@article{12566, abstract = {Approximate agreement is one of the few variants of consensus that can be solved in a wait-free manner in asynchronous systems where processes communicate by reading and writing to shared memory. In this work, we consider a natural generalisation of approximate agreement on arbitrary undirected connected graphs. Each process is given a node of the graph as input and, if non-faulty, must output a node such that – all the outputs are within distance 1 of one another, and – each output value lies on a shortest path between two input values. From prior work, it is known that there is no wait-free algorithm among processes for this problem on any cycle of length , by reduction from 2-set agreement (Castañeda et al., 2018). In this work, we investigate the solvability of this task on general graphs. We give a new, direct proof of the impossibility of approximate agreement on cycles of length , via a generalisation of Sperner's Lemma to convex polygons. We also extend the reduction from 2-set agreement to a larger class of graphs, showing that approximate agreement on these graphs is unsolvable. On the positive side, we present a wait-free algorithm for a different class of graphs, which properly contains the class of chordal graphs.}, author = {Alistarh, Dan-Adrian and Ellen, Faith and Rybicki, Joel}, issn = {0304-3975}, journal = {Theoretical Computer Science}, number = {2}, publisher = {Elsevier}, title = {{Wait-free approximate agreement on graphs}}, doi = {10.1016/j.tcs.2023.113733}, volume = {948}, year = {2023}, } @inproceedings{13053, abstract = {Deep neural networks (DNNs) often have to be compressed, via pruning and/or quantization, before they can be deployed in practical settings. In this work we propose a new compression-aware minimizer dubbed CrAM that modifies the optimization step in a principled way, in order to produce models whose local loss behavior is stable under compression operations such as pruning. Thus, dense models trained via CrAM should be compressible post-training, in a single step, without significant accuracy loss. Experimental results on standard benchmarks, such as residual networks for ImageNet classification and BERT models for language modelling, show that CrAM produces dense models that can be more accurate than the standard SGD/Adam-based baselines, but which are stable under weight pruning: specifically, we can prune models in one-shot to 70-80% sparsity with almost no accuracy loss, and to 90% with reasonable (∼1%) accuracy loss, which is competitive with gradual compression methods. Additionally, CrAM can produce sparse models which perform well for transfer learning, and it also works for semi-structured 2:4 pruning patterns supported by GPU hardware. The code for reproducing the results is available at this https URL .}, author = {Peste, Elena-Alexandra and Vladu, Adrian and Kurtic, Eldar and Lampert, Christoph and Alistarh, Dan-Adrian}, booktitle = {11th International Conference on Learning Representations }, location = {Kigali, Rwanda }, title = {{CrAM: A Compression-Aware Minimizer}}, year = {2023}, } @phdthesis{13074, abstract = {Deep learning has become an integral part of a large number of important applications, and many of the recent breakthroughs have been enabled by the ability to train very large models, capable to capture complex patterns and relationships from the data. At the same time, the massive sizes of modern deep learning models have made their deployment to smaller devices more challenging; this is particularly important, as in many applications the users rely on accurate deep learning predictions, but they only have access to devices with limited memory and compute power. One solution to this problem is to prune neural networks, by setting as many of their parameters as possible to zero, to obtain accurate sparse models with lower memory footprint. Despite the great research progress in obtaining sparse models that preserve accuracy, while satisfying memory and computational constraints, there are still many challenges associated with efficiently training sparse models, as well as understanding their generalization properties. The focus of this thesis is to investigate how the training process of sparse models can be made more efficient, and to understand the differences between sparse and dense models in terms of how well they can generalize to changes in the data distribution. We first study a method for co-training sparse and dense models, at a lower cost compared to regular training. With our method we can obtain very accurate sparse networks, and dense models that can recover the baseline accuracy. Furthermore, we are able to more easily analyze the differences, at prediction level, between the sparse-dense model pairs. Next, we investigate the generalization properties of sparse neural networks in more detail, by studying how well different sparse models trained on a larger task can adapt to smaller, more specialized tasks, in a transfer learning scenario. Our analysis across multiple pruning methods and sparsity levels reveals that sparse models provide features that can transfer similarly to or better than the dense baseline. However, the choice of the pruning method plays an important role, and can influence the results when the features are fixed (linear finetuning), or when they are allowed to adapt to the new task (full finetuning). Using sparse models with fixed masks for finetuning on new tasks has an important practical advantage, as it enables training neural networks on smaller devices. However, one drawback of current pruning methods is that the entire training cycle has to be repeated to obtain the initial sparse model, for every sparsity target; in consequence, the entire training process is costly and also multiple models need to be stored. In the last part of the thesis we propose a method that can train accurate dense models that are compressible in a single step, to multiple sparsity levels, without additional finetuning. Our method results in sparse models that can be competitive with existing pruning methods, and which can also successfully generalize to new tasks.}, author = {Peste, Elena-Alexandra}, issn = {2663-337X}, pages = {147}, publisher = {Institute of Science and Technology Austria}, title = {{Efficiency and generalization of sparse neural networks}}, doi = {10.15479/at:ista:13074}, year = {2023}, } @article{14364, abstract = {We introduce extension-based proofs, a class of impossibility proofs that includes valency arguments. They are modelled as an interaction between a prover and a protocol. Using proofs based on combinatorial topology, it has been shown that it is impossible to deterministically solve -set agreement among processes or approximate agreement on a cycle of length 4 among processes in a wait-free manner in asynchronous models where processes communicate using objects that can be constructed from shared registers. However, it was unknown whether proofs based on simpler techniques were possible. We show that these impossibility results cannot be obtained by extension-based proofs in the iterated snapshot model and, hence, extension-based proofs are limited in power.}, author = {Alistarh, Dan-Adrian and Aspnes, James and Ellen, Faith and Gelashvili, Rati and Zhu, Leqi}, issn = {1095-7111}, journal = {SIAM Journal on Computing}, number = {4}, pages = {913--944}, publisher = {Society for Industrial and Applied Mathematics}, title = {{Why extension-based proofs fail}}, doi = {10.1137/20M1375851}, volume = {52}, year = {2023}, } @inproceedings{14458, abstract = {We show for the first time that large-scale generative pretrained transformer (GPT) family models can be pruned to at least 50% sparsity in one-shot, without any retraining, at minimal loss of accuracy. This is achieved via a new pruning method called SparseGPT, specifically designed to work efficiently and accurately on massive GPT-family models. We can execute SparseGPT on the largest available open-source models, OPT-175B and BLOOM-176B, in under 4.5 hours, and can reach 60% unstructured sparsity with negligible increase in perplexity: remarkably, more than 100 billion weights from these models can be ignored at inference time. SparseGPT generalizes to semi-structured (2:4 and 4:8) patterns, and is compatible with weight quantization approaches. The code is available at: https://github.com/IST-DASLab/sparsegpt.}, author = {Frantar, Elias and Alistarh, Dan-Adrian}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, issn = {2640-3498}, location = {Honolulu, Hawaii, HI, United States}, pages = {10323--10337}, publisher = {ML Research Press}, title = {{SparseGPT: Massive language models can be accurately pruned in one-shot}}, volume = {202}, year = {2023}, } @inproceedings{14460, abstract = {We provide an efficient implementation of the backpropagation algorithm, specialized to the case where the weights of the neural network being trained are sparse. Our algorithm is general, as it applies to arbitrary (unstructured) sparsity and common layer types (e.g., convolutional or linear). We provide a fast vectorized implementation on commodity CPUs, and show that it can yield speedups in end-to-end runtime experiments, both in transfer learning using already-sparsified networks, and in training sparse networks from scratch. Thus, our results provide the first support for sparse training on commodity hardware.}, author = {Nikdan, Mahdi and Pegolotti, Tommaso and Iofinova, Eugenia B and Kurtic, Eldar and Alistarh, Dan-Adrian}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, issn = {2640-3498}, location = {Honolulu, Hawaii, HI, United States}, pages = {26215--26227}, publisher = {ML Research Press}, title = {{SparseProp: Efficient sparse backpropagation for faster training of neural networks at the edge}}, volume = {202}, year = {2023}, } @inproceedings{14461, abstract = {Communication-reduction techniques are a popular way to improve scalability in data-parallel training of deep neural networks (DNNs). The recent emergence of large language models such as GPT has created the need for new approaches to exploit data-parallelism. Among these, fully-sharded data parallel (FSDP) training is highly popular, yet it still encounters scalability bottlenecks. One reason is that applying compression techniques to FSDP is challenging: as the vast majority of the communication involves the model’s weights, direct compression alters convergence and leads to accuracy loss. We present QSDP, a variant of FSDP which supports both gradient and weight quantization with theoretical guarantees, is simple to implement and has essentially no overheads. To derive QSDP we prove that a natural modification of SGD achieves convergence even when we only maintain quantized weights, and thus the domain over which we train consists of quantized points and is, therefore, highly non-convex. We validate this approach by training GPT-family models with up to 1.3 billion parameters on a multi-node cluster. Experiments show that QSDP preserves model accuracy, while completely removing the communication bottlenecks of FSDP, providing end-to-end speedups of up to 2.2x.}, author = {Markov, Ilia and Vladu, Adrian and Guo, Qi and Alistarh, Dan-Adrian}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, issn = {2640-3498}, location = {Honolulu, Hawaii, HI, United States}, pages = {24020--24044}, publisher = {ML Research Press}, title = {{Quantized distributed training of large models with convergence guarantees}}, volume = {202}, year = {2023}, } @inproceedings{14771, abstract = {Pruning—that is, setting a significant subset of the parameters of a neural network to zero—is one of the most popular methods of model compression. Yet, several recent works have raised the issue that pruning may induce or exacerbate bias in the output of the compressed model. Despite existing evidence for this phenomenon, the relationship between neural network pruning and induced bias is not well-understood. In this work, we systematically investigate and characterize this phenomenon in Convolutional Neural Networks for computer vision. First, we show that it is in fact possible to obtain highly-sparse models, e.g. with less than 10% remaining weights, which do not decrease in accuracy nor substantially increase in bias when compared to dense models. At the same time, we also find that, at higher sparsities, pruned models exhibit higher uncertainty in their outputs, as well as increased correlations, which we directly link to increased bias. We propose easy-to-use criteria which, based only on the uncompressed model, establish whether bias will increase with pruning, and identify the samples most susceptible to biased predictions post-compression. Our code can be found at https://github.com/IST-DASLab/pruned-vision-model-bias.}, author = {Iofinova, Eugenia B and Peste, Elena-Alexandra and Alistarh, Dan-Adrian}, booktitle = {2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition}, issn = {2575-7075}, location = {Vancouver, BC, Canada}, pages = {24364--24373}, publisher = {IEEE}, title = {{Bias in pruned vision models: In-depth analysis and countermeasures}}, doi = {10.1109/cvpr52729.2023.02334}, year = {2023}, } @inproceedings{12299, abstract = {Transfer learning is a classic paradigm by which models pretrained on large “upstream” datasets are adapted to yield good results on “downstream” specialized datasets. Generally, more accurate models on the “upstream” dataset tend to provide better transfer accuracy “downstream”. In this work, we perform an in-depth investigation of this phenomenon in the context of convolutional neural networks (CNNs) trained on the ImageNet dataset, which have been pruned-that is, compressed by sparsifiying their connections. We consider transfer using unstructured pruned models obtained by applying several state-of-the-art pruning methods, including magnitude-based, second-order, regrowth, lottery-ticket, and regularization approaches, in the context of twelve standard transfer tasks. In a nutshell, our study shows that sparse models can match or even outperform the transfer performance of dense models, even at high sparsities, and, while doing so, can lead to significant inference and even training speedups. At the same time, we observe and analyze significant differences in the behaviour of different pruning methods. The code is available at: https://github.com/IST-DASLab/sparse-imagenet-transfer.}, author = {Iofinova, Eugenia B and Peste, Elena-Alexandra and Kurtz, Mark and Alistarh, Dan-Adrian}, booktitle = {2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition}, issn = {2575-7075}, location = {New Orleans, LA, United States}, pages = {12256--12266}, publisher = {Institute of Electrical and Electronics Engineers}, title = {{How well do sparse ImageNet models transfer?}}, doi = {10.1109/cvpr52688.2022.01195}, year = {2022}, } @inproceedings{11180, abstract = {Designing and implementing efficient parallel priority schedulers is an active research area. An intriguing proposed design is the Multi-Queue: given n threads and m ≥ n distinct priority queues, task insertions are performed uniformly at random, while, to delete, a thread picks two queues uniformly at random, and removes the observed task of higher priority. This approach scales well, and has probabilistic rank guarantees: roughly, the rank of each task removed, relative to remaining tasks in all other queues, is O (m) in expectation. Yet, the performance of this pattern is below that of well-engineered schedulers, which eschew theoretical guarantees for practical efficiency. We investigate whether it is possible to design and implement a Multi-Queue-based task scheduler that is both highly-efficient and has analytical guarantees. We propose a new variant called the Stealing Multi-Queue (SMQ), a cache-efficient variant of the Multi-Queue, which leverages both queue affinity---each thread has a local queue, from which tasks are usually removed; but, with some probability, threads also attempt to steal higher-priority tasks from the other queues---and task batching, that is, the processing of several tasks in a single insert / remove step. These ideas are well-known for task scheduling without priorities; our theoretical contribution is showing that, despite relaxations, this design can still provide rank guarantees, which in turn implies bounds on total work performed. We provide a general SMQ implementation which can surpass state-of-the-art schedulers such as OBIM and PMOD in terms of performance on popular graph-processing benchmarks. Notably, the performance improvement comes mainly from the superior rank guarantees provided by our scheduler, confirming that analytically-reasoned approaches can still provide performance improvements for priority task scheduling.}, author = {Postnikova, Anastasiia and Koval, Nikita and Nadiradze, Giorgi and Alistarh, Dan-Adrian}, booktitle = {Proceedings of the 27th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming}, isbn = {9781450392044}, location = {Seoul, Republic of Korea}, pages = {353--367}, publisher = {Association for Computing Machinery}, title = {{Multi-queues can be state-of-the-art priority schedulers}}, doi = {10.1145/3503221.3508432}, year = {2022}, } @inproceedings{11183, abstract = {Subgraph detection has recently been one of the most studied problems in the CONGEST model of distributed computing. In this work, we study the distributed complexity of problems closely related to subgraph detection, mainly focusing on induced subgraph detection. The main line of this work presents lower bounds and parameterized algorithms w.r.t structural parameters of the input graph: - On general graphs, we give unconditional lower bounds for induced detection of cycles and patterns of treewidth 2 in CONGEST. Moreover, by adapting reductions from centralized parameterized complexity, we prove lower bounds in CONGEST for detecting patterns with a 4-clique, and for induced path detection conditional on the hardness of triangle detection in the congested clique. - On graphs of bounded degeneracy, we show that induced paths can be detected fast in CONGEST using techniques from parameterized algorithms, while detecting cycles and patterns of treewidth 2 is hard. - On graphs of bounded vertex cover number, we show that induced subgraph detection is easy in CONGEST for any pattern graph. More specifically, we adapt a centralized parameterized algorithm for a more general maximum common induced subgraph detection problem to the distributed setting. In addition to these induced subgraph detection results, we study various related problems in the CONGEST and congested clique models, including for multicolored versions of subgraph-detection-like problems.}, author = {Nikabadi, Amir and Korhonen, Janne}, booktitle = {25th International Conference on Principles of Distributed Systems}, editor = {Bramas, Quentin and Gramoli, Vincent and Milani, Alessia}, isbn = {9783959772198}, issn = {1868-8969}, location = {Strasbourg, France}, publisher = {Schloss Dagstuhl - Leibniz-Zentrum für Informatik}, title = {{Beyond distributed subgraph detection: Induced subgraphs, multicolored problems and graph parameters}}, doi = {10.4230/LIPIcs.OPODIS.2021.15}, volume = {217}, year = {2022}, } @inproceedings{11184, abstract = {Let G be a graph on n nodes. In the stochastic population protocol model, a collection of n indistinguishable, resource-limited nodes collectively solve tasks via pairwise interactions. In each interaction, two randomly chosen neighbors first read each other’s states, and then update their local states. A rich line of research has established tight upper and lower bounds on the complexity of fundamental tasks, such as majority and leader election, in this model, when G is a clique. Specifically, in the clique, these tasks can be solved fast, i.e., in n polylog n pairwise interactions, with high probability, using at most polylog n states per node. In this work, we consider the more general setting where G is an arbitrary regular graph, and present a technique for simulating protocols designed for fully-connected networks in any connected regular graph. Our main result is a simulation that is efficient on many interesting graph families: roughly, the simulation overhead is polylogarithmic in the number of nodes, and quadratic in the conductance of the graph. As a sample application, we show that, in any regular graph with conductance φ, both leader election and exact majority can be solved in φ^{-2} ⋅ n polylog n pairwise interactions, with high probability, using at most φ^{-2} ⋅ polylog n states per node. This shows that there are fast and space-efficient population protocols for leader election and exact majority on graphs with good expansion properties. We believe our results will prove generally useful, as they allow efficient technology transfer between the well-mixed (clique) case, and the under-explored spatial setting.}, author = {Alistarh, Dan-Adrian and Gelashvili, Rati and Rybicki, Joel}, booktitle = {25th International Conference on Principles of Distributed Systems}, editor = {Bramas, Quentin and Gramoli, Vincent and Milani, Alessia}, isbn = {9783959772198}, issn = {1868-8969}, location = {Strasbourg, France}, publisher = {Schloss Dagstuhl - Leibniz-Zentrum für Informatik}, title = {{Fast graphical population protocols}}, doi = {10.4230/LIPIcs.OPODIS.2021.14}, volume = {217}, year = {2022}, } @inproceedings{11844, abstract = {In the stochastic population protocol model, we are given a connected graph with n nodes, and in every time step, a scheduler samples an edge of the graph uniformly at random and the nodes connected by this edge interact. A fundamental task in this model is stable leader election, in which all nodes start in an identical state and the aim is to reach a configuration in which (1) exactly one node is elected as leader and (2) this node remains as the unique leader no matter what sequence of interactions follows. On cliques, the complexity of this problem has recently been settled: time-optimal protocols stabilize in Θ(n log n) expected steps using Θ(log log n) states, whereas protocols that use O(1) states require Θ(n2) expected steps. In this work, we investigate the complexity of stable leader election on general graphs. We provide the first non-trivial time lower bounds for leader election on general graphs, showing that, when moving beyond cliques, the complexity landscape of leader election becomes very diverse: the time required to elect a leader can range from O(1) to Θ(n3) expected steps. On the upper bound side, we first observe that there exists a protocol that is time-optimal on many graph families, but uses polynomially-many states. In contrast, we give a near-time-optimal protocol that uses only O(log2n) states that is at most a factor log n slower. Finally, we show that the constant-state protocol of Beauquier et al. [OPODIS 2013] is at most a factor n log n slower than the fast polynomial-state protocol. Moreover, among constant-state protocols, this protocol has near-optimal average case complexity on dense random graphs.}, author = {Alistarh, Dan-Adrian and Rybicki, Joel and Voitovych, Sasha}, booktitle = {Proceedings of the Annual ACM Symposium on Principles of Distributed Computing}, isbn = {9781450392624}, location = {Salerno, Italy}, pages = {246--256}, publisher = {Association for Computing Machinery}, title = {{Near-optimal leader election in population protocols on graphs}}, doi = {10.1145/3519270.3538435}, year = {2022}, } @inproceedings{13147, abstract = {We investigate fast and communication-efficient algorithms for the classic problem of minimizing a sum of strongly convex and smooth functions that are distributed among n different nodes, which can communicate using a limited number of bits. Most previous communication-efficient approaches for this problem are limited to first-order optimization, and therefore have \emph{linear} dependence on the condition number in their communication complexity. We show that this dependence is not inherent: communication-efficient methods can in fact have sublinear dependence on the condition number. For this, we design and analyze the first communication-efficient distributed variants of preconditioned gradient descent for Generalized Linear Models, and for Newton’s method. Our results rely on a new technique for quantizing both the preconditioner and the descent direction at each step of the algorithms, while controlling their convergence rate. We also validate our findings experimentally, showing faster convergence and reduced communication relative to previous methods.}, author = {Alimisis, Foivos and Davies, Peter and Alistarh, Dan-Adrian}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, isbn = {9781713845065}, issn = {2640-3498}, location = {Virtual}, pages = {196--206}, publisher = {ML Research Press}, title = {{Communication-efficient distributed optimization with quantized preconditioners}}, volume = {139}, year = {2021}, } @article{8286, abstract = {We consider the following dynamic load-balancing process: given an underlying graph G with n nodes, in each step t≥ 0, one unit of load is created, and placed at a randomly chosen graph node. In the same step, the chosen node picks a random neighbor, and the two nodes balance their loads by averaging them. We are interested in the expected gap between the minimum and maximum loads at nodes as the process progresses, and its dependence on n and on the graph structure. Variants of the above graphical balanced allocation process have been studied previously by Peres, Talwar, and Wieder [Peres et al., 2015], and by Sauerwald and Sun [Sauerwald and Sun, 2015]. These authors left as open the question of characterizing the gap in the case of cycle graphs in the dynamic case, where weights are created during the algorithm’s execution. For this case, the only known upper bound is of 𝒪(n log n), following from a majorization argument due to [Peres et al., 2015], which analyzes a related graphical allocation process. In this paper, we provide an upper bound of 𝒪 (√n log n) on the expected gap of the above process for cycles of length n. We introduce a new potential analysis technique, which enables us to bound the difference in load between k-hop neighbors on the cycle, for any k ≤ n/2. We complement this with a "gap covering" argument, which bounds the maximum value of the gap by bounding its value across all possible subsets of a certain structure, and recursively bounding the gaps within each subset. We provide analytical and experimental evidence that our upper bound on the gap is tight up to a logarithmic factor. }, author = {Alistarh, Dan-Adrian and Nadiradze, Giorgi and Sabour, Amirmojtaba}, issn = {1432-0541}, journal = {Algorithmica}, location = {Virtual, Online; Germany}, publisher = {Springer Nature}, title = {{Dynamic averaging load balancing on cycles}}, doi = {10.1007/s00453-021-00905-9}, year = {2021}, } @article{8723, abstract = {Deep learning at scale is dominated by communication time. Distributing samples across nodes usually yields the best performance, but poses scaling challenges due to global information dissemination and load imbalance across uneven sample lengths. State-of-the-art decentralized optimizers mitigate the problem, but require more iterations to achieve the same accuracy as their globally-communicating counterparts. We present Wait-Avoiding Group Model Averaging (WAGMA) SGD, a wait-avoiding stochastic optimizer that reduces global communication via subgroup weight exchange. The key insight is a combination of algorithmic changes to the averaging scheme and the use of a group allreduce operation. We prove the convergence of WAGMA-SGD, and empirically show that it retains convergence rates similar to Allreduce-SGD. For evaluation, we train ResNet-50 on ImageNet; Transformer for machine translation; and deep reinforcement learning for navigation at scale. Compared with state-of-the-art decentralized SGD variants, WAGMA-SGD significantly improves training throughput (e.g., 2.1× on 1,024 GPUs for reinforcement learning), and achieves the fastest time-to-solution (e.g., the highest score using the shortest training time for Transformer).}, author = {Li, Shigang and Tal Ben-Nun, Tal Ben-Nun and Nadiradze, Giorgi and Girolamo, Salvatore Di and Dryden, Nikoli and Alistarh, Dan-Adrian and Hoefler, Torsten}, issn = {10459219}, journal = {IEEE Transactions on Parallel and Distributed Systems}, number = {7}, publisher = {IEEE}, title = {{Breaking (global) barriers in parallel stochastic optimization with wait-avoiding group averaging}}, doi = {10.1109/TPDS.2020.3040606}, volume = {32}, year = {2021}, } @inproceedings{10217, abstract = {This paper gives tight logarithmic lower bounds on the solo step complexity of leader election in an asynchronous shared-memory model with single-writer multi-reader (SWMR) registers, for both deterministic and randomized obstruction-free algorithms. The approach extends to lower bounds for deterministic and randomized obstruction-free algorithms using multi-writer registers under bounded write concurrency, showing a trade-off between the solo step complexity of a leader election algorithm, and the worst-case number of stalls incurred by a processor in an execution.}, author = {Alistarh, Dan-Adrian and Gelashvili, Rati and Nadiradze, Giorgi}, booktitle = {35th International Symposium on Distributed Computing}, isbn = {9-783-9597-7210-5}, issn = {1868-8969}, location = {Freiburg, Germany}, publisher = {Schloss Dagstuhl - Leibniz Zentrum für Informatik}, title = {{Lower bounds for shared-memory leader election under bounded write contention}}, doi = {10.4230/LIPIcs.DISC.2021.4}, volume = {209}, year = {2021}, } @inproceedings{10219, abstract = {We show that any algorithm that solves the sinkless orientation problem in the supported LOCAL model requires Ω(log n) rounds, and this is tight. The supported LOCAL is at least as strong as the usual LOCAL model, and as a corollary this also gives a new, short and elementary proof that shows that the round complexity of the sinkless orientation problem in the deterministic LOCAL model is Ω(log n).}, author = {Korhonen, Janne and Paz, Ami and Rybicki, Joel and Schmid, Stefan and Suomela, Jukka}, booktitle = {35th International Symposium on Distributed Computing}, isbn = {9-783-9597-7210-5}, issn = {1868-8969}, location = {Freiburg, Germany}, publisher = {Schloss Dagstuhl - Leibniz Zentrum für Informatik}, title = {{Brief announcement: Sinkless orientation is hard also in the supported LOCAL model}}, doi = {10.4230/LIPIcs.DISC.2021.58}, volume = {209}, year = {2021}, } @phdthesis{10429, abstract = {The scalability of concurrent data structures and distributed algorithms strongly depends on reducing the contention for shared resources and the costs of synchronization and communication. We show how such cost reductions can be attained by relaxing the strict consistency conditions required by sequential implementations. In the first part of the thesis, we consider relaxation in the context of concurrent data structures. Specifically, in data structures such as priority queues, imposing strong semantics renders scalability impossible, since a correct implementation of the remove operation should return only the element with highest priority. Intuitively, attempting to invoke remove operations concurrently creates a race condition. This bottleneck can be circumvented by relaxing semantics of the affected data structure, thus allowing removal of the elements which are no longer required to have the highest priority. We prove that the randomized implementations of relaxed data structures provide provable guarantees on the priority of the removed elements even under concurrency. Additionally, we show that in some cases the relaxed data structures can be used to scale the classical algorithms which are usually implemented with the exact ones. In the second part, we study parallel variants of the stochastic gradient descent (SGD) algorithm, which distribute computation among the multiple processors, thus reducing the running time. Unfortunately, in order for standard parallel SGD to succeed, each processor has to maintain a local copy of the necessary model parameter, which is identical to the local copies of other processors; the overheads from this perfect consistency in terms of communication and synchronization can negate the speedup gained by distributing the computation. We show that the consistency conditions required by SGD can be relaxed, allowing the algorithm to be more flexible in terms of tolerating quantized communication, asynchrony, or even crash faults, while its convergence remains asymptotically the same.}, author = {Nadiradze, Giorgi}, issn = {2663-337X}, pages = {132}, publisher = {Institute of Science and Technology Austria}, title = {{On achieving scalability through relaxation}}, doi = {10.15479/at:ista:10429}, year = {2021}, } @inproceedings{10432, abstract = {One key element behind the recent progress of machine learning has been the ability to train machine learning models in large-scale distributed shared-memory and message-passing environments. Most of these models are trained employing variants of stochastic gradient descent (SGD) based optimization, but most methods involve some type of consistency relaxation relative to sequential SGD, to mitigate its large communication or synchronization costs at scale. In this paper, we introduce a general consistency condition covering communication-reduced and asynchronous distributed SGD implementations. Our framework, called elastic consistency, decouples the system-specific aspects of the implementation from the SGD convergence requirements, giving a general way to obtain convergence bounds for a wide variety of distributed SGD methods used in practice. Elastic consistency can be used to re-derive or improve several previous convergence bounds in message-passing and shared-memory settings, but also to analyze new models and distribution schemes. As a direct application, we propose and analyze a new synchronization-avoiding scheduling scheme for distributed SGD, and show that it can be used to efficiently train deep convolutional models for image classification.}, author = {Nadiradze, Giorgi and Markov, Ilia and Chatterjee, Bapi and Kungurtsev, Vyacheslav and Alistarh, Dan-Adrian}, booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence}, location = {Virtual}, number = {10}, pages = {9037--9045}, title = {{Elastic consistency: A practical consistency model for distributed stochastic gradient descent}}, volume = {35}, year = {2021}, }