GPU-CFR: 80x Faster Counterfactual Regret Minimization by Compiling the Game to Static Dataflow and CUDA Graph Replay
GPU-CFR compiles counterfactual regret minimization into static dataflow with CUDA Graph Replay, achieving 29.8-80.4x speedups over prior GPU solvers.
The paper presents a compiler and runtime that turns any fixed game's CFR iteration into a static dataflow graph of flat arrays and precomputed indices, cutting framework operations by up to 18.1x. Because shapes and buffer addresses never change, CUDA Graph Replay records the iteration once and replays it with a single launch. On one A100 across an eight-game suite, GPU-CFR runs 29.8-80.4x faster than the fastest prior GPU CFR and 14-258x faster than the CPU implementation LiteEFG on the four largest games, while reproducing reference iterates bitwise on CPU.
- Static dataflow compilation reduces framework operations by up to 18.1x per iteration.
- Runs 14-258x faster than LiteEFG, a leading open-source CPU solver, on the largest games.
- CUDA Graph Replay enables single-launch replay because shapes, indices and buffers are fixed.
- Compiled representation alone is 2.2-51.1x faster than the GPU baseline on eight CPU threads.
- Tags
- compilerscounterfactual-regret-minimizationcudagame-solvinggpuhigh-performance-computingnvidia-a100
- Vendors
- NVIDIA
From the original article
1 min read at arxiv.orgCounterfactual regret minimization (CFR) is one of the few large numerical workloads that still runs faster on CPUs than on GPUs. Each iteration sweeps a game tree with up to billions of states in millions of small, interdependent gather and scatter steps issued through a generic tree interface.
By Boning Li · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.