Murali Annavaram

Murali Annavaram

Auto-generated profile, based on publicly available information

Murali Annavaram

University of Southern California, Los Angeles, CA, USA

Research Focus & Metrics

Academic Career Length

15Years (2012-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 June 2015
Verified
Conference Paper
Closed access
Mohammad Abdel-MajeedMohammad Abdel-Majeed
Mohammad Abdel-Majeed
WDWaleed Dweik
Waleed Dweik
HJHyeran Jeon
Hyeran Jeon
MAMurali Annavaram
Murali Annavaram
Graphics processing units (GPUs) are now the dominant computing fabric within many supercomputers. As such many mission critical applications run on GPUs, which demand stringent reliability and computational correctness guarantees from GPUs. Prior approaches to GPU reliability have tackled solely either error detection, or error correction assuming error detection is already present. In this paper we present Warped Redundant Execution (Warped-RE), a unified framework that is capable of detecting and then correcting transient and nontransient errors in the GPU execution lanes. Our work exploits two critical properties of applications running on GPUs. First, we observe that neighboring execution lanes in GPUs may operate on the same values. Thus when neighboring lanes execute the same instruction using the same values then these lanes provide inherent DMR (dual modular redundancy) or even inherent TMR (triple modular redundancy) opportunities. The second property we exploit is that due to insufficient parallelism or due to branch divergence, applications do not fully utilize all the available execution lanes. In this case it is possible to force DMR or TMR on unused execution lanes, when inherent redundancy is insufficient. During error-free execution, Warped-RE uses a combination of inherent and forced DMR to guarantee that every thread computation within every warp instruction will be verified. When an error is detected in a warp instruction, the instruction is re-executed in TMR mode in order to correct the error and identify execution lanes with potential non-transient errors. Our evaluations show 8.4% and 29% average performance overhead during the DMR and TMR operation modes, respectively. Compared to traditional DMR and TMR, Warped-RE reduces the power overhead by 42% and 40%, respectively.

1 September 2021
Verified
Conference Paper
Closed access
KNKrishna Giri Narra
Krishna Giri Narra
ZLZhifeng Lin
Zhifeng Lin
YWYongqin Wang
Yongqin Wang
KBKeshav Balasubramanian
Keshav Balasubramanian
MAMurali Annavaram
Murali Annavaram
This work presents Origami, a framework which provides privacy-preserving inference for large deep neural network (DNN) models through a combination of enclave execution, cryptographic blinding, interspersed with accelerator-based computation. Origami partitions the ML model into multiple partitions. The first partition receives the encrypted user input within an SGX enclave. The enclave decrypts the input and then applies cryptographic blinding to the input data and the model parameters. The layer computation is offloaded to a GPU/CPU and the computed output is returned to the enclave, which decodes the computation on noisy data using the unblinding factors privately stored within SGX. This process may be repeated for each DNN layer, as has been done in prior work Slalom. However, the overhead of blinding and unblinding the data is a limiting factor to scalability. Origami relies on the empirical observation that the feature maps after the first several layers can not be used, even by a powerful conditional GAN adversary to reconstruct input. Hence, Origami dynamically switches to executing the rest of the DNN layers directly on an accelerator. We empirically demonstrate that using Origami, a conditional GAN adversary, even with an unlimited inference budget, cannot reconstruct the input. Compared to running the entire VGG-19 model within SGX, Origami inference improves the performance of private inference from 11x while using Slalom to 15. 1x.

17 June 2023
Verified
Conference Paper
Open access
RRRachit Rajat
Rachit Rajat
YWYongqin Wang
Yongqin Wang
MAMurali Annavaram
Murali Annavaram
Memory access patterns have been demonstrated to leak critical information such as security keys and a program's spatial and temporal information. This information leak poses a significant privacy challenge in machine learning models with embedding tables. Embedding tables are used to learn categorical features from training data. The address of an embedding table entry carries privacy sensitive information since the address of an entry discloses features associated with a user. Oblivious RAM (ORAM), and its enhanced variants, such as PathORAM, have emerged as viable solutions to hide leakage from memory access streams. PathORAM fetches an entire path of memory blocks for every memory fetch request, thereby leading to substantial bandwidth and performance overheads. In this work, we present Look Ahead ORAM (LAORAM), an ORAM framework designed to protect user privacy during embedding table training. LAORAM exploits the unique property of ML training, namely the training samples that are going to be used in the future are known beforehand. LAORAM preprocesses the training samples to identify the memory blocks which are accessed together in the near future. LAORAM combines multiple blocks accessed together as superblocks and tries to assign all blocks in a superblock to few paths. Thus, future accesses to a collection of blocks can be satisfied from a few paths, effectively reducing the number of reads and writes to the ORAM. To further increase performance, LAORAM uses a fat-tree structure for PathORAM, i.e. a tree with variable bucket size, effectively reducing the number of background evictions required, which improves the stash usage. We have evaluated LAORAM using both a recommendation model (DLRM) and an NLP model (XLM-R) embedding table configurations. LAORAM performs 5 times faster than PathORAM on a recommendation dataset (Kaggle) and 5.4 times faster on an NLP dataset (XNLI) while guaranteeing the same security guarantees as the original PathORAM.

1 November 2012
Verified
Conference Paper
Closed access
SLSang-Won Lee
Sang-Won Lee
MAMurali Annavaram
Murali Annavaram
Wireless Body Area Networks (WBANs) promise to revolutionize health care in the near future. By integrating bio-sensors with a mobile phone it is possible to monitor an individual's health and related behaviors. Monitoring is done by analyzing the sensor data either on a mobile phone or on a remote server by relaying this information over a wireless network. However, the “wireless” aspect of WBAN is being limited by the battery life of the mobile phone. A WBAN designer has a range of options to trade-off limited battery with many important metrics. From the choice of programming languages to dynamically choosing between computation versus communication under varying signal strengths, there are several non-obvious choices that can have dramatic impact on battery life. In this research we use an in-field deployed WBAN called KNOWME to present a comprehensive quantification of a mobile phone's energy consumption. We quantify the energy impact of different programming paradigms, sensing modalities, data storage, and conflicting computation and communication demands. Based on the knowledge gained from the measurement studies, we propose an Active Energy Profiling strategy that uses short profiling periods to automatically determine the most energy efficient choices for running a WBAN.

8 October 2024
Verified
Conference Paper
Open access
KBKeshav Balasubramanian
Keshav Balasubramanian
AAAbdulla Alshabanah
Abdulla Alshabanah
EMElan Markowitz
Elan Markowitz
Greg Ver SteegGreg Ver Steeg
Greg Ver Steeg
MAMurali Annavaram
Murali Annavaram
Recommendation systems connect users to items and create value chains in the internet economy. Recommendation systems learn from past user-item interaction histories. As such, items that have short interaction histories, either because they are new or not popular, have been shown to be disproportionately under-recommended. This long-tail item problem can exacerbate model bias, and reinforce poor recommendation of tail items. In this paper, we propose biased user history synthesis, to not only address this problem but also achieve better personalization in recommendation systems. As a result, we concurrently improve tail and head item recommendation performance. Our approach is built on a tail item biased User Interaction History (UIH) sampling strategy and a synthesis model that produces an augmented user representation from the sampled user history. We provide a theoretical justification for our approach using information theory and demonstrate through extensive experimentation, that our model outperforms state-of-the-art baselines on tail, head, and overall recommendation. The source code is available at https://github.com/lkp411/BiasedUserHistorySynthesis.

14 September 2024 in ACM Transactions on Architecture and Code Optimization
Verified
Journal Article
Open access
HFHajar Falahati
Hajar Falahati
MSMohammad Sadrosadati
Mohammad Sadrosadati
QXQiumin Xu
Qiumin Xu
JLJuan Gómez Luna
Juan Gómez Luna
BLBanafsheh Saber Latibari
Banafsheh Saber Latibari
MPMasoud Pedram
Masoud Pedram
Graphics Processing Units (GPUs) are the accelerator of choice in a variety of application domains, because they can accelerate massively parallel workloads and can be easily programmed using general-purpose programming frameworks such as CUDA and OpenCL. Each Streaming Multiprocessor (SM) contains an L1 data cache (L1D) to exploit the locality in data accesses. L1D misses are costly for GPUs for two reasons. First, L1D misses consume a lot of energy as they need to access the L2 cache (L2) via an on-chip network and the off-chip DRAM in case of L2 misses. Second, L1D misses impose performance overhead if the GPU does not have enough active warps to hide the long memory access latency. We observe that threads running on different SMs share 55% of the data they read from the memory. Unfortunately, as the L1Ds are in the non-coherent memory domain, each SM independently fetches data from the L2 or the off-chip memory into its L1D, even though the data may be currently available in the L1D of another SM. Our goal is to service L1D read misses via other SMs, as much as possible, to cut down costly accesses to the L2 or the off-chip DRAM. To this end, we propose a new data-sharing mechanism, called Cross-Core Data Sharing (CCDS) . CCDS employs a predictor to estimate whether the required cache block exists in another SM. If the block is predicted to exist in another SM’s L1D, then CCDS fetches the data from the L1D that contain the block. Our experiments on a suite of 26 workloads show that CCDS improves average energy and performance by 1.30× and 1.20×, respectively, compared to the baseline GPU. Compared to the state-of-the-art data-sharing mechanism, CCDS improves average energy and performance by 1.37× and 1.11×, respectively.