Skip to content

ROADMAP.md

Latest commit

 

History

History
62 lines (50 loc) · 3.3 KB

ROADMAP.md

File metadata and controls

62 lines (50 loc) · 3.3 KB
Raw

Nexus Roadmap

Current State

  • Current phase: Phase 11 completed
  • Final validation status: verified on 2026-04-22
  • Repository state: bounded educational/research-grade platform with a passing 56/56 low-memory test run

Phase Sequence

  1. Phase 1: repository bootstrap and build scaffolding
  2. Phase 2: language spec, frontend, AST, semantics
  3. Phase 3: IR, CFG, dominators, baseline analysis
  4. Phase 4: MIPS backend and functional execution
  5. Phase 5: single-cycle and multi-cycle organization models
  6. Phase 6: 5-stage pipeline, hazards, forwarding, branch handling
  7. Phase 7: memory/cache/I/O/DMA/interrupts and metrics
  8. Phase 8: advanced architecture sandbox and Amdahl evaluation
  9. Phase 9: advanced compiler prototypes and toolchain comparison
  10. Phase 10: multicore/parallel simulation, coherence/consistency-lite, synchronization, OpenMP/MPI/SIMD integration
  11. Phase 11: HDL modules/testbenches, optional GPU demos, literature completion, and final validation

Final Gap-Closure Status

Topic Status Exit evidence
FPU-lite experimentally implemented src/common/src/fpu_lite.cpp, nexus_fpu_lite_test
L1+L2 hierarchy experimentally implemented src/sim/memory/, nexus_memory_cache_test, nexus_pipeline_memory_system_test, CLI coverage
Flex/Bison LR parser path experimentally implemented src/compiler/experimental_parallel_parsing/, nexus_experimental_parse_test
scoreboard scheduler experimentally implemented src/sim/advanced/src/model.cpp, nexus_advanced_model_test
affine/locality compiler slice experimentally implemented src/compiler/analysis/src/affine_analysis.cpp, src/compiler/passes/src/affine_stripmine.cpp, focused tests
stronger GPU demo experimentally implemented optional CUDA demo source plus validated clean skip path
HDL CPU slice experimentally implemented src/hdl/cpu_slice/, nexus_hdl_cpu_slice, nexus_hdl_all
core compiler/simulator/CLI baseline fully implemented existing end-to-end stack and full-suite coverage
broader theory-only material such as Tomasulo, generalized ambiguity-supporting parsing, and industrial polyhedral scheduling documented with worked examples docs/ and docs/reports/
full synthesizable CPU, mandatory GPU execution, industrial out-of-order core, industrial optimizer depth still outside bounded scope explicit final-scope limits

Final Verdict Frame

The roadmap is complete in bounded educational/research-grade form. The repository now covers the course domains through a deliberate mix of fully implemented core flows, experimentally implemented extensions, and documented worked examples, while still naming the topics that remain outside final bounded scope.

Final Submission Set

  • README.md
  • STATUS.md
  • ROADMAP.md
  • docs/reports/final_nexus_system_paper.md
  • docs/reports/validation_report.md
  • docs/reports/course_mapping.md
  • docs/reports/full_syllabus_checklist.md
  • docs/reports/repository_wide_truth_audit.md

Post-Project Future Work

  • full synthesizable HDL CPU integration
  • stronger heterogeneous and device-backed GPU evaluation
  • Tomasulo/reorder-buffer out-of-order execution experiments
  • industrial SSA, register allocation, and deeper optimization infrastructure
  • larger manycore and richer coherence studies