dse-loop skill

DSE Loop: Autonomous Design Space Exploration

Autonomously explore a design space: run → analyze → pick next parameters → repeat, until the objective is met or timeout is reached. Designed for computer architecture and EDA problems.

Context: $ARGUMENTS

Safety Rules — READ FIRST

NEVER do any of the following:

• sudo anything
• rm -rf, rm -r, or any recursive deletion
• rm any file you did not create in this session
• Overwrite existing source files without reading them first
• git push, git reset --hard, or any destructive git operation
• Kill processes you did not start

If a step requires any of the above, STOP and report to the user.

Constants (override via $ARGUMENTS)

Override inline: /dse-loop "task desc — timeout: 4h, max_iterations: 100, patience: 15"

Typical Use Cases

Workflow

Phase 0: Parse Task & Setup

1. Parse $ARGUMENTS to extract:

• Program: what to run (command, script, or Makefile target)
• Parameter space: which knobs to tune and their ranges/options (may be incomplete — see step 2)
• Objective metric: what to optimize (and how to extract it from output)
• Constraints: hard limits that must not be violated (e.g., timing must close)
• Timeout: wall-clock budget
• Success criteria: when is the result "good enough" to stop early?

1. Infer missing parameter ranges — If the user provides parameter names but NOT ranges/options, you MUST infer them before exploring:

a. Read the source code — search for the parameter names in the codebase:

• Look for argparse/click definitions, config files, Makefile variables, module parameters, #define, parameter (SystemVerilog), localparam, etc.
• Extract defaults, types, and any comments hinting at valid values

b. Apply domain knowledge to set reasonable ranges:

c. Start conservative — begin with 3-5 values per parameter. Expand range later if the best result is at a boundary.

d. Log inferred ranges — write the inferred parameter space to dseresults/inferredparams.md so the user can review:

# Inferred Parameter Space

      | Parameter | Source | Default | Inferred Range | Reasoning |
      |-----------|--------|---------|---------------|-----------|
      | CACHE_SIZE | config.py:42 | 32768 | [8192, 16384, 32768, 65536, 131072] | powers of 2, ±2x from default |
      | ASSOC | config.py:43 | 4 | [1, 2, 4, 8] | standard associativities |
      | BMC_DEPTH | run_bmc.py:15 | 10 | [5, 10, 20, 50] | geometric, common BMC depths |

e. Boundary expansion — during the search, if the best result is at the min or max of a range, automatically extend that range by one step in that direction (but log the extension).

1. Read the project to understand:

• How to run the program
• Where results are produced (stdout, log files, reports)
• How to parse the objective metric from output
• Current/baseline configuration (if any)

1. Create working directory: dse_results/ in project root

• dseresults/dselog.csv — one row per design point
• dseresults/DSEREPORT.md — final report
• dseresults/DSESTATE.json — state for recovery
• dseresults/inferredparams.md — inferred parameter space (if ranges were not provided)
• dse_results/configs/ — config files for each run
• dse_results/outputs/ — raw output for each run

1. Write a parameter extraction script (dseresults/parseresult.py or similar) that takes a run's output and returns the objective metric as a number. Test it on a baseline run first.

1. Run baseline (iteration 0): run the program with default/current parameters. Record the baseline metric. This is the point to beat.

Phase 1: Initial Exploration

Goal: Quickly survey the space to understand which parameters matter most.

Strategy: Latin Hypercube Sampling or structured sweep of key parameters.

1. Pick 5-10 diverse design points that span the parameter ranges
2. Run them (in parallel if independent, via background processes or sequential)
3. Record all results in dse_log.csv:

iteration,param1,param2,...,metric,constraint_met,timestamp,notes
   0,default,default,...,baseline_val,yes,2026-03-13T10:00:00,baseline
   1,val1a,val2a,...,result1,yes,2026-03-13T10:05:00,initial sweep
   ...

1. Analyze: which parameters have the most impact on the objective?
2. Narrow the search to the most sensitive parameters

Phase 2: Directed Search

Goal: Converge toward the optimum by making informed choices.

Strategy: Adaptive — pick the approach that fits the problem:

• Few parameters (≤3): Fine-grained grid search around the best region from Phase 1
• Many parameters (>3): Coordinate descent — optimize one parameter at a time, holding others at current best
• Binary/categorical params: Enumerate promising combinations
• Continuous params: Binary search or golden section between best neighbors
• Multi-objective: Track Pareto frontier, explore along the front

For each iteration:

1. Select next design point based on results so far:

• Look at the trend: which direction improves the metric?
• Avoid re-running configurations already evaluated
• Balance exploration (untested regions) vs exploitation (near current best)

1. Modify parameters: edit config file, command-line args, or source constants

1. Run the program: execute and capture output

1. Parse results: extract the objective metric and check constraints

1. Log to dselog.csv**: append the new row

1. Check stopping conditions:

• Timeout reached? → stop
• Max iterations reached? → stop
• Patience exhausted (no improvement in N iterations)? → stop
• Success criteria met (metric is "good enough")? → stop
• Constraint violation pattern detected? → adjust search bounds

1. Update DSESTATE.json**:

{
     "iteration": 15,
     "status": "in_progress",
     "best_metric": 1.23,
     "best_params": {"cache_size": 32768, "assoc": 4, "pipeline_width": 2},
     "total_iterations": 15,
     "start_time": "2026-03-13T10:00:00",
     "timeout": "2h",
     "patience_counter": 3
   }

1. Decide next step → back to step 1

Phase 3: Refinement (if time allows)

If the search converged and there's still time budget:

1. Local perturbation: try ±1 step on each parameter from the best point
2. Sensitivity analysis: which parameters can be relaxed without hurting the metric?
3. Constraint boundary: if a constraint is nearly binding, explore near-feasible points

Phase 4: Report

Write dseresults/DSEREPORT.md:

# Design Space Exploration Report

**Task**: [description]
**Date**: [start] → [end]
**Total iterations**: N
**Wall-clock time**: X hours Y minutes

## Objective
- **Metric**: [what was optimized]
- **Direction**: minimize / maximize
- **Baseline**: [value]
- **Best found**: [value] ([improvement]% better than baseline)

## Best Configuration
| Parameter | Baseline | Best |
|-----------|----------|------|
| param1    | default  | best_val |
| param2    | default  | best_val |
| ...       | ...      | ... |

## Search Trajectory
| Iteration | param1 | param2 | ... | Metric | Notes |
|-----------|--------|--------|-----|--------|-------|
| 0 (baseline) | ... | ... | ... | ... | baseline |
| 1 | ... | ... | ... | ... | initial sweep |
| ... | ... | ... | ... | ... | ... |
| N (best) | ... | ... | ... | ... | ★ best |

## Parameter Sensitivity
- **param1**: [high/medium/low impact] — [brief explanation]
- **param2**: [high/medium/low impact] — [brief explanation]

## Pareto Frontier (if multi-objective)
[Table or description of non-dominated points]

## Stopping Reason
[timeout / max_iterations / patience / success_criteria_met]

## Recommendations
- [actionable insights from the explo

Also generate a summary plot if matplotlib is available:

• Convergence curve (metric vs iteration)
• Parameter sensitivity bar chart
• Pareto frontier scatter (if multi-objective)

State Recovery

If the context window compacts mid-run, the loop recovers from DSESTATE.json + dselog.csv: