-
Notifications
You must be signed in to change notification settings - Fork 1
Expand file tree
/
Copy pathresearch.html
More file actions
582 lines (579 loc) · 36.7 KB
/
research.html
File metadata and controls
582 lines (579 loc) · 36.7 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>DaSH Lab - Research</title>
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta name="description" content="DaSH Lab, Research, BITS Pilani, K K Birla Goa Campus, India">
<meta name="author" content="">
<link rel="icon" type="image/png" href="thumbs/dashlab_logo_white_bg.png?v=1">
<script>(function(){try{if(localStorage.getItem("dashlab-theme")==="dark"){document.documentElement.classList.add("dark-mode");}}catch(e){}})();</script>
<!-- Le styles -->
<link href="css/bootstrap.min.css" rel="stylesheet">
<link href="css/bootstrap-responsive.min.css" rel="stylesheet">
<link href="css/theme.css?v=3" rel="stylesheet">
<link href="css/dark-mode.css" rel="stylesheet">
<link href="css/scroll.css" rel="stylesheet">
<script type="text/javascript">
(function(c,l,a,r,i,t,y){
c[a]=c[a]||function(){(c[a].q=c[a].q||[]).push(arguments)};
t=l.createElement(r);t.async=1;t.src="https://www.clarity.ms/tag/"+i;
y=l.getElementsByTagName(r)[0];y.parentNode.insertBefore(t,y);
})(window, document, "clarity", "script", "mn8xvap9au");
</script>
</head>
<!-- Google tag (gtag.js) -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-NW9LBZ5NWD"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-NW9LBZ5NWD');
</script>
<body>
<div class="container">
<header class="jumbotron subhead" id="overview">
<h1>Data, Systems and HPC (DaSH) Lab</h1>
<p class="lead"> BITS Pilani K. K. Birla Goa Campus, India </p>
</header>
<div class="masthead">
<div class="navbar">
<div class="navbar-inner">
<div class="container">
<button type="button" class="btn btn-navbar" data-toggle="collapse" data-target=".nav-collapse">
<span class="icon-bar"></span>
<span class="icon-bar"></span>
<span class="icon-bar"></span>
</button>
<a class="brand d-none-md-full" href="#" style="margin-left: 3px;">Navigation</a>
<div class="nav-collapse collapse" style="height: 0; overflow: hidden;">
<ul class="nav">
<li><a href="index.html">Home</a></li>
<li><a href="people.html">People</a></li>
<li class="active"><a href="#">Research</a></li>
<li><a href="publications.html">Publications</a></li>
<li><a href="gallery.html">Gallery</a></li>
<li><a href="collaborations.html">Partners</a></li>
<li><a href="equipment.html">Equipment</a></li>
<li><a href="joinus.html">Join Us</a></li>
<li><a href="contact.html">Contact</a></li>
</ul>
</div>
</div>
</div>
</div>
</div>
<div class="row-fluid">
<div class="span3" id="navparent">
<ul class="nav nav-list bs-docs-sidenav" data-spy="affix" data-offset-top="260" data-offset-bottom="260">
<!---<li><a href="#statemet"> Lorem Ipsum </a></li>--->
<li><a href="#data">Data and I/O</a></li>
<li><a class="subhead" href="#iouring"> I/O Uring </a></li>
<li><a class="subhead" href="#compr"> Data Compression </a></li>
<li><a class="subhead" href="#dataspaces"> DataSpaces</a></li>
<li><a href="#sysml">Systems for ML</a></li>
<li><a class="subhead" href="#privfl"> Privacy-aware FL </a></li>
<li><a class="subhead" href="#multimodalfl"> Multi-Modal FL </a></li>
<li><a class="subhead" href="#flav"> FL for Autonomous Vehicles </a></li>
<li><a class="subhead" href="#potfl"> Proof of Training in FL </a></li>
<li><a class="subhead" href="#mlops"> Operation Pipeline in FL </a></li>
<li><a class="subhead" href="#llmsched"> Scheduling LLM Inference </a></li>
<li><a class="subhead" href="#slmedge"> SLMs on Edge Devices </a></li>
<li><a class="subhead" href="#bcfl"> Cross-Silo FL </a></li>
<li><a class="subhead" href="#tinyml"> Data Caching in TinyML </a></li>
<li><a href="#distributed">Distributed Systems</a></li>
<li><a class="subhead" href="#serverless"> Trust in Serverless </a></li>
<li><a class="subhead" href="#crosscloud"> Cross Cloud Compatibility </a></li>
<li><a class="subhead" href="#smartcontract"> Blockchain Smart Contracts </a></li>
<li><a class="subhead" href="#containerdedup"> Container deduplication </a></li>
<li><a href="#hpc"> High Performance Computing (HPC) </a></li>
<li><a class="subhead" href="#pfs"> Parallel File System </a></li>
<li><a class="subhead" href="#datacache"> Data Caching </a></li>
<li><a class="subhead" href="#resourcesched"> Resource Scheduling </a></li>
<li><a class="subhead" href="#ornl"> Supercomputer Log Analysis </a></li>
</ul>
</div>
<div class="span8 offset1">
<!---<section id="statemet">
<div class="page-header">
<h3>Lorem Ipsum</h3>
<hr>
<p>At vero eos et accusamus et iusto odio dignissimos ducimus qui blanditiis praesentium voluptatum deleniti atque corrupti quos dolores et quas
molestias excepturi sint occaecati cupiditate non provident, similique sunt in culpa qui officia deserunt mollitia animi, id est laborum et
dolorum fuga. Et harum quidem rerum facilis est et expedita distinctio. Nam libero tempore, cum soluta nobis est eligendi optio cumque nihil
impedit quo minus id quod maxime placeat facere possimus, omnis voluptas assumenda est, omnis dolor repellendus. Temporibus autem quibusdam
et aut officiis debitis aut rerum necessitatibus saepe eveniet ut et voluptates repudiandae sint et molestiae non recusandae. Itaque earum
rerum hic tenetur a sapiente delectus, ut aut reiciendis voluptatibus maiores alias consequatur aut perferendis doloribus asperiores repellat.
</p>
</div>
</section>--->
<section id="data">
<div class="page-header">
<h3>Data and I/O</h3>
<p>DaSH Lab work on various data optimization techniques like compression or prefetching on various platforms like scientific workflows or database.</p>
</div>
<div class="row-fluid">
<div class="span12">
<section id="iouring">
<h4>I/O Uring (Ongoing)</h4>
<br/>
<img src="images/iouring.png" class="float-right" style="height: 4cm;" alt="pfs"/>
<p>
<ul>
<li>How can kernel I/O optimizations affect different large-scale use-cases?</li>
<li>Which aspects of key value stores and databases can benefit from I/O uring optimizationsat the kernel level?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Arnav Gupta, Druva Dakshinamoorthy, Arjun Jagtap
<br/>
<b>Few Published Papers: </b><a href="https://ieeexplore.ieee.org/iel8/10740708/10740804/10740888.pdf">Cluster2024</a>
</p>
</section>
<hr>
<section id="compr">
<h4>Data Compression (Completed)</h4>
<br/>
<img src="images/zstd.jpg" class="float-right" style="height: 3cm;" alt="pfs"/>
<p>
<ul>
<li>Can we have a cost optimization model for running data compression algorithms on the cloud vs the end devices?</li>
<li>How do the different optimization parameters of zstd fare against each other?</li>
<li>How do large VM images fare with various levels of zstd?</li>
</ul>
</p>
<p>
<b>Collaborators: </b><a href="https://www.druva.com/">Druva Inc. </a>
<br/>
<b>BITS Pilani Personnel: </b>Pinki Yadav, <a href="https://www.vinayaknaik.info/">Vinayak Naik</a>
<br/>
</p>
</section>
<hr>
<section id="dataspaces">
<h4>DataSpaces (Completed)</h4>
<br/>
<p>
<ul>
<li><a href="https://dataspaces.sci.utah.edu/index.php/home">DataSpaces</a> is a data management framework for scientific workflows.</li>
<li>Can optimizations be done for integrating HPC file systems into dataspaces?</li>
<li>How can I/O intensive tasks be optimized in such a workflow?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>University of Utah - USA
<br/>
<b>BITS Pilani Personnel: </b>Joel Tony
<br/>
<b>Other Personnel: </b><a href="https://www.manishparashar.org/">Manish Parashar</a>
<br/>
</p>
</section>
<hr>
</div>
</div>
</section>
<hr>
<section id="sysml">
<div class="page-header">
<h3>Systems for ML</h3>
<p>DaSH Lab tries to optimize systems and build frameworks for machine learning (ML) workloads. We aim at both large-scale ML workloads as well as tiny ML workloads suitable for resource-constrained edge devices.</p>
</div>
<div class="row-fluid">
<div class="span12">
<section id="privfl">
<h4>Privacy-aware FL (Ongoing)</h4>
<br/>
<p>
<ul>
<li>How can we adaptively tune parameters for different privacy-preserving methods in FL?</li>
<li>How does privacy preserving techniques get impacted by heterogeneity in the FL clusters?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Adithya Balasubramanian, Nischit Kumar, Aishwarya Jayashankar
<br/>
</p>
</section>
<hr>
<section id="multimodalfl">
<h4>Multimodal FL (Ongoing)</h4>
<br/>
<img src="images/blossom.png" class="float-right" style="height: 3cm;" alt="blossom"/>
<p>
<ul>
<li>How can we aggregate FL clusters holding different modes of data?</li>
<li>How can FL aggregation handle missing modalities?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>Queen Mary University of London - UK
<br/>
<b>BITS Pilani Personnel: </b>Pranav M R, Jayant Chandwani
<br/>
<b>Other Personnel: </b><a href="https://eecs.qmul.ac.uk/~ahmed/">Ahmed Sayed</a>
<br/>
</p>
</section>
<hr>
<section id="flav">
<h4>FL for Autonomous Vehicles (Ongoing)</h4>
<br/>
<img src="images/autofl.png" class="float-right" style="height: 3cm;" alt="flops"/>
<p>
<ul>
<li>How can FL be used for vehicles with streaming data?</li>
<li>How the FL techniques adapt to the changing networking environment of vehicles?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Vimarsh Shah, Harsh Chothani, Marichamy Kasi
<br/>
<b>Few Published Papers: </b>HiPC’25 Poster
<br/>
</p>
</section>
<hr>
<section id="potfl">
<h4>Proof of Training in FL (Ongoing)</h4>
<br/>
<p>
<ul>
<li>How can clusters verify that clients actually do the training?</li>
<li>How can cross-silo FL get validation that all participating clusters are training well?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Ojas Marathe, Surya Datta, Abhishek Kumar
<br/>
</p>
</section>
<hr>
<section id="mlops">
<h4>Operation Pipeline in FL (Ongoing)</h4>
<br/>
<img src="images/flops.png" class="float-right" style="height: 3cm;" alt="flops"/>
<p>
<ul>
<li>How to integrate automation of fault tolerance in MLOps pipeline?</li>
<li>Can optimizations be done for deploying FL workloads using FLOps?</li>
<li>Can intelligent Infrastructure-As-Code (IAC) be deployed for MLOps?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>Queen Mary University of London - UK
<br/>
<b>BITS Pilani Personnel: </b>Vimarsh Shah, Aryan Bethmangalkar, Sarang S, Harsh Chothani, Aishwarya Jayashankar, Ayush Bharadwaj
<br/>
<b>Other Personnel: </b><a href="https://eecs.qmul.ac.uk/~ahmed/">Ahmed Sayed</a>
<br/>
<b>Few Published Papers: </b><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10898668">HiPC 2024 Poster</a>, <a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=11262612">ICDCS 2025 Workshop</a>
</p>
</section>
<hr>
<section id="llmsched">
<h4>Scheduling LLM Inference Tasks (Ongoing)</h4>
<br/>
<p>
<ul>
<li>How can you design a scheduler that is resource-aware for a heterogeneous cluster of GPUs while scheduling LLM inference jobs?</li>
<li>Furthermore, can you make the scheduler network-aware?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Aditya Shiva Sharma, Atharva Pandit
<br/>
</p>
</section>
<hr>
<section id="slmedge">
<h4>Small Language Models on Edge Devices (Ongoing)</h4>
<br/>
<p>
<ul>
<li>How can SLMs be resource-aware on edge devices?</li>
<li>Can the SLM tasks be be application aware on mobile devices?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Soham Dambalkar, Utkarsh Sharma, Kunal Mishra
<br/>
</p>
</section>
<hr>
<section id="bcfl">
<h4>Cross-Silo FL (Completed)</h4>
<br/>
<img src="images/bcfl.jpg" class="float-right" style="height: 3cm;" alt="pfs"/>
<p>
<ul>
<li>How can multiple organizations collabate in a data-privacy manner in FL?</li>
<li>Can knowledge distillation process be improved in FL?</li>
<li>Are all edge devices needed to participate in a FL training process?</li>
<li>Can we optimize energy in an FL process?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>Queen Mary University of London - UK, Rochester Institute of Technology- USA
<br/>
<b>BITS Pilani Personnel: </b>Sarang S, Druva Dhakshinamoorthy, Yuvraj Singh Bhadauria, Arihant Bansal, Pinki Yadav, Aditya Shiva Sharma, Sidharth C. Vivek, Vijay Dharmaji, Subroto Majumder, Manit Tanwar
<br/>
<b>Other Personnel: </b><a href="https://eecs.qmul.ac.uk/~ahmed/">Ahmed Sayed</a>, <a href="https://cs.rit.edu/~mrafique/">M. Mustafa Rafique</a>
<br/>
<b>Few Published Papers: </b><a href="https://dl.acm.org/doi/10.1145/3721462.3730955">Middleware 2025</a>, <a href="https://doi.org/10.1109/HiPCW61695.2023.00021">Poster@HiPC23</a>
</p>
</section>
<hr>
<section id="tinyml">
<h4>Data Caching in TinyML (Completed)</h4>
<br/>
<img src="images/lessismore.png" class="float-right" style="height: 3cm;" alt="lessismore"/>
<p>
<ul>
<li>How can distributed tinyML training be made aware of the heterogeneous edge environment?</li>
<li>Can the datasets used for tinyML training be cached given the low resources on the devices?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Advik Raj, Sidharth C. Vivek, Advaith Krishna
<br/>
<b>Few Published Papers: </b><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10884396">HiPC'24</a>
</p>
</section>
<hr>
</div>
</div>
</section>
<hr>
<section id="distributed">
<div class="page-header">
<h3>Distributed Systems</h3>
<p>DaSH Lab works on emerging areas of distributed systems, such as, serverless compute, blockchain, and cloud computing. All projects use state-of-the-art technologies like Kafka, Ethereum, Kubernetes, Docker.</p>
</div>
<div class="row-fluid">
<div class="span12">
<section id="serverless">
<h4>Trust in Serverless</h4>
<br/>
<img src="images/serverless.jpg" class="float-right" style="height: 5cm;" alt="pfs"/>
<p>
<ul>
<li>How do you form a framework for FaaS using personal devices?</li>
<li>How do you enable trust among the geo-distributed personal devices?</li>
<li>How does one schedule FaaS on these devices?</li>
<li>How is the cost of running FaaS provider services calculated?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Aalhad Sawane, Pranav Anand, Satyam Bansal, Pranay Varshney, Paras Mittal, Chinmay Rao, Amogh Balepur
<br/>
<b>Few Published Papers: </b><a href="https://doi.org/10.1109/HiPCW61695.2023.00021">Poster@HiPC23 (Best Undergraduate Poster Award)</a>
</p>
</section>
<hr>
<section id="crosscloud">
<h4>Cross Cloud Compatibility</h4>
<br/>
<!---<img src="images/hvac.jpg" class="float-right" style="height: 7cm;" alt="Quam Id Leo"/>--->
<p>
<ul>
<li>How do you overcome a vendor locking situation with regards to public cloud providers?</li>
<li>Can we have a middleware to have cross compatibility among major cloud providers, such as AWS, Azure and GCP?</li>
</ul>
</p>
</section>
<hr>
<section id="smartcontract">
<h4>Blockchain Smart Contracts</h4>
<br/>
<img src="images/smart_contract.jpg" class="float-right" style="height: 5cm;" alt="sched"/>
<p>
<ul>
<li>How does one remove bugs in smart contracts after deployment?</li>
<li>Are the upgrade mechanisms given by Ethereum usable for all smart contracts?</li>
<li>Which upgrade patterns are most popular and why?</li>
</ul>
</p>
<p>
<b>BITS Pilani Personnel: </b>Aishwarya Parab, Nirmal Govindaraj, Siddhant Kulkarni, Pranay Varshney, Kunal Korgaonkar
<br/>
<b>Few Published Papers: </b><a href="https://doi.org/10.1109/HiPCW61695.2023.00021">Poster@HiPC23</a>
</p>
</section>
<hr>
<section id="containerdedup">
<h4>Container deduplication</h4>
<br/>
<img src="images/container.jpg" class="float-right" style="height: 5cm;" alt="sched"/>
<p>
<ul>
<li>What are the common layers in containers?</li>
<li>Can common layers be deduplicated in large clusters using large amounts of containers?</li>
<li>How can deduplication be solved in edge clusters with low resources?</li>
<li>Can we devise caching methods for caching updates in edge clusters?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>Virginia Tech - USA, Northwestern Polytechnical University - China, Kuwait University, IBM Research - USA, University of Minnesota - USA
<br/>
<b>BITS Pilani Personnel: </b>Naman Agarwal, Prithvi Vishak, Nakul Bhachawat
<br/>
<b>Other Personnel: </b><a href="https://hadeelalbahar.github.io/"> Hadeel Albahar</a>, <a href="https://www.linkedin.com/in/nannan-zhao-882617126/"> Nannan Zhao</a>, <a href="https://cse.umn.edu/cs/ali-anwar">Ali Anwar</a>
<br/>
<b>Few Published Papers: </b><a href="https://dl.acm.org/doi/pdf/10.1145/3643819">ACM TOS24</a>, <a href="https://doi.org/10.1109/TPDS.2020.3034517">TPDS20</a>
</p>
</section>
<hr>
</div>
</div>
</section>
<hr>
<section id="hpc">
<div class="page-header">
<h3>High Performance Computing (HPC)</h3>
<p>DaSH Lab works on the data and file system aspect of HPC. Most emerging HPC applications have complicated I/O patterns which result in sub-optimal application performance.</p>
</div>
<div class="row-fluid">
<div class="span12">
<section id="pfs">
<h4>Parallel File System</h4>
<br/>
<img src="images/pfs.jpg" class="float-right" style="height: 5cm;" alt="pfs"/>
<p>
<ul>
<li>How do you load balance I/O on to storage servers in a seamless manner?</li>
<li>What is an optimal striping layout for a file within an application?</li>
<li>Where to place files in a heterogeneous storage setup?</li>
<li>How does network and I/O interact with each other in a parallel file system?</li>
<li>How does I/O performance differ in parallel file system and object stores?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>Virginia Tech - USA, Johannes Gutenberg University Mainz - Germany
<br/>
<b>BITS Pilani Personnel: </b>Joel Tony, Yash Bhisikar, Bhavya Bajaj, Kaaviya Uthirapandian, Arnav Borkar, Hari Vamsi, Tushar Barman, Sreenath M.
<br/>
<b>Other Personnel: </b>Debasmita Biswas, <a href="https://researchprofiles.uni-mainz.de/19180-sarah-neuwirth">Sarah Neuwirth</a>
<br/>
<b>PFS investigated: </b>Lustre, IBM Spectrum Scale, BeeGFS, Ceph
<br/>
<b>Few Published Papers: </b><a href="https://dl.acm.org/doi/abs/10.1145/3641885">ACM TOS 2024</a>, <a href="https://doi.org/10.1109/CLUSTERWorkshops61457.2023.00010">Cluster 2023 (1)</a>, <a href="https://doi.org/10.1109/CLUSTERWorkshops61457.2023.00014">Cluster 2023 (2)</a>, <a href="https://doi.org/10.1109/INDIS54524.2021.00011">INDIS@SC21</a>, <a href="https://doi.org/10.1109/CLUSTER.2019.8891045">Cluster19</a>, <a href="https://doi.org/10.1109/IPDPS.2019.00070">IPDPS19</a>
</p>
</section>
<hr>
<section id="datacache">
<h4>Data Caching</h4>
<br/>
<img src="images/hvac.jpg" class="float-right" style="height: 7cm;" alt="Quam Id Leo"/>
<p>
<ul>
<li>How to remove I/O bottleneck for large-scale deep learning applications?</li>
<li>How can we use importance sampling for Distributed DL applications?</li>
<li>Can I/O bottleneck be reduced for HPC applications by data prefetching?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>Oak Ridge National Laboratory - USA, University of Viginia - USA, Virginia Tech - USA, Georgia State University
<br/>
<b>Other Personnel: </b><a href="https://www.olcf.ornl.gov/directory/staff-member/ahmad-maroof-karimi/">Ahmad Maroof Karimi</a> , <a href="https://www.linkedin.com/in/redwankhan/"> Redwan Khan</a>, <a href="https://tddg.github.io/">Yue Cheng</a>, <a href="https://www.ornl.gov/staff-profile/jong-youl-choi">Jong Choi</a>, <a href="https://lwan86.github.io/"> Lipeng Wan</a>
<br/>
<b>Few Published Papers: </b><a href="https://www.usenix.org/conference/fast23/presentation/khan">FAST23</a>, <a href="https://doi.org/10.1109/CLUSTER51413.2022.00044">Cluster22</a>
</p>
</section>
<hr>
<section id="resourcesched">
<h4>Resource Scheduling</h4>
<br/>
<img src="images/sched.jpg" class="float-right" style="height: 3cm;" alt="sched"/>
<p>
<ul>
<li>How can distributed deep learning jobs be made GPU heterogenity aware?</li>
<li>How can distributed inferencing jobs be made GPU Heterogeneity aware?</li>
<li>How can containers be used in HPC environments?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>Virginia Tech - USA, Purdue University - USA, IIT Indore, Kuwait University, Northwestern Polytechnical University - China
<br/>
<b>BITS Pilani Personnel: </b>Aditya Shiva Sharma, Kinshuk Goel, Amey Patil, Mukul Krishnan
<br/>
<b>Other Personnel: </b><a href="https://davisjam.github.io/"> James Davis</a>, <a href="http://people.iiti.ac.in/~sidharth/"> Siddharth Sharma</a>, <a href="https://hadeelalbahar.github.io/">Hadeel Albahar</a>, <a href="https://www.linkedin.com/in/nannan-zhao-882617126/"> Nannan Zhao</a>
<br/>
<b>Few Published Papers: </b><a href="https://doi.org/10.1109/CCGrid54584.2022.00079">CCGrid22</a>, <a href="https://doi.org/10.1109/CLOUD49709.2020.00048">Cloud20</a>
</p>
</section>
<hr>
<section id="ornl">
<h4>Supercomputer Log Analysis</h4>
<br/>
<img src="images/darshan.jpg" class="float-right" style="height: 3cm;" alt="sched"/>
<p>
<ul>
<li>How can darshan I/O characterization logs be used to characterize ML workloads?</li>
<li>How can darshan logs be used for HPC trace generation?</li>
<li>How are different supercomputing stacks used by varying HPC workloads?</li>
<li>How do DL workloads differ in academia, industry and national labortories?</li>
</ul>
</p>
<p>
<b>Collaborators: </b>Virginia Tech - USA, Oak Ridge National Laboratory, Argonne National Laboratory, Lawrence Berkeley Laboratory
<br/>
<b>BITS Pilani Personnel: </b>Natasha Meena Joseph, S Sai Vineet, Kunal Korgaonkar, Snehanshu Saha
<br/>
<b>Other Personnel: </b><a href="https://www.linkedin.com/in/ahmad-hossein-yazdani-6560b3138/"> Ahmad Yazdani</a>,<a href="https://www.olcf.ornl.gov/directory/staff-member/ahmad-maroof-karimi/">Ahmad Maroof Karimi</a>, <a href="https://www.ornl.gov/staff-profile/jong-youl-choi">Jong Choi</a>, <a href="https://www.anl.gov/profile/phil-carns">Phil Carns</a>, <a href="https://www.microsoft.com/en-us/research/people/bingxie/">Bing Xie</a>, <a href="https://cse.osu.edu/people/byna.1">Suren Byna</a>, <a href="https://www.ornl.gov/staff-profile/feiyi-wang">Feiyi Wang</a>, <a href="https://crd.lbl.gov/divisions/scidata/sdm/staff/jean-luca-bez/">Jean Luca Bez</a>
<br/>
<b>Few Published Papers: </b><a href="https://doi.org/10.1145/3571306.3571414">ICDCN23</a>, <a href="https://doi.org/10.1109/MASCOTS53633.2021.9614303">MASCOTS21</a>, <a href="https://doi.org/10.1145/3502181.3531461">HPDC22</a>, <a href="https://doi.org/10.1145/3502181.3531457">HPDC22</a>
</p>
</section>
<hr>
</div>
</div>
</section>
<hr>
</div>
</div>
</div>
<footer id="footer">
<div class="container-fluid">
<div class="row-fluid">
<div class="span5">
<h3>Contact Information</h3>
<h5 style="color:white;"><b>Lab Meeting: </b>Tuesdays (6:30 PM - 8:30 PM IST)</h5>
<h5 style="color:white;"><b>PI: </b> <a href = "https://arnabkrpaul.github.io/">Arnab K. Paul</a></h5>
<h5 style="color:white;"><b>Email 1: </b>dashlab.cs@gmail.com</h5>
<h5 style="color:white;"><b>Email 2: </b>arnabp@goa.bits-pilani.ac.in</h5>
<!---<p><b>Cell: </b>999-999-9999</p>
<a href="mailto:your.email@uni.edu">Email</a>--->
</div>
<div class="span2">
<a href="index.html"><img src = "images/FullLogo.png" alt="research-lab-logo"/></a>
</div>
<div class="span5">
<h3 class="">Lab Address</h3>
<h5 style="color:white">D-241, Department of CSIS<br>
BITS Pilani K. K. Birla Goa Campus,<br>
Sancoale, Goa<br>
India - 403726
</h5>
</div>
<!-- <a href="http://maps.google.com/">Show Map</a> -->
</div>
</div>
</div>
</footer>
<a href='#top' class="stt" title="scroll to top"></a>
<!-- Le javascript
================================================== -->
<!-- Placed at the end of the document so the pages load faster -->
<script src="js/jquery-1.9.1.min.js"></script>
<script src="js/bootstrap.min.js"></script>
<script src="js/theme-toggle.js"></script>
<script>
$(document).ready(function() {
$(document.body).scrollspy({
target: "#navparent"
});
});
</script>
</body>
</html>