A Survey of End-to-End Modeling for Distributed DNN Training: Workloads, Simulators, and TCO

Jonas Svedas; Hannah Watson; Nathan Laubeuf; Diksha Moolchandani; Abubakr Nada; Arjun Singh; Dwaipayan Biswas; James Myers; Debjyoti Bhattacharjee

arXiv · 2506.09275

A Survey of End-to-End Modeling for Distributed DNN Training: Workloads, Simulators, and TCO

Jonas Svedas·Hannah Watson·Nathan Laubeuf·Diksha Moolchandani·Abubakr Nada·Arjun Singh·Dwaipayan Biswas·James Myers·Debjyoti Bhattacharjee

View original source ↗Read source document ↗Export metadata ↓

Abstract

Distributed deep neural networks (DNNs) have become a cornerstone for scaling machine learning to meet the demands of increasingly complex applications. However, the rapid growth in model complexity far outpaces CMOS technology scaling, making sustainable and efficient system design a critical challenge. Addressing this requires coordinated co-design across software, hardware, and technology layers. Due to the prohibitive cost and complexity of deploying full-scale training systems, simulators play a pivotal role in enabling this design exploration. This survey reviews the landscape of distributed DNN training simulators, focusing on three major dimensions: workload representation, simulation infrastructure, and models for total cost of ownership (TCO) including carbon emissions. It covers how workloads are abstracted and used in simulation, outlines common workload representation methods, and includes comprehensive comparison tables covering both simulation frameworks and TCO/emissions models, detailing their capabilities, assumptions, and areas of focus. In addition to synthesizing existing tools, the survey highlights emerging trends, common limitations, and open research challenges across the stack. By providing a structured overview, this work supports informed decision-making in the design and evaluation of distributed training systems.

Explore related subjects

cs.DC

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Jonas Svedas, Hannah Watson, Nathan Laubeuf, Diksha Moolchandani, Abubakr Nada, Arjun Singh, Dwaipayan Biswas, James Myers, Debjyoti Bhattacharjee. 2025-06-10. A Survey of End-to-End Modeling for Distributed DNN Training: Workloads, Simulators, and TCO. https://arxiv.org/abs/2506.09275

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

KEEP EXPLORING

Related papers

arXiv2026

Reforge: Low-Latency Distributed GNN Serving with Selective Embedding Recomputation

Graph Neural Networks (GNNs) have been widely adopted for their ability to compute expressive node representations in graph datasets. However, serving GNNs on large graphs is challenging due to the high communication, computation, and memory overheads of constructing and executing computation graphs, which represent information flow across large neighborhoods. Existing approximation techniques in training can mitigate the overheads but, in serving, still lead to high latency and/or accuracy loss. To this end, we propose Reforge, a system that enables low-latency GNN serving for large graphs with minimal accuracy loss through two key ideas. First, Reforge employs selective recomputation of precomputed embeddings, which allows for reusing precomputed computation subgraphs while selectively recomputing a small fraction to minimize accuracy loss. Second, we develop computation graph parallelism, which reduces communication overhead by parallelizing the creation and execution of computation graphs across machines. Our evaluation with large graph datasets and GNN models shows that Reforge significantly outperforms state-of-the-art techniques.

cs.DC↗

arXiv2026

Agentic AI Workload Characteristics

Agentic AI shifts LLM serving from isolated prompt-generation requests to stateful, multi-turn executions that repeatedly invoke the model, call tools, and grow context over time. This paper characterizes ReAct-style agents from both the LLM-serving and tool-execution perspectives using an end-to-end tracing infrastructure across reasoning and non-reasoning Gemma and Qwen configurations on five agentic benchmarks. Our study shows that agentic workloads are not simply long-prompt workloads: with effective context caching, most input tokens are reused across turns, making execution decode-dominated while increasing dependence on long-lived KV-cache state. We also find that tool use has a clear temporal structure, with agents shifting from read/explore behavior early in execution to execute/write behavior later. These results show that efficient agentic serving must jointly manage repeated model re-entry, persistent context state, and workload-dependent tool behavior.

cs.DC↗

arXiv2026

Byzantine Causal Reliable Broadcast (BCRB) with Constant-Size Message Metadata

Asynchronous Byzantine Reliable Broadcast (BRB) is a fundamental primitive that guarantees agreement and validity in distributed systems subject to Byzantine faults, but it lacks ordering guarantees. In this paper, we address Byzantine Causal Reliable Broadcast (BCRB), which builds on BRB to enforce causal message ordering. We present a novel BCRB protocol that decouples causal ordering from the BRB layer, achieving constant-size $\mathcal{O}(1)$ message metadata overhead and $\mathcal{O}(n^2)$ communication word complexity as against $\mathcal{O}(n^3)$ communication word complexity of existing protocols; here $n$ is the number of processes. We present two variants of our protocol: a cryptographic version using a threshold encryption scheme and sequence gating, and its non-cryptographic version. In the cryptographic version, senders broadcast ciphertexts immediately, and decryption shares are piggybacked on out-of-band ACKs, preventing early decryption and front-running. In both versions, causal safety is achieved probabilistically. We evaluate the probability of causal safety violations using a random variable path analysis under independent exponential link delay distributions. We show that both variants satisfy liveness and the probability of weak safety violation is bounded by $\mathcal{O}(f^{-3}\cdot\ln^3 f)$, where $f$ is the upper bound on the number of Byzantine processes, and $f < n/3$ and $f=\mathcal{O}(n)$. Further, for the crypto version, we show that the probability of strong safety violation is bounded by $\mathcal{O}(f^{-1} \cdot \ln^2 f)$. We also show how to modify our two protocols to guarantee 100\% weak safety keeping $\mathcal{O}(1)$ message space overhead but with $\mathcal{O}(n^3)$ messages and $\mathcal{O}(n^3)$ communication word complexity.

cs.DC↗