AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Characterizing Vision-Language-Action Models across XPUs: Constraints and Acceleration for On-Robot Deployment

Kaijun Zhou, Qiwei Chen, Da Peng, Zhiyang Li, Xijun Li, Jinyu Gu

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Distributed LLM inference systems · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
compile π0 with torch.compile / ATCπ0 · RTX 4090latency (ms): 102.3 -> 35.2Improved
compile π0 with torch.compileπ0 · Jetson Thorlatency (ms): 246.0 -> 163.0Improved
add DP-Cache (S=4) to diffusion policyDiffusion Policy · LIBERO Spatial/Object/Long, RTX 4090latency (ms): 378 -> 200Improved
increase DP-Cache cache steps from S=4 to S=8 (ablation)DP-Cache · LIBERO Spatialsuccess rate (%): 78.9 -> 71.6Hurt
add DP-Cache (S=4) to π0.5-droid on Frankaπ0.5-droid · Franka table cleaning (50 trials)success rate (%): 76 -> 70Hurt
add ATC compilation to DP-Cache on Ascend 310P (ablation)DP-Cache · Diffusion Policy, Ascend 310Pspeedup: 3.5× -> 6.0×Improved
add 4-bit quantization to Spec-VLA (ablation)OpenVLA · LIBERO averagesuccess rate (%): 75.1 -> 68.5 (Spec. vs Spec.+Cache); Quant. alone 76.5 -> 67.7Hurt
combine Spec-VLA with Cache-VLA (ablation)OpenVLA · LIBERO averagespeedup: 1.11× -> 1.29× speedup; SR 75.1 -> 68.5Mixed
increase stale denoising steps in V-AEFusion from 5 to 7 (ablation)V-AEFusion · JAKA S5 Putting Block in Bowlsuccess rate (%): 88 -> 50Hurt
use 1 to 3 stale denoising steps in V-AEFusion (ablation)V-AEFusion · JAKA S5 Putting Block in Bowlsuccess rate (%): 92 -> 90No clear effect
add V-AEFusion pipeline to π0π0 · RTX 4090latency (ms): 102 -> 77Improved
add V-AEFusion pipeline to π0π0 · Ascend 310Platency (ms): 818 -> 820No clear effect

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy