摘要:
A programmable system is disclosed having multiple configurable floating point units (“FPU”) that are coupled to multiple programmable logic and routing blocks and multiple memories. Each floating point unit has static configuration blocks and dynamic configuration blocks, where the dynamic configuration blocks can be reconfigured to perform a different floating point unit function. A floating point unit includes a pre-normalization for shifting an exponent calculation as well as shifting and aligning a mantissa, and a post-normalization for normalizing and rounding a received input. The post-normalization receives an input Z and realigns the input, normalizes the input and rounds the input Z.
摘要:
A programmable system is disclosed having multiple configurable floating point units (“FPU”) that are coupled to multiple programmable logic and routing blocks and multiple memories. Each floating point unit has static configuration blocks and dynamic configuration blocks, where the dynamic configuration blocks can be reconfigured to perform a different floating point unit function. A floating point unit includes a pre-normalization for shifting an exponent calculation as well as shifting and aligning a mantissa, and a post-normalization for normalizing and rounding a received input. The post-normalization receives an input Z and realigns the input, normalizes the input and rounds the input Z.
摘要:
A programmable system is disclosed having multiple configurable floating point units (“FPU”) that are coupled to multiple programmable logic and routing blocks and multiple memories. Each floating point unit has static configuration blocks and dynamic configuration blocks, where the dynamic configuration blocks can be reconfigured to perform a different floating point unit function. A floating point unit includes a pre-normalization for shifting an exponent calculation as well as shifting and aligning a mantissa, and a post-normalization for normalizing and rounding a received input. The post-normalization receives an input Z and realigns the input, normalizes the input and rounds the input Z.
摘要:
A programmable logic device has a plurality of levels of programmable logic modules with fixed interconnections. The outputs of a level connect to inputs of the next level of programmable logic modules. The first level is fed from a bank of memory elements and the inputs to this bank of memory elements are derived from the last level. Crossbar switches are optionally inserted between a carefully chosen pairs of levels.
摘要:
A function generator is described that can be configured as a combinational logic, sequential logic, or routing cell. The function generator couples to a plurality of selector blocks that select a wire from a plurality of inputs. The selected wires are inputs to a function generator. The function generator, when configured as a combinational logic cell, can generate any function of its inputs. The function generator, when configured as a sequential logic cell, behaves as a register, where any of the inputs can be directed to input data, clear, clock enable, or reset signals. The register is configurable to a falling or rising edge flip-flop or a positive or negative level sensitive latch. As a routing element, logic cells selects one of its inputs. The output of the programmable cell can fan out to one or more inputs of another integrated cell.
摘要:
A programmable logic array for use in a field programmable application specific integrated circuit (ASIC) implementation is provided. The programmable logic array includes programmable logic blocks, and programmable logic interfaces. The programmable logic interfaces couple the programmable logic blocks to external interfaces of the field programmable ASIC, and enable the programmable logic array to be inserted into the field programmable ASIC as a hard macro block.
摘要:
A memory is disclosed that can be utilized with a field programmable gate array. In some embodiments, the memory can include a memory array comprising a plurality of memory banks, each memory bank including at least one memory block, each of the at least one memory block including an array of memory cells; an address decoder coupled to each of the at least one memory block, the address decoder including a comparator coupled to receive an input address and a block address and provide a compare bit that indicates when a portion of the input address matches the block address, and an OR gate coupled to receive the compare bit and a wildcard bit, the OR gate providing an enable to the memory block when either the compare bit or the wildcard bit is asserted; and a logic unit that receives a mode value and the input address and provides the wildcard bit to each of the address decoders. Data can be simultaneously written into the memory array in patterns in accordance with the mode value. For example, in some embodiments the mode value indicates one of four patterns, a normal pattern, a block checkerboard pattern, a bank checkerboard pattern, and an all banks pattern.
摘要:
A system and method are disclosed for providing highly parallel, FFT calculations in a circuit including a plurality of RADIX-2 elements. Partitioned RAM resources allow RADIXes at all stages to have optimal bandwidth memory access. Preferably more memory is made available for early RADIX stages and a “critical” stage. RADIXes within stages beyond the critical stage preferably each need only a single RAM partition, and can therefore simultaneously operate without fighting for memory resources. In a preferred configuration having P RAM partitions and P RADIX stages, the critical stage is stage number log2P, and until the critical stage, only P/2 RADIX elements can simultaneously operate within each stage. After the critical stage, all RADIXes within each stage can simultaneously operate.
摘要:
A programmable logic device (PLD) is provided that supports multi-gigabit transceivers (MGTs). The PLD includes one or more pairs of shared clock pads for receiving one or more high-quality differential clock signals. Dedicated clock traces couple each pair of shared clock pads to one or more MGTs on the PLD. Each MGT includes a clock multiplexer circuit, which allows one of the high-quality differential clock signals to be routed as a reference clock signal for the MGT. The clock multiplexer circuits are designed such that no significant jitter is added to the high-quality clock signals. The clock multiplexer circuits can also route general-purpose clock signals received by the PLD as lower quality reference clock signals for the MGTs. The reference clock signal routed by the clock multiplexer circuit can be stepped down to provide a reference clock for a physical coding sublayer of the MGT.
摘要:
Disclosed are methods and structures for preparing data for transmission over a network. In an embodiment consistent with the OSI network model, transmit and receive CRC generators are moved from the link layer to the physical layer, which frees up valuable programmable logic resources when a programmable logic device is employed to perform the functions of the link layer. The CRC generators of the physical layer comply with a plurality of network communication standards.