Programmable logic systems and methods employing configurable floating point units
View Patent ↗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.
1. A programmable integrated circuit comprising:
a configurable floating point unit for performing floating point calculations;
a logic and routing block having a plurality of dedicated logic cells, each dedicated logic cell having a first logic and routing cell and a second logic and routing cell;
one or more first dedicated lines connecting from the first logic and routing cell in the first dedicated logic cell of the logic and routing block to a first plurality of inputs in the configurable floating point unit; and
one or more second dedicated lines connecting from a first plurality of outputs in the configurable floating point unit to the first logic and routing cell in a first dedicated logic cell of the logic and routing block.
2. The programmable system of claim 1 , further comprising a memory; and
one or more third dedicated lines connecting from the memory to a second plurality of inputs in the configurable floating point unit.
3. The programmable system of claim 2 , further comprising one or more fourth dedicated lines connecting from a second plurality outputs in the configurable floating point unit to the memory.
4. The programmable integrated circuit of claim 1 , wherein the floating point unit comprises:
a pre-normalization block for receiving a first input including a sign bit, an exponent field and a mantissa field from a dedicated logic cell, the pre-normalization block shifting the exponent field and shifting the mantissa in generating an output F A ; and
a post-normalization block coupled to the pre-normalization block for receiving the output Z from the pre-normalization block, the post-normalization block normalizing and rounding the received input Z.
5. The programmable integrated circuit of claim 4 , wherein the pre-normalization block receives a second input comprising a sign bit, an exponent field and a mantissa field from the dedicated logic cell, the pre-normalization block shifting the exponent field and shifting the mantissa in generating an output F B .
6. The programmable integrated circuit of claim 5 , further comprising a set of multiplexers coupled to the post-normalization block for selecting a load signal, a clear signal, and an enable signal.
7. A programmable system for direct memory connections, comprising:
a floating point unit for performing floating point calculations;
a logic and routing block having a plurality of dedicated logic cells, each dedicated logic cell having a first logic and routing cell and a second logic and routing cell; and
a first memory coupled to the floating point unit, the first memory having one or more first dedicated lines connecting to at least one input of the floating point unit, one or more second dedicated lines connecting to at least one output of the floating point unit, one or more third dedicated lines connecting from at least one input of the first memory to the first logic and routing cell, and one or more fourth dedicated lines connecting from the first logic and routing cell to at least one output of the first memory.
8. The programmable system of claim 7 further comprising a second memory that has one or more dedicated connections to the floating point unit.
9. The programmable system of claim 8 further comprising a second column of dedicated logic cells coupled to the floating point unit.