1. CS203 – Advanced Computer Architecture
Computer Architecture Simulators

2. Why use simulators? Designing and fabricating chips are expensive
Would take years to test new microarchitecture design
Abstract performance/queuing models are simplistic
Require a middle-ground
Fast, accurate, configurable

3. Why use simulators?
Leverage faster, more flexible S/W development cycles
Permits more design space exploration
Facilitates validation before H/W becomes available
Level of abstraction can be throttled to design task
Possible to increase/improve system instrumentation
Adapted from SimpleScalar Tutorial

4. What is an architectural simulator?
Tool that reproduces the behavior of a computing device
Adapted from SimpleScalar Tutorial

5. Taxonomy of Simulator Tools
Architectural Simulators
Functional
Trace-driven
Exec-driven
Interpreters
Direct Execution
Performance
Inst. Schedulers
Cycle timers
Adapted from SimpleScalar Tutorial

6. Functional vs Performance Simulators
Functional simulators implement the architecture
The architecture is what the programmer’s see
Performance simulators implement the microarchitecture
Model system internals (microarchitecture)
Often concerned with time
Adapted from SimpleScalar Tutorial

7. Execution vs Trace-driven Simulation
Trace-based simulation:
Reads a “trace” of instructions saved from previous execution
Easiest to implement, no funcitonal component needed
No feedback into trace
Execution-driven simulation:
Simulator “runs” the program, generating stream dynamically
More difficult to implement, many advantages
Direct execution: instrumented program runs on host
Adapted from SimpleScalar Tutorial

8. Instruction schedulers vs cycle timers
Constraint-based instruction schedulers
Simulator schedules instructions based on resource availability
Instructions processed one at a time, in order
Simpler to implement/modify, generally less detailed
Cycle-timer simulators
Simulator tracks microarch. State each cycle
Many instructions in various stages at any time
Simulator state == microarch. State
Good for detailed microarchitecture simulation

9. Functional vs performance simulators
Functional simulators implement the architecture
Perform the actual execution
Implement what programmers see
Performance (or timing) simulators implement the microarchitecture
Model system resources/internals
Measure time
Implement what the programmers do not see
Adapted from SimpleScalar Tutorial

10. Application-only vs Full-system
Adapted from Multi2Sim Tutorial

11. The Zen of Simulator Design
Design goals will drive which aspects are optimized
Adapted from SimpleScalar Tutorial

12. Example Simulators - CPU
SimpleScalar (Wisconsin/Michigan)
SESC (UIUC)
ESESC (UCSC)
gem5 (Michigan, Wisconsin, ARM, etc.)
Merger of M5 (Michigan) + GEMS (Wisconsin)
Simics (Wind River Systems)
Sniper (Ghent)
zsim (Stanford/MIT)
MARSSx86 (Binghamton)

13. Example Simulators Temperature Memory Disk Power Reliability
Hotspot (Virginia)
Memory
DramSim2 (Maryland)
CACTI (HP)
NVSim (PSU)
Disk
DiskSim (CMU)
FlashSim (PSU)
Power
McPAT (HP)
Wattch (Harvard)
Reliability
VARIUS (UIUC)

14. Example Simulators Graphical Processing Units (GPU) GPU Energy
GPGPU-Sim (UBC)
Multi2Sim (Northeastern)
MacSim (GaTech)
GPU Energy
GPUWattch (UT-Austin/Wisconsin/UBC)

15. Example Simulators Heterogeneous Architectures Datacenters Network
Multi2Sim (Northeastern) - CPU/GPU
CPU-GPU Simulator (PSU) – CPU/GPU
gem5-gpu (Wisconsin) – CPU/GPU
Datacenters
BigHouse (Michigan)
Network
DARSIM (MIT)
NoC-SIM (TAMU)