A pipelined RV32I processor implemented in SystemVerilog as UIUC ECE411 Computer Organization and Design FA20 final project.

Features

• 5-stage pipeline with data forwarding and hazard detection
• Branch prediction
  • Local Branch History Table
  • Global Branch History Table
  • Tournament Branch Predictor
• Caches
  • One-cycle L1 I-Cache and D-Cache
  • Two-cycle L2 I-Cache and D-Cache using BRAM
  • Parameterized number of ways (2, 4 or 8) with pseudo LRU
  • Parameterized set number
  • Single unified parameterized module with all the configurable parameters above
• Prefetch
  • Passing signal down in normal read state
  • Fetch next line immediately after the current read (in prefetch state)
  • Block new read during the prefetch
• Performance counters

Authors

Team 0_warning_0_error:

• Zikai Liu (zikail2@illinois.edu)
• Kerui Zhu (keruiz2@illinois.edu)
• Kaiwen Cao (kaiwenc2@illinois.edu)

Organization

Miscellaneous

• See our presentation for an overview of the performance of our processor.
• Original ECE411 MP4 README.