# Gem5 Architecture

## Basic Info
- Time: [time=20221218]
- Members: [name=宏諺], [name=Cherie], [name=峻元], [name=岑晏]

## Discussion Conclusion
- [name=宏諺]
- [name=Cherie]
- [name=峻元]
- [name=岑晏]

## Topic
### Matrix Extension Programming Model
- Finished Matrix Instruction Programming Model
- Commit: https://github.com/nycu-caslab/gem5/commit/16e9cb8c62490b533df4487810f98daa9ccf17ab
- Implementing the Matrix Instruction
- Config Instruction
- Load/Store Instruction
- Matrix Computation Instruction
- Workload Establishment
- Matrix Multiplication
- Assembly Macro

### Vector Performance Profiling
- [ ] Utilization
- [ ] Function-Unit Utilization
- [ ] Load-to-use latency
- [ ] Vector Load/Store latency
- [ ] Stall (FU Idle)
- [ ] Depedency Instrution
- [ ] Load/Store
- [ ] Performance Counter
- [ ] Cache Optimization
- [ ] Context-awared Cache
- [ ] Cache Group

# lmul = 4
vle8 v1, (0x80000000)
vmult z5, v1, v0
vmult z6, v2, v0
vmult z7, v3, v0
vmult z8, v4, v0
vle8 v1, (0x80000064)
vmult v5, v1, v0
vmult v6, v2, v0
vmult v7, v3, v0
vmult v8, v4, v0

- [ ] Computing Unit Optimization