Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2609.08231

SemBridge: Compiling Consumer Observations into Cross-Stack Communication Plans

Genlang Chen·Junyi Zhu·Yuanshan Lin
View original source ↗Read source document ↗Export metadata ↓

Abstract

Distributed-tensor systems specify where values reside, while collective systems optimize how requested operations execute. At a boundary between vendor runtimes that cannot share a native communicator, neither abstraction states what a remote consumer must observe. SemBridge fills this gap by compiling graph and runtime facts into a typed contract for the consumer-visible result and its delivery obligations. The contract captures provenance, substitutability, completion, authority, demand, and native-domain locality. A deterministic lowerer constructs backend-neutral communication plans, and a symbolic checker validates each plan before execution across CUDA/NCCL and CANN/HCCL. An independent layout-only planner handles all 72 structural transitions but establishes only 54 complete obligations; a byte-only minimizer proposes 40 semantically invalid candidates, all rejected by SemBridge. On nine real edges, SemBridge produces distinct observation-aware plans that reduce startups on all nine and payload bytes on the three result edges. A live CUDA/CANN run derives and executes full-logit reconstruction, source projection, and owner-token delivery from log-probability, token-only, and owner-scoped requests. On a measured two-host 1-GbE capacity-spillover deployment, source projection cuts result traffic by more than 99.97% and increases throughput by 8.92-80.20% across Dense, MoE, and MiniMax workloads. All 18 MiniMax restart pairs at concurrency 1, 8, and 16 favor source projection. A Qwen3-14B MLP slice additionally verifies bitwise activation-shard delivery and HCCL completion of row-parallel partials. These results establish consumer observation as a semantic layer between placement and collective execution.

Explore related subjects

cs.DC

Keep this discovery

BibTeXRIS

Genlang Chen, Junyi Zhu, Yuanshan Lin. 2026-09-08. SemBridge: Compiling Consumer Observations into Cross-Stack Communication Plans. https://arxiv.org/abs/2609.08231

Cite the original work for its findings. Save a collection to share your selection of sources.

Discover connections

cs.DC subject

Connections use source metadata and explicit phrase matches, not verified experimental comparisons.

KEEP EXPLORING

Related papers

arXiv2026

DC-Gen: Post-Training Diffusion Acceleration with Deeply Compressed Latent Space

Existing text-to-image diffusion models excel at generating high-quality images, but face significant efficiency challenges when scaled to high resolutions, like 4K image generation. While previous research accelerates diffusion models in various aspects, it seldom handles the inherent redundancy within the latent space. To bridge this gap, this paper introduces DC-Gen, a general framework that accelerates text-to-image diffusion models by leveraging a deeply compressed latent space. Rather than a costly training-from-scratch approach, DC-Gen uses an efficient post-training pipeline to preserve the quality of the base model. A key challenge in this paradigm is the representation gap between the base model's latent space and a deeply compressed latent space, which can lead to instability during direct fine-tuning. To overcome this, DC-Gen first bridges the representation gap with a lightweight embedding alignment training. Once the latent embeddings are aligned, only a small amount of LoRA fine-tuning is needed to unlock the base model's inherent generation quality. We verify DC-Gen's effectiveness on SANA and FLUX.1-Krea. The resulting DC-Gen-SANA and DC-Gen-FLUX models achieve quality comparable to their base models but with a significant speedup. Specifically, DC-Gen-FLUX reduces the latency of 4K image generation by 53x on the NVIDIA H100 GPU. When combined with NVFP4 SVDQuant, DC-Gen-FLUX generates a 4K image in just 3.5 seconds on a single NVIDIA 5090 GPU, achieving a total latency reduction of 138x compared to the base FLUX.1-Krea model. Code: https://github.com/dc-ai-projects/DC-Gen.

cs.CV↗
arXiv2026

On Reduction and Synthesis of Petri's Cycloids

Cycloids are particular Petri nets for modelling processes of actions and events, belonging to the fundaments of Petri's general systems theory. Defined by four parameters they provide an algebraic formalism to describe strongly synchronized sequential processes. To further investigate their structure, reduction systems of cycloids are defined in the style of rewriting systems and properties of irreducible cycloids are proved. In particular the synthesis of cycloid parameters from their Petri net structure is derived, leading to an efficient method for a decision procedure for cycloid isomorphism.

cs.DC↗
arXiv2026

On Solving Problems of Substantially Super-linear Complexity in $N^{o(1)}$ Rounds in the MPC Model

We study the possibility of designing $N^{o(1)}$-round protocols for problems of substantially super-linear polynomial-time (sequential) complexity in the model of Massively Parallel Computation, where $N$ is the input size. We show that if the machines are not equipped with relatively large local memory and their number does not exceed $N$, then the exponent of the average time complexity of the local computation performed by a machine in a round (in terms of local memory size) in such protocols must be larger than the exponent of the time complexity of the given problem.

cs.DC↗