the second (K-1)-input LUT having an output; and ii) a first multiplexer, the first multiplexer having a first input coupled to the output of the first (K-1)-input LUT, a second input coupled to the output of the second (K-1)-input LUT, and a selection input coupled to the first input of the K-input LUT circuit. The first multiplexer provides a primary output Y of the logic cell, where Y is an arbitrary independent function of the K inputs. The logic unit includes a carry circuit having an X multiplexer coupled to provide an output selected from one of a constant logic reference and a primary output Y; and an XOR gate providing a sum output S, wherein the XOR gate is coupled to receive the output of the X multiplexer as a first input and to receive the carry input as a second input.

[0011] In one embodiment, a logic cell for a programmable logic integrated circuit device includes a programmable routing network, wherein a K-input lookup table circuit receives K inputs from the programmable routing network, and wherein the programmable routing network receives a primary output and a sum output.

[0012] In one embodiment, the logic cell includes a K-input lookup table circuit for providing a primary output; and a carry circuit coupled to receive one or more signals from the lookup table and a carry input, and to provide a carry output and a sum output. In one embodiment, the carry input is not received from the programmable routing network, and the carry output is not provided to the programmable routing network. In one embodiment, the carry input is received from a previous logic cell in a chain of logic cells, and the carry output is provided to a subsequent logic cell in the chain.

[0013] In one embodiment, the logic cell can be selectively configured to provide the value of the carry-in at the carry-out while also providing an arbitrary function of the K inputs at the primary output.

[0014] In one embodiment, the logic cell can be selectively configured to provide the value of the carry input at the sum output while also providing an arbitrary function of the K inputs at the primary output.

[0015] In one embodiment, the logic unit can be selectively configured to provide a single bit of an adder, where the sum appears at the sum output and the carry appears at the carry output. In one embodiment, the logic unit can be selectively configured to implement the last bit of a multi-bit adder, where the sum appears at the sum output and the carry output is selectively driven to a predetermined determined value (e.g., 0 or 1).

[0016] In various embodiments, K ≤ 4, such as K = 3 or K = 4. In another embodiment, K ≥ 4, such as 4 or 6. K is not limited to these values, and can be any non-negative integer value.

[0017] A method comprises the steps of a): providing a K-input lookup table (LUT) circuit, the K-input LUT circuit comprising: i) a first (K-1)-input LUT and a second (K-1)-input LUT, both of which share second to K-th inputs to the K-input LUT, each of the first (K-1)-input LUT and the second (K-1)-input LUT having an output; and ii) a first multiplexer having a first input coupled to the output of the first (K-1)-input LUT, a second input coupled to the output of the second (K-1)-input LUT, and a select input coupled to the first input of the K-input LUT circuit, the first multiplexer providing a primary output Y of a logic cell. The method comprises the step b) of providing a carry circuit coupled to receive a carry input (CI) and generating a carry output (CO) and a sum output (S), wherein the carry output is selectively independent of Y, wherein the carry circuit comprises: i) an X multiplexer coupled to provide an output selected from one of a constant logic reference and a primary output; and ii) an XOR gate providing the sum output, wherein the XOR gate is coupled to receive the output of the X multiplexer as a first input and the carry input as a second input. The method comprises the step c) of generating a carry output selectively independent of Y; and the step d) of propagating the carry input to a selected one of the carry output and the sum output of the logic unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present invention is illustrated by way of example and not limitation in the figures of the accompanying drawings, in which like reference numerals indicate similar elements, and in which:

[0019] Figure 1 is a block diagram of a logic unit according to one aspect of the present invention;

[0020] Figure 2 Two adders implemented using chains of logic cells are shown;

[0021] Figure 3One embodiment of a P multiplexer and an X multiplexer of a logic cell and associated configuration circuitry is shown according to an aspect of the present invention;

[0022] Figure 4 shows the configuration for Figure 1 The configuration settings of the logic units to implement the functions described;

[0023] Figure 5A According to one aspect of the present invention, Figure 1 A block diagram of a 3:2 compressor implemented by a logic unit;

[0024] Figure 5B According to one aspect of the present invention, Figure 1 Detailed view of the 3:2 compressor implemented by the logic unit;

[0025] Fig. 6A It is the functional block diagram of a 4:2 compressor;

[0026] Figure 6B According to one aspect of the present invention, Figure 1 A block diagram of a 4:2 compressor implemented by a logic unit;

[0027] Figure 7 According to one aspect of the present invention, Figure 1 A block diagram of a 2-input adder implemented by a logic unit of

[0028] Figure 8 Shows the configuration Figure 1 One embodiment of a method of a logic unit is shown. DETAILED DESCRIPTION

[0029] Those skilled in the art will recognize that the following description is illustrative only and not limiting in any way. Other aspects will readily occur to those skilled in the art.

[0030] The present invention improves the efficiency of existing LUT-based FPGA logic cells in implementing accumulators and counters while generating primary outputs that are functions of LUT inputs.

[0031] According to one aspect of the present invention, Figure 1 An illustrative LUT-based logic unit 190 that allows for improved efficiency of counters and accumulators is shown in FIG.

[0032] Figure 1The logic unit 190 includes a K-input LUT circuit formed by a first (K-1)-input LUT 124a and a second (K-1)-input LUT 124b shown within the dashed line 120. Small squares 126 and 128 represent configuration inputs for defining the functions of LUT 124a and LUT 124b, respectively. Both LUT 124a and LUT 124b share the K-1 function input In 2 、In 3 、..In K , shown at reference numerals 132, 134, and 136, respectively. The first (2-input) multiplexer 122 ("Y multiplexer") has a first data input coupled to the output F0 of the first (K-1)-input LUT 124a, a second data input coupled to the output F1 of the second (K-1)-input LUT 124b, and a function input In coupled to the LUT 120 at reference numeral 130. 1 The data output of the Y multiplexer 122 is used as the output of the K-input LUT circuit 120 and the primary output (Y) of the logic unit 190, as shown by reference numeral 48. In one specific example of the present invention, K=4.

[0033] Figure 1 The logic unit 190 also includes a carry input CI 42 , a carry output CO 44 , and a sum output S 50 . Figure 1 The carry circuit 160 in the logic unit 190 includes Figure 1 1 , a second multiplexer ("G multiplexer") is shown as a 4-input multiplexer 32 having a first data input coupled to the F0 output of the first K-1 input LUT 124a, a second data input coupled to the F1 output of the second K-1 input LUT 124b, a third data input coupled to a logic high reference, a fourth data input coupled to a logic low reference, and a data output denoted as G. The select input of the small block 34, shown as configuration circuitry coupled to the multiplexer 32, determines which of the inputs of the G multiplexer 32 will be passed to its data output.

[0034] The third multiplexer ("P multiplexer") Figure 1 1 and 1 , 3-input multiplexer 36 in carry circuit 160 is shown as logic cell 190. P multiplexer 36 has a first data input coupled to primary output Y 48 of K-input LUT circuit 120 from Y multiplexer 122, a second data input coupled to a logic high reference, and a third data input coupled to a logic low reference. Configuration circuit 38 selects which of the inputs to P multiplexer 36 will be passed to its data output, shown as P39.

[0035] The fourth multiplexer, the carry-out multiplexer ("CO multiplexer"), is located in Figure 1 1 is a 2-input multiplexer 40 in the carry circuit 160 of the logic unit 190. The CO multiplexer 40 has a first data input coupled to the data output G35 of the G multiplexer 32, a second data input coupled to the carry input CI42 of the logic unit 190, a select input coupled to the output P 39 of the P multiplexer 36, and a data output coupled to provide the carry output CO 44 of the logic unit 190.

[0036] A fifth multiplexer 86 ("X multiplexer") in the carry circuit of logic cell 190 has a first data input receiving primary output Y 48 of first multiplexer 122 and a second data input coupled to a logic low reference. Configuration circuit 88 selects which of the inputs to the X multiplexer will be passed to its data output, designated X 89.

[0037] XOR gate 46 in carry circuit 160 of logic unit 190 has a first input coupled to receive carry input CI 42, and a second input coupled to data output X 89 of X multiplexer 86. The output of the XOR gate is sum output S 50 of logic unit 190.

[0038] In some implementations of the present invention, either the F0 or F1 inputs from LUT 124a and LUT 124b and one of the logic reference inputs may be omitted from the G multiplexer or the P multiplexer (i.e., the second multiplexer 32 or the third multiplexer 36). In various implementations of the logic cell chain, the carry circuits of one or more consecutive logic cells may use a carry circuit that is functionally equivalent to the carry circuit shown in FIG. Figure 1 The carry function shown but could be a carry lookahead, carry select or similar logic operating at increased speeds.

[0039] Figure 1 The logic unit 190 has K function inputs and one carry input. Figure 1 The specific example shown includes K function inputs, In 1 130.In 2 132.IN 3 134 to In K136. To implement a logic cell with a larger value of K, additional function inputs may be added that drive both LUT 124a and LUT 124b. Primary output Y 48 of logic cell 190 presents the value of any function that is a function of the K-1 function inputs. The second output (carry out CO 44) presents the value present at carry input CI 42 or at the output from second multiplexer 32, as selected by the output of third multiplexer 36. Input In 1 、In 2 、In 3 、..In K And outputs S and Y are connected to the programmable routing network 52.

[0040] Figure 2 A chain of logic cells, such as logic cell 190 within an FPGA, is shown. The carry-out (CO) of one logic cell directly drives the carry-in (CI) of the next cell in the chain, and neither CI nor CO need to be connected to a programmable routing network. Figure 2 Also shown is a pair of adders 200 including adder 1 210 and adder 2 220 implemented using a chain of logic cells. Figure 1 The programmable routing network of can be configured to place multiple logic cells such as logic cell 190 in a carry chain. Each adder is composed of multiple serially coupled logic cells in a carry chain. Each logic cell generates a corresponding sum digit S and a carry out (CO) output signal to provide to the carry input (CI) of the next cell. For example, logic cell 212 includes a carry input CI, a sum output S, and a carry output CO coupled to the carry input CI of logic cell 214. Logic cell 216 associated with the most significant bit of adder 1 210 produces an output S representing the most significant bit of adder 1. Logic cell 216 also provides an appropriate CO signal 218 to the carry input of the logic cell corresponding to the least significant bit of adder 2 220.

[0041] The configuration of the P multiplexer, G multiplexer, and X multiplexer can be changed between the logic cell 212 associated with the least significant bit, the logic cell 216 associated with the most significant bit, and the logic cell (such as 214) associated with the middle bit of the adder 1 210. The P multiplexer, G multiplexer, and X multiplexer are configured by respective configuration circuits 38, 34, 88 to ensure that each logic cell within the adder outputs the appropriate signal. Specifically, the X multiplexer of the logic cells in the adder is set to select Y so that the sum output S=EXOR(Y,CI), where "EXOR" is a Boolean exclusive OR function applied to the value set (Y,CI). In all logic cells except the last logic cell of the adder, the P multiplexer is set to select Y so that if Y=1 the carry output CO=CI, and if Y=0 the carry output CO=G. This causes a ripple carry chain to propagate through the adder. However, in the last logic cell of an adder (such as logic cell 216 of adder 1), the P multiplexer is configured to prevent the carry from propagating to the first logic cell of the next adder (such as logic cell 222 of adder 2). In the last logic cell of the adder, the P multiplexer is set to select "0" so that CO=G, and the G multiplexer is set to select 0 or 1 (which is the appropriate constant to initialize the carry chain of the next adder).

[0042] Figure 3 (refer to Figure 1 ) shows a combined implementation 300 of a P multiplexer (36) and its configuration circuit (38) as circuit 310 and an X multiplexer (86) and its configuration circuit (88) as circuit 320. Because the P multiplexer has three inputs and the X multiplexer has two inputs, there are six possible combinations of P multiplexer and X multiplexer selections. However, only four of the six possible combinations are of interest. These combinations are shown in FIG. Figure 3 330. Therefore, the configuration circuits for the P multiplexer and the X multiplexer can be implemented with as few as two configuration bits (CFG0 and CFG1) using appropriate logic.

[0043] The P multiplexer and configuration circuit 310 includes a NOR gate 312 that receives the primary output Y and a first configuration bit CFG0. The NOR gate 312 is coupled to a NOR gate 314. The output of the NOR gate 312 is provided as one of the inputs of the NOR gate 314. The NOR gate 314 also receives a second configuration bit CFG1. The output of the NOR gate 314 is the output P 39. The Boolean function of the P multiplexer and configuration circuit 310 is P=((Y+CFG0)′+CFG1)′ or any equivalent (where the prime indicates logical negation).

[0044] X multiplexer and configuration circuit 320 includes NAND gate 322, which receives two configuration bits CFG0 and CFG1. NAND gate 322 is coupled to NAND gate 324. NAND gate 324 receives the output of NAND gate 322 and the primary output Y as inputs. NAND gate 324 is coupled to inverter gate 326, so that the output of NAND gate 324 is provided to inverter gate 326. The output of inverter gate 326 is output X 89. In the illustrated embodiment, NAND gate 324 is followed by inverter gate 326 to implement an AND function. In an alternative embodiment, gate 324 is replaced by an AND gate, so that inverter gate 326 is not required. The Boolean function of X multiplexer and configuration circuit 320 is X=Y·(CFG0·CFG1)′ or any logical equivalent (where the prime sign indicates logical negation).

[0045] Truth table 330 shows the values ​​of configuration bits CFG0 and CFG1 provided to P multiplexer and configuration circuit 310 and X multiplexer and configuration circuit 320, respectively. Figure 1 The P multiplexer 36 and the P 39 output and the X 89 output of the X multiplexer 88.

[0046] Figure 4 shows the configuration for Figure 1 to complete the configuration settings for the various modes shown in the first column of Table 400. As indicated in the first row of Table 400, Figure 1 The illustrated logic unit 190 of the present invention can be configured to provide any function of the K inputs as Y by setting the P multiplexer 36 to select 1 and the X multiplexer 86 to select Y and propagate the CI unchanged to the CO (and leaving the S output unused).

[0047] As indicated in the second row of function table 400, Figure 1 The logic cell 190 of the present invention shown can be configured to provide any function of the K input at Y and propagate CI to the S output. At the same time, the output of the G multiplexer is propagated to the CO output. This is accomplished by setting the P multiplexer 36 to select 0 and setting the X multiplexer 86 to select 0. If the next logic cell in the chain enables an adder (such as Figure 2 222 in the logic unit), then this mode is available.

[0048] As indicated in the third row of function table 400, Figure 1 The illustrated logic unit 190 of the present invention can be configured to implement one bit of an adder by setting the P multiplexer 36 to select Y and the X multiplexer 86 to select Y, with S reflecting the sum and CO as the carry out to the next more significant bit of the adder.

[0049] As indicated in the fourth row of function table 400, Figure 1 The logic unit 190 of can be configured to implement the most significant bit of the adder, where S reflects the sum. At the same time, the output of the G multiplexer is propagated to the CO output. This is accomplished by setting the P multiplexer 36 to select 0 and the X multiplexer 86 to select Y. If the next logic unit in the chain enables another adder (such as Figure 2 , then this is available.

[0050] Figure 5A Shows the use of Figure 1 An embodiment of a 3:2 compressor 500 implemented by logic unit 190 of FIG. A suitably implemented 3:2 compressor 500 receives three signals (U, V, W) from a programmable routing network, calculates two new signals EXOR(U, V, W) and MAJ(U, V, W), and sends them back to the programmable routing network. The logic units designated as "190-n" are Figure 1 1, wherein the value of "n" is used to distinguish between the instances. In this example, logic cell 190-0 is coupled to logic cell 190-1, wherein the carry output CO of logic cell 190-0 is coupled to the carry input CI of logic cell 190-1. No additional connection is required between CI or CO of logic cell 190-0 or logic cell 190-1 and the programmable routing network. This capability may, for example, allow a more efficient implementation of a compressor. In this embodiment, additional reference is made to Figure 1 , LUT 124a of logic unit 190-0 is configured to provide F0=MAJ(In 2 ,In 3 ,In 4 ),like Figure 5A "MAJ" is a majority function. A majority function produces a single output from multiple inputs. A majority function produces "1" when more than 1 / 2 of its inputs are "1", and produces "0" otherwise. LUT 124b of logic unit 190-0 is configured to provide F1=XOR(In 2 ,In 3 ,In 4 ),like Figure 5AAs indicated in Table 502 of . "XOR" refers to a Boolean exclusive OR function that produces "1" if and only if an odd number of its inputs are "1", and produces "0" otherwise. As indicated in Table 502, the G multiplexer 32 of logic unit 190-0 provides an F0 input at its corresponding data output G 35; the P multiplexer 36 of each of logic units 190-0 and 190-1 provides a logic low (i.e., 0) at its corresponding data output P39; and the X multiplexer 86 of each of logic units 190-0 and 190-1 provides a logic low (i.e., 0) at its corresponding data output X 89. The LUT 124a, LUT 124b and G multiplexer 32 of logic unit 190-1 are free to be used for other purposes, or can be configured as described in connection with logic unit 190-0 to implement the first unit of another independent 3:2 compressor, as described below with reference to Figure 5B Description.

[0051] Figure 5B shows how a single logic unit 190 can simultaneously implement the second stage of a compressor ( Figure 5A 190-1 in the logic unit 190-1 and the first stage of the second independent compressor ( Figure 5A The internal details of the logic unit 190-0 in FIG. Figure 5A 1 is configured to provide the required value of the second stage of the first compressor at the S output. LUT 124a and LUT 124b, P multiplexer 36 and G multiplexer 32 are as described above. Figure 5A 190-0 of the logic unit 190-1 is configured to provide the required values ​​of the first stage of the second compressor at the Y and CO outputs. In this way, M compressors can be implemented in M+1 logic units. Therefore, the added cost of a 3:2 compressor is only one logic unit 190.

[0052] Fig. 6A and Figure 6B Shows Figure 1 Another example of the flexibility of logic unit 190. Fig. 6AA block diagram of an embodiment of a 4:2 compressor 600 is shown, with full adders 612, 614 coupled together (i.e., the S output 50 of the full adder 612 is fed to one of the inputs of the full adder 614) as shown. The W, V, and U inputs are fed to the respective inputs of the full adder 614, and the Z and T inputs are fed to the respective inputs of the full adder 612. The N output is provided by the CO output of the full adder 614, the M output is provided by the CO output of the full adder 612, and the L output is provided by the S output of the full adder 612. Each full adder 612, 614 has three equivalent inputs and a sum and carry output. The compressor 600 is fed from inputs arbitrarily designated as inputs T, U, V, W, and Z. In implementation, multiple instances of this compressor will be used, connected in a chain similar to a carry chain, where the input T of one compressor is driven by the output N of the previous compressor in the chain. Figure 6B is a diagram showing how only two logic units 190-0 and 190-1 are needed to implement adders 612 and 614 to form Fig. 6A Table 602 shows the configuration of logic units 190-0 and 190-1 used to implement this function.

[0053] Fig. 6A The top full adder 614 is implemented as Figure 6B The logic cell 190-1 has an additional transfer function as described below so that it can be used in a programmable routing network. The lower full adder 612 is implemented in the logic cell 190-0 and includes generating the signal M, which then passes from the CI input through the logic cell 190-1 to the S output. Figure 1 , the LUT 124a of the logic unit 190-0 is configured to generate F0=In 2 The LUT 124b of the logic unit 190-0 is configured to generate F1=In 2 ! (where "!" indicates logical negation) The G, P, and X multiplexers are set to select G=F0, P=Y, and X=Y. The LUT 124a of the logic unit 190-1 is configured to generate F0=MAJ (In 2 ,In 3 ,In 4 ). The LUT 124b of the logic unit 190-1 is configured to generate F1=XOR(In 2 ,In 3 ,In 4 ), where the G, P, and X multiplexers are set to select G = F0, P = 0, and X = 0. The N output of one 4:2 compressor implemented using logic cells 190-0 and 190-1 can be used to drive the T input of another 4:2 compressor implemented using the next two logic cells in the carry chain.

[0054] Figure 7 1 is a block diagram of a 2-input, N-bit adder 700 implemented using logic cells 190 according to one aspect of the present invention. The N-bit adder 700 employs N+1 logic cells shown as 190-0 through 190-N connected as shown. The CI input of logic cell 190-0 is coupled to a constant logic low "0" level. The CO output of each logic cell 190-0 through 190-(N-1) is coupled to the CI input of the next logic cell 190 in the adder 700.

[0055] In logic units 190-0 to 190-(N-1) 1 andIn 2 The inputs are coupled to the N operand inputs U[0] to U[N-1] and V[0] to V[N-1] of adder 700, respectively. The S outputs of each logic unit 190-0 to 190-N form the sum output Z[0] to Z[N] of each of the N+1 output bits of the adder. Logic unit 190-N only needs to pass its carry input (CI) to its sum output (S) and to the programmable routing network. Advantageously, this allows its LUT to be used to implement any unrelated function of the K inputs in parallel.

[0056] Figure 7 Table 702 in FIG. 7 shows the configuration of each of the logic cells in adder 700. Specifically, the lookup table of logic cells 190 indexed from 0 to N-1 (ie, 190-0 to 190-(N-1)) is configured to calculate F0=In 2 and F1=In 2 !

[0057] Figure 8 Shown is a method for implementing Figure 4 A method for using a pattern in a function table 400 of a logic cell. The method includes step 810: providing a K-input lookup table (LUT) circuit, the K-input LUT circuit including: i) a first (K-1)-input LUT and a second (K-1)-input LUT, both of which share second to K-th inputs to the K-input LUT, each of the first (K-1)-input LUT and the second (K-1)-input LUT having an output; and ii) a first multiplexer, the first multiplexer having a first input coupled to the output of the first (K-1)-input LUT, a second input coupled to the output of the second (K-1)-input LUT, and a select input coupled to the first input of the K-input LUT circuit, the first multiplexer providing a primary output Y of a logic cell. In various applications, K≤4, such as 3 or 4. In other applications, K≥6. Although typical values ​​of K in various applications are 3, 4, or 6, K is not limited to these values. In general, K is any non-negative integer.

[0058] Figure 8 The method includes step 820 of providing a carry circuit coupled to receive a carry input CI and generating a carry output CO and a sum output S, wherein the carry output is selectively independent of Y, wherein the carry circuit includes: i) an X multiplexer coupled to provide an output Y selected from one of a constant logic reference and a primary output; and ii) an XOR gate providing the sum output S, wherein the XOR gate is coupled to receive the output of the X multiplexer as a first input and the carry input as a second input. Step 830 includes generating the carry output CO selectively independent of Y, i.e., by setting the P multiplexer 36 to 0 so that the CO output 44 reflects the data output G 35 regardless of the value of Y, or by setting the P multiplexer 36 to 1 so that the CO output 44 reflects the CI input 42 regardless of the value of Y. Carry output CO is also selectively independent of CI, i.e., by setting P multiplexer 36 to 0 so that CO output 44 reflects data output G 35 excluding the value of CI. Step 840 includes configuring the X multiplexer to propagate the carry input as a selected one of the carry output and the sum output of the logic unit.

[0059] A logic cell for a programmable logic integrated circuit device comprises i) a K-input look-up table (LUT) circuit having a primary output Y and at least one additional output F (such as Figure 1 and ii) a carry circuit coupled to receive the output of the LUT and a carry input CI of the logic unit, wherein the carry circuit generates a sum output S and a carry output CO of the logic unit, and wherein the carry circuit can be configured in various ways. Figure 1 The example provides the option of utilizing F0 or F1 as an additional output, however this is not meant to be limiting. One way is to provide S=CI and provide CO as a value selected from the set {0,1,F}. The carry circuit may alternatively be configured to provide S=EXOR(Y,CI) and CO as a value selected from the set {0,1,F}. Alternatively, the carry circuit may be configured to provide S=EXOR(Y,CI) and provide CO=CI in the case of Y=q, and provide CO as a value selected from the set {0,1,F} in the case of Y≠q, where q is a predetermined value (e.g., such as 0 and 1).

[0060] While aspects and applications of the present invention have been shown and described, it will be apparent to those skilled in the art that many more modifications than those described above are possible without departing from the inventive concepts herein.

Claims

1. A logic unit for a programmable logic integrated circuit device, the logic unit include: A K-input lookup table circuit, the K-input lookup table circuit comprising: a first (K-1)-input lookup table and a second (K-1)-input lookup table, both of which commonly share second to K-th inputs to the K-input lookup table, each of the first (K-1)-input lookup table and the second (K-1)-input lookup table having an output; and a first multiplexer having a first data input coupled to an output of the first (K-1)-input lookup table, a second data input coupled to an output of the second (K-1)-input lookup table, and a select input coupled to a first input of the K-input lookup table circuit, the data output of the first multiplexer providing a primary output of the logic cell, wherein the primary output is an arbitrary independent function of the K inputs; A carry circuit, the carry circuit comprising: a second multiplexer having a respective data input coupled to a respective output of each of the first (K-1)-input lookup table and the second (K-1)-input lookup table, coupled to receive one or more data inputs of one or more respective constant logic references, a select input, and a data output; a third multiplexer having a first data input coupled to the data output of the first multiplexer to receive the primary output, one or more data inputs coupled to receive one or more corresponding constant logic references, a select input, and a data output; a fourth multiplexer having a first data input coupled to the data output of the second multiplexer, a second data input for receiving a carry input, a select input coupled to the data output of the third multiplexer, and a data output providing a carry output of the logic unit; a fifth multiplexer having a first data input coupled to the data output of the first multiplexer, a second data input coupled to receive a constant logic reference, a select input, and a data output; and An XOR gate having a first input for receiving the carry input, a second input coupled to the data output of the fifth multiplexer, and an output providing the sum output of the logic unit.

2. The logic cell for a programmable logic integrated circuit device according to claim 1, wherein the sum output of the logic cell corresponds to one bit of an adder.

3. The logic unit for a programmable logic integrated circuit device according to claim 1, wherein the one or more data inputs of the second multiplexer coupled to receive one or more corresponding constant logic references include a third data input coupled to receive a logic high reference and a fourth data input coupled to receive a logic low reference, and Wherein the one or more data inputs of the third multiplexer coupled to receive one or more corresponding constant logic references include a second data input coupled to receive a logic high reference and a third data input coupled to receive a logic low reference. 4 . The logic cell for a programmable logic integrated circuit device according to claim 3 , wherein the carry circuit comprises a configuration circuit coupled to a select input of the third multiplexer. 5 . The logic cell for a programmable logic integrated circuit device according to claim 3 , wherein the carry circuit comprises a configuration circuit coupled to a select input of the fifth multiplexer. 6 . The logic cell for a programmable logic integrated circuit device according to claim 1 , wherein K=3. 7 . The logic cell for a programmable logic integrated circuit device according to claim 1 , wherein K=4. 8 . The logic cell for a programmable logic integrated circuit device according to claim 1 , wherein K=6.

9. The logic unit for a programmable logic integrated circuit device according to claim 1, wherein the logic unit include: A programmable routing network, wherein the K-input lookup table circuit receives the K inputs from the programmable routing network, wherein the programmable routing network receives the primary output of the logic cell and the sum output of the logic cell.

10. The logic cell for a programmable logic integrated circuit device according to claim 9, wherein the carry input is not received from the programmable routing network and wherein the carry output of the logic cell is not provided to the programmable routing network.

11. A logic unit, wherein the logic unit include: A K-input lookup table circuit, the K-input lookup table circuit comprising: a first (K-1)-input lookup table and a second (K-1)-input lookup table, both of which commonly share second to K-th inputs to the K-input lookup table, each of the first (K-1)-input lookup table and the second (K-1)-input lookup table having a corresponding output; a first multiplexer having a first data input coupled to the output of the first (K-1)-input lookup table, a second data input coupled to the output of the second (K-1)-input lookup table, and a select input coupled to the first input of the K-input lookup table circuit, the first multiplexer providing a primary output of the logic cell; and a carry circuit coupled to receive a carry input and to generate a carry output and a sum output, wherein the carry output is selectively independent of the primary output, wherein the carry circuit comprises: a second multiplexer coupled to receive a respective output of each of the first (K-1) input lookup table and the second (K-1) input lookup table; a third multiplexer coupled to receive the primary output of the logic unit; a fourth multiplexer coupled to provide one of the output of the second multiplexer and the carry input as the carry output according to the output of the third multiplexer; a fifth multiplexer coupled to provide an output selected from one of a constant logic reference and the primary output of the logic cell; and An XOR gate is provided for the sum output, wherein the XOR gate is coupled to receive the output of the fifth multiplexer as a first input and the carry input as a second input.

12. The logic cell of claim 11 , wherein the output of the third multiplexer is derived from a Boolean function of the primary output of the logic cell and two configuration bits CFG0 and CFG1, wherein P=((Y+CFG0)′+CFG1)′, where Y is the primary output of the logic unit and P is the output of the third multiplexer.

13. The logic cell of claim 11 , wherein the output of the fifth multiplexer is derived from a Boolean function of the primary output of the logic cell and two configuration bits CFG0 and CFG1, wherein X=Y·(CFG0·CFG1)′, where Y is the primary output of the logic unit and X is the output of the fifth multiplexer.

14. The logic cell of claim 11 , the logic cell comprising a programmable routing network, wherein the K-input lookup table circuit receives the K inputs from the programmable routing network, wherein the programmable routing network receives the primary output and the sum output of the logic cell, wherein the carry output and the carry input are not connected to the programmable routing network.

15. The logic cell of claim 11, wherein the logic cell is selectively configurable to provide a value of the carry-in as the carry-out.

16. The logic cell of claim 11, wherein the logic cell is selectively configurable to provide a value of the carry input as the sum output.

17. The logic cell of claim 11, wherein the logic cell is selectively configurable to provide the primary output of the logic cell as an arbitrary function of the K inputs. The logic cell of claim 11 , wherein K≤6.

19. A method, the method comprising: The following steps are involved: A K-input lookup table circuit is provided, the K-input lookup table circuit comprising: a first (K-1)-input lookup table and a second (K-1)-input lookup table, both of which commonly share second to K-th inputs to the K-input lookup table, each of the first (K-1)-input lookup table and the second (K-1)-input lookup table having an output; and a first multiplexer having a first data input coupled to the output of the first (K-1)-input lookup table, a second data input coupled to the output of the second (K-1)-input lookup table, and a select input coupled to the first input of the K-input lookup table circuit, the first multiplexer providing a primary output of a logic cell; and A carry circuit is provided, the carry circuit being coupled to receive a carry input and to generate a carry output and a sum output, wherein the carry circuit comprises: a second multiplexer coupled to receive a respective output of each of the first (K-1) input lookup table and the second (K-1) input lookup table; a third multiplexer coupled to receive the primary output of the logic unit; a fourth multiplexer coupled to provide one of the output of the second multiplexer and the carry input as the carry output according to the output of the third multiplexer; a fifth multiplexer coupled to provide an output selected from one of a constant logic reference and the primary output of the logic cell; and an XOR gate providing the sum output, wherein the XOR gate is coupled to receive the output of the fifth multiplexer as a first input and to receive the carry input as a second input; generating a carry output selectively independent of the primary output of the logic unit; and The carry input is propagated to a selected one of a carry output and a sum output of the logic unit.