{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,5]],"date-time":"2026-08-05T06:45:19Z","timestamp":1785912319393,"version":"3.56.0"},"reference-count":70,"publisher":"IEEE","license":[{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,6,1]],"date-time":"2026-06-01T00:00:00Z","timestamp":1780272000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,6]]},"DOI":"10.1109\/isca66397.2026.00170","type":"proceedings-article","created":{"date-parts":[[2026,8,4]],"date-time":"2026-08-04T19:08:07Z","timestamp":1785870487000},"page":"2425-2441","source":"Crossref","is-referenced-by-count":0,"title":["Scalable Synthesis of Distributed Llm Workloads Through Symbolic Tensor Graphs"],"prefix":"10.1109","author":[{"given":"Changhai","family":"Man","sequence":"first","affiliation":[{"name":"Georgia Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Joongun","family":"Park","sequence":"additional","affiliation":[{"name":"Georgia Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hanjiang","family":"Wu","sequence":"additional","affiliation":[{"name":"Georgia Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Huan","family":"Xu","sequence":"additional","affiliation":[{"name":"Georgia Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Srinivas","family":"Sridharan","sequence":"additional","affiliation":[{"name":"NVIDIA Inc."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tushar","family":"Krishna","sequence":"additional","affiliation":[{"name":"Georgia Institute of Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Deepseek-v2: A strong, economical, and efficient mixture-of-experts language model.","journal-title":"[Online]"},{"key":"ref2","article-title":"TensorFlow: Large-scale machine learning on heterogeneous systems","volume-title":"software available","author":"Abadi","year":"2015"},{"key":"ref3","volume-title":"Gqa: Training generalized multi-query transformer models from multi-head checkpoints","author":"Ainslie","year":"2023"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/micro61859.2024.00021"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3531348.3532176"},{"key":"ref6","volume-title":"Language models are few-shot learners","author":"Brown","year":"2020"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3620666.3655592"},{"key":"ref8","year":"2025","journal-title":"CS-3 System"},{"key":"ref9","volume-title":"Tvm: An automated end-to-end optimizing compiler for deep learning","author":"Chen","year":"2018"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/iiswc63097.2024.00012"},{"key":"ref11","volume-title":"Palm: Scaling language modeling with pathways","author":"Chowdhery","year":"2022"},{"key":"ref12","volume-title":"Deepseekmoe: Towards ultimate expert specialization in mixture-of-experts language models","author":"Dai","year":"2024"},{"key":"ref13","year":"2025","journal-title":"Deepseek v3\/r1 inference system overview. GitHub: Open Infra Index, Day 6 of 2025 Open Source Week"},{"key":"ref14","author":"DeepSeek-AI","year":"2025","journal-title":"Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning"},{"key":"ref15","volume-title":"Deepseek-v3 technical report","author":"DeepSeek-AI","year":"2025"},{"key":"ref16","volume-title":"Bert: Pre-training of deep bidirectional transformers for language understanding","author":"Devlin","year":"2019"},{"key":"ref17","volume-title":"Proteus: Simulating the performance of distributed dnn training","author":"Duan","year":"2023"},{"key":"ref18","volume-title":"Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity","author":"Fedus","year":"2022"},{"key":"ref19","volume-title":"Zamba: A compact 7b ssm hybrid model","author":"Glorioso","year":"2024"},{"key":"ref20","volume-title":"The llama 3 herd of models","author":"Grattafiori","year":"2024"},{"key":"ref21","volume-title":"Mamba: Linear-time sequence modeling with selective state spaces","author":"Gu","year":"2024"},{"key":"ref22","volume-title":"Pipedream: Fast and efficient pipeline parallel dnn training","author":"Harlap","year":"2018"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA59077.2024.00064"},{"key":"ref24","volume-title":"Gpipe: Efficient training of giant neural networks using pipeline parallelism","author":"Huang","year":"2019"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1145\/3581784.3607102"},{"key":"ref26","volume-title":"Beyond data and model parallelism for deep neural networks","author":"Jia","year":"2018"},{"key":"ref27","volume-title":"Mistral 7b","author":"Jiang","year":"2023"},{"key":"ref28","volume-title":"Mixtral of experts","author":"Jiang","year":"2024"},{"key":"ref29","volume-title":"Scaling laws for neural language models","author":"Kaplan","year":"2020"},{"key":"ref30","volume-title":"Reducing activation recomputation in large transformer models","author":"Korthikanti","year":"2022"},{"key":"ref31","volume-title":"Gshard: Scaling giant models with conditional computation and automatic sharding","author":"Lepikhin","year":"2020"},{"key":"ref32","doi-asserted-by":"crossref","DOI":"10.14778\/3415478.3415530","volume-title":"Pytorch distributed: Experiences on accelerating data parallel training","author":"Li","year":"2020"},{"key":"ref33","volume-title":"Mystique: Enabling accurate and scalable generation of production ai benchmarks","author":"Liang","year":"2023"},{"key":"ref34","volume-title":"Lumos: Efficient performance modeling and estimation for large-scale 1lm training","author":"Liang","year":"2025"},{"key":"ref35","volume-title":"Jamba: A hybrid transformer-mamba language model","author":"Lieber","year":"2024"},{"key":"ref36","year":"2025","journal-title":"The llama 4 herd: The beginning of a new era of natively multimodal ai innovation"},{"key":"ref37","volume-title":"Chakra working group","year":"2023"},{"key":"ref38","volume-title":"Chakra schema","year":"2024"},{"key":"ref39","author":"Nguyen","year":"2024","journal-title":"Minibatch coresets for memory-efficient training of large language models"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1145\/3453483.3454083"},{"key":"ref41","year":"2024","journal-title":"Nvidia cupti - cuda profiling tools interface"},{"key":"ref42","year":"2024","journal-title":"Cuda profiling tools interface (cupti)"},{"key":"ref43","year":"2025","journal-title":"NVIDIA HGX Platform"},{"key":"ref44","volume-title":"Pytorch: An imperative style, high-performance deep learning library","author":"Paszke","year":"2019"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1109\/9780470544839.ch42"},{"key":"ref46","author":"Contributors","year":"2024","journal-title":"Pytorch profiler recipe"},{"key":"ref47","author":"Team","year":"2025","journal-title":"Kineto: Performance profiling library for pytorch"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1145\/3470496.3527382"},{"key":"ref50","doi-asserted-by":"crossref","DOI":"10.1109\/ISCA45697.2020.00045","volume-title":"Mlperf inference benchmark","author":"Reddi","year":"2020"},{"key":"ref51","volume-title":"Torch.fx: Practical program capture and transformation for deep learning in python","author":"Reed","year":"2022"},{"key":"ref52","author":"Research","year":"2023","journal-title":"Param: A trace abstraction for ml workloads"},{"key":"ref53","volume-title":"Scale-sim: Systolic cnn accelerator simulator","author":"Samajdar","year":"2019"},{"key":"ref54","volume-title":"Megatron-lm: Training multi-billion parameter language models using model parallelism","author":"Shoeybi","year":"2020"},{"key":"ref55","article-title":"Mlcommons chakra: Advancing performance benchmarking and co-design using standardized execution traces","volume-title":"Proceedings of the Ninth Annual Conference on Machine Learning and Systems (MLSys 2026)","author":"Sridharan"},{"key":"ref56","year":"2023","journal-title":"Kineto: A cpu+gpu profiling library for pytorch"},{"key":"ref57","volume-title":"Llama: Open and efficient foundation language models","author":"Touvron","year":"2023"},{"key":"ref58","first-page":"267","article-title":"Unity: Accelerating DNN training through joint optimization of algebraic transformations and parallelization","volume-title":"16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22)","author":"Unger","year":"2022"},{"key":"ref59","volume-title":"Tensor comprehensions: Framework-agnostic highperformance machine learning abstractions","author":"Vasilache","year":"2018"},{"key":"ref60","volume-title":"Attention is all you need","author":"Vaswani","year":"2023"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.1109\/hoti63208.2024.00013"},{"key":"ref62","first-page":"541","article-title":"SimAI: Unifying architecture design and performance tuning for Large-Scale large language model training with scalability and precision","volume-title":"22nd USENIX Symposium on Networked Systems Design and Implementation (NSDI 25)","author":"Wang","year":"2025"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.1109\/ISPASS57527.2023.00035"},{"key":"ref64","author":"Won","year":"2023","journal-title":"Astra-sim2.0: Modeling hierarchical networks and disaggregated systems for large-model training at scale"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1145\/3617232.3624858"},{"key":"ref66","author":"Yoo","year":"2026","journal-title":"Towards easy and realistic network infrastructure testing for large-scale machine learning"},{"key":"ref67","first-page":"1","article-title":"Apollo: Automatic partition-based operator fusion through layer by layer optimization","volume-title":"Proceedings of Machine Learning and Systems","volume":"4","author":"Zhao","year":"2022"},{"key":"ref68","volume-title":"Pytorch fsdp: Experiences on scaling fully sharded data parallel","author":"Zhao","year":"2023"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.1145\/3689031.3717461"},{"key":"ref70","first-page":"761","article-title":"Resiliency at scale: Managing Google\u2019s TPUv4 machine learning supercomputer","volume-title":"21st USENIX Symposium on Networked Systems Design and Implementation (NSDI 24)","author":"Zu","year":"2024"}],"event":{"name":"2026 ACM\/IEEE 53rd Annual International Symposium on Computer Architecture (ISCA)","location":"Raleigh, NC, USA","start":{"date-parts":[[2026,6,27]]},"end":{"date-parts":[[2026,7,1]]}},"container-title":["2026 ACM\/IEEE 53rd Annual International Symposium on Computer Architecture (ISCA)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11617393\/11617394\/11617571.pdf?arnumber=11617571","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,5]],"date-time":"2026-08-05T05:51:46Z","timestamp":1785909106000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11617571\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6]]},"references-count":70,"URL":"https:\/\/doi.org\/10.1109\/isca66397.2026.00170","relation":{},"subject":[],"published":{"date-parts":[[2026,6]]}}}