The PIPE is a double register plus a small state machine that enables a fully synchronous stall-able pipeline to be built. A parameter defines the ratio of the input width to the output width. For example, if the input width is 32-bits and the Reduction Factor is 4 the the output width is 8-bits. The ORDY ready signal will indicate when at least a whole output word is available. The interface is fully compatible with a Standard FIFO interface and they may be mixed and matched. A “Stall” may be generated by gating the POP and ORDY signals to arrest the normal flow of data down the pipeline and allows any given stage to take multiple cycles when necessary. In doing so the IRDY signal is removed in the following cycle and the data that was to replace the data in the primary register is saved into the secondary register. When the “Stall” is removed the the secondary register contents are copied to the primary register and the IRDY is re-asserted for the next clock edge.
Stallable pipeline stage with width contraction
Overview
Key Features
- Fully Synthesizable RTL - Verilog
- Parameterized Reduction Factor
- Static Timing Analysis compatible
- Double Register for synchronous pipeline
- Configurable width
- Standard FIFO handshake interface
Block Diagram

Applications
- Fully Synchronous pipelines
- Datapath Width Reduction
Deliverables
- verilog RTL and Testbench
Technical Specifications
Short description
Stallable pipeline stage with width contraction
Vendor
Vendor Name
Maturity
Multiple Uses
Availability
Now
Related IPs
- Stallable pipeline stage with width expansion
- Stallable pipeline stage with protocol for multiway pipeline fork and join capability
- Open-source compact microcontroller core with a 4-stage in-order pipeline for deeply embedded applications
- Efficient microcontroller core with a 5-stage in-order pipeline, privilege modes, an MPU, L1 and L2 caches
- Efficient microcontroller core with a 5-stage in-order pipeline, privilege modes, an FPU, an MPU, L1 and L2 caches
- Efficient Linux-capable application core with a 9-stage in-order pipeline, an MMU, L1 and L2 caches, and cache coherency