Parallel programs on shared memory multiprocessors suffer from a slowdown of an order of magnitude due to false sharing. False sharing occurs when multiple threads demand exclusive access to a cache line even though their data locate on non-overlapping portions of the cache line. In this project, we proposed Huron, a hybrid in-house/in-production false sharing detection and repair system. Huron detects and repairs as much false sharing as it can in-house, and relies on its lightweight in-production mechanism for remaining cases. The key idea behind Huron’s in-house false sharing repair is to group together data that is accessed by the same set of threads, to shift falsely-shared data to different cache lines. Huron's in-house repair technique can generalize to previously-unobserved inputs. Our evaluation shows that Huron can detect more false sharing bugs than all state-of-the-art techniques, and with a lower overhead. Huron improves runtime performance by 3.82× on average (up to 11×), which is 2.11-2.27× better than the state of the art.

Publications

Huron: Hybrid False Detection and Repair
by Tanvir Ahmed Khan, Yifan Zhao, Gilles Pokam, Barzan Mozafari, and Baris Kasikci
In Proceedings of PLDI, 2019.

Dynamic spectrum access through Cognitive Radio Networks (CRNs) and exploiting multiple radios on a single node are two different well accepted techniques for enhancing network performance. Simultaneous usage of both the techniques, i.e., augmenting dynamic spectrum access with multiple radios can improve delay, however, makes throughput worse. Therefore, in this study, we propose a novel approach to improve network throughput for multi-radio cognitive radio networks. Through ns-3 simulation, we show that our approach can boost throughput without degrading the delay.

Publications

Poster: Overcoming Throughput Degradation in Multi-Radio Cognitive Radio Networks
by Tanvir Ahmed Khan and A. B. M. Alim Al Islam
In Proceedings of MobiSys Companion, Singapore, 2016.

In this study, our objective was to reduce the delay CR users experience due to channel sensing, reporting, and negotiation. The main problem of reducing the delay is that we can not reduce any one of these individual components. Therefore, to reduce the delay we had to look at the bigger picture. Our ns-2 simulation results showed us that even though CR users experience these three delays, their delay performance mainly gets degraded due to sudden unavailability of their currently used communication medium. What happens is that when they lost their currently used channel, they have to start the whole process of channel reservation. And this results in longer delay. Therefore, to reduce the delay, we needed to decrease this overhead due to channel switching. We tried to mitigate this overhead by exploiting multiple radios for each user. Our basic intuition was that if we can supply the user with an already negotiated channel when it needs one after losing the previous one, we will be able to reduce the delay. Consequently, we mathematically modeled this multi-radio, multi-channel system with an M/G/k queue and simulated it in ns-2.

Publications

Towards exploiting a synergy between cognitive and multi-radio networking
by Tanvir Ahmed Khan, Chowdhury Sayeed Hyder, and A. B. M. Alim Al Islam
In Proceedings of WiMob, Abu Dhabi, UAE, 2015.