Research map / Learning theory and algorithms / Combinatorial and search optimization
A Parallel CPU-GPU Framework for Batching Heuristic Operations in Depth-First Heuristic Search
Ehsan Futuhi, Nathan R. Sturtevant
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase batch size 1 -> 800 (ablation) | SingleGPU Batch IDA* · 4x4 STP | avg time (s): 16.80 -> 0.54 | Improved |
| increase batch size 80 -> 8,000 (ablation) | SingleGPU Batch IDA* · 3x3 Rubik's Cube | avg time (s): 58.07 -> 3.46 | Improved |
| increase batch size 800 -> 8,000 (ablation) | SingleGPU Batch IDA* · 4x4 STP | avg time (s): 0.54 -> 0.71 | Hurt |
| add second GPU (ablation) | Batch IDA* batch 8000 · 3x3 Rubik's Cube | avg time (s): 3.46 -> 2.51 | Improved |
| increase batch size 10 -> 2000 (ablation) | Batch BTS · Rubik's Cube length 9, regression heuristic | avg time (s): 169.73 -> 3.01 | Improved |
| use BTS instead of IDA* with regression heuristic | Batch search · Rubik's Cube length 8, batch 2000 | avg time (s): 144.91 -> 0.83 | Improved |
| increase GPUs 1 -> 4 (ablation) | Batch IDA* · 3x3 Rubik's Cube (GPU count) | avg time (s): 6.11 -> 2.56 | Improved |
| use batch IDA* with neural heuristic instead of AIDA* | Batch IDA* · 4x4 STP learned heuristic | avg time (s): 1.21 -> 11.78 | Hurt |
| use 2GPU Batch IDA* instead of Batch A* | Batch IDA* · 3x3 Rubik's Cube learned heuristic | avg time (s): 42.07 -> 5.11 | Improved |
| use Batch A* instead of 2GPU Batch IDA* | Batch A* · 4x4 STP learned heuristic | avg time (s): 11.78 -> 2.26 | Improved |
Closest papers by meaning
- DOGE-Train: Discrete Optimization on GPU with End-to-End Training (AAAI 2024 · similarity 0.84)
- Parallel Greedy Best-First Search with a Bound on Expansions Relative to Sequential Search (AAAI 2025 · similarity 0.83)
- B$^3$-PWL: GPU-Batched Branch-and-Bound for Piecewise-Linear Optimization with SOS2 Constraints (NeurIPS 2026 · similarity 0.83)
- Integrated Hardware Architecture and Device Placement Search (ICML 2024 · similarity 0.82)
- Rectangle Search: An Anytime Beam Search (AAAI 2024 · similarity 0.82)
- FlashRNN: I/O-Aware Optimization of Traditional RNNs on modern hardware (ICLR 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
