arXiv subjects

Rachee Singh

Publications and source records attributed to Rachee Singh.

22 records · Page 2Linked to original sources

arXiv2022

TACCL: Guiding Collective Algorithm Synthesis using Communication Sketches

Machine learning models are increasingly being trained across multiple GPUs and servers. In this setting, data is transferred between GPUs using communication collectives such as AlltoAll and AllReduce, which can become a significant bottleneck in training large models. Thus, it is important to use efficient algorithms for collective communication. We develop TACCL, a tool that enables algorithm designers to guide a synthesizer into automatically generating algorithms for a given hardware configuration and communication collective. TACCL uses a novel communication sketch abstraction to get crucial information from the designer to significantly reduce the search space and guide the synthesizer towards better algorithms. TACCL also uses a novel encoding of the problem that allows it to scale beyond single-node topologies. We use TACCL to synthesize algorithms for three collectives and two hardware topologies: DGX-2 and NDv2. We demonstrate that the algorithms synthesized by TACCL outperform the Nvidia Collective Communication Library (NCCL) by up to 6.7x. We also show that TACCL can speed up end-to-end training of Transformer-XL and BERT models by 11%--2.3x for different batch sizes.

cs.DC↗

arXiv2022

Glowing in the Dark Uncovering IPv6 Address Discovery and Scanning Strategies in the Wild

In this work we identify scanning strategies of IPv6 scanners on the Internet. We offer a unique perspective on the behavior of IPv6 scanners by conducting controlled experiments leveraging a large and unused /56 IPv6 subnet. We selectively make parts of the subnet visible to scanners by hosting applications that make direct or indirect contact with IPv6- capable servers on the Internet. By careful experiment design, we mitigate the effects of hidden variables on scans sent to our /56 subnet and establish causal relationships between IPv6 host activity types and the scanner attention they evoke. We show that IPv6 host activities e.g., Web browsing, membership in the NTP pool and Tor network, cause scanners to send a magnitude higher number of unsolicited IP scans and reverse DNS queries to our subnet than before. DNS scanners focus their scans in narrow regions of the address space where our applications are hosted whereas IP scanners broadly scan the entire subnet. Even after the host activity from our subnet subsides, we observe persistent residual scanning to portions of the address space that previously hosted applications

cs.NI↗

arXiv2016

The Politics of Routing: Investigating the Relationship Between AS Connectivity and Internet Freedom

The Internet's importance for promoting free and open communication has led to widespread crackdowns on its use in countries around the world. In this study, we investigate the relationship between national policies around freedom of speech and Internet topology in countries around the world. We combine techniques from network measurement and machine learning to identify features of Internet structure at the national level that are the best indicators of a country's level of freedom. We find that IP density and path lengths to other countries are the best indicators of a country's freedom. We also find that our methods predict freedom categories for countries with 91% accuracy.

cs.NI↗

arXiv2016

Holding all the ASes: Identifying and Circumventing the Pitfalls of AS-aware Tor Client Design

Traffic correlation attacks to de-anonymize Tor users are possible when an adversary is in a position to observe traffic entering and exiting the Tor network. Recent work has brought attention to the threat of these attacks by network-level adversaries (e.g., Autonomous Systems). We perform a historical analysis to understand how the threat from AS-level traffic correlation attacks has evolved over the past five years. We find that despite a large number of new relays added to the Tor network, the threat has grown. This points to the importance of increasing AS-level diversity in addition to capacity of the Tor network. We identify and elaborate on common pitfalls of AS-aware Tor client design and construction. We find that succumbing to these pitfalls can negatively impact three major aspects of an AS-aware Tor client -- (1) security against AS-level adversaries, (2) security against relay-level adversaries, and (3) performance. Finally, we propose and evaluate a Tor client -- Cipollino -- which avoids these pitfalls using state-of-the-art in network-measurement. Our evaluation shows that Cipollino is able to achieve better security against network-level adversaries while maintaining security against relay-level adversaries and

cs.CR↗