-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
759 lines (724 loc) · 38.4 KB
/
index.html
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
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
---
title: Home
---
<!DOCTYPE HTML>
<!-- GH
Linear by TEMPLATED
templated.co @templatedco
Released for free under the Creative Commons Attribution 3.0 license (templated.co/license)
-->
<html>
<head>
<title>{{ site.title }}: {{ page.title }}</title>
{% include header.html %}
</head>
<body class="homepage">
<!-- Header -->
<div id="header">
{% include menu.html %}
<div class="container">
<!-- Logo -->
<div id="logo">
<h1><a href="#"><span class="lift-light">Lift</span></a></h1>
<span class="tag">Where High-Level Programming Meets Performance Portability</span>
</div>
</div>
</div>
<!-- Featured -->
<div id="featured">
<div class="container">
<header>
<h2>Welcome to Lift</h2>
</header>
<p><span class="lift-bold">Lift</span> is a novel approach to achieving performance portability on parallel accelerators.
<span class="lift-bold">Lift</span> combines a high-level functional data parallel language with a system of rewrite rules which encode algorithmic and hardware-specific optimisation choices.
Applications written in <span class="lift-bold">Lift</span> are able to take advantage of GPUs (and in the future other accelerators), transparently from the user.</p>
</p>
</div>
</div>
<!-- Main -->
<!--
<div id="example">
<div id="content" class="container">
<header>
<h2>Example</h2>
</header>
<p>
Matrix multiplication high-level code example
Plots of results on different machines
</p>
</div>
</div>
-->
<!-- Positions -->
<div id="positions">
<div id="content" class="container">
<header>
<h2>Open Positions and Collaborations</h2>
</header>
<p>
We are always looking for bright and enthusiastic researchers to join our team.
We have funding opportunities for PhD studies and visits/internships both at the University of Edinburgh
(<a href="mailto:[email protected]">[email protected]</a>) and the University of Glasgow
(<a href="mailto:[email protected]">[email protected]</a>).
Please feel free to contact us if you are interested.
</p>
<!-- p>
You can also find below some specific positions currently available:
<ul>
<li>Fully funded <b>PhD position</b> at Edinburgh University: <a href="phdmsr">High-Level Synthesis of Neural Networks for FPGAs with LIFT</a></li>
</ul>
</p -->
</div>
</div>
<!-- Events -->
<div id="events">
<div id="content" class="container">
<header>
<h2>Past events</h2>
</header>
<div style="display: flex; padding-bottom: -100px">
<section class="1u" style="margin-left: 3%; padding-left: 0; text-align: right; margin-bottom: 0">
<a href="http://www.ispass.org/ispass2018/">
<img src="images/ispass.png" width="51" height="175" alt="ISPASS 2018" >
</a>
</section>
<section class="10u" style="padding-left: 1em; padding-top: 0.3em; text-align: left;">
<header>
<h3><a href="http://www.ispass.org/ispass2018/">ISPASS-2018</a> tutorial</h3>
<span class="subtitle">April 2, 2018 @ Belfast, Northern Ireland, United Kingdom</span>
</header>
We brought <span class="lift-bold">Lift</span> to ISPASS with this tutorial titled <i>"<span class="lift-bold">Lift</span> — performance portable code generation on parallel accelerators"</i>. We went through the fundamental components of <span class="lift-bold">Lift</span>: language primitives and their properties, <span class="lift-bold">Lift</span> IR, type system, parallelism exploitation and memory management. We dived deep into rewrite rules and the arithmetic expression simplifier behind <span class="lift-bold">Lift</span>; we covered the existing applications of the language along with some practical examples for those interested to try them out.
</section>
</div>
<center style="margin-top: 20px">
<span class="button_wrapper" style="align: center;">
<a href="ispass2018" class="button button-style1">Read More</a>
</span>
</center>
</div>
</div>
<!-- Source Code -->
<div id="code">
<div id="content" class="container">
<header>
<h2>Source Code</h2>
</header>
<p>
The <span class="lift-bold">Lift</span> source code is available at <a href="https://github.com/lift-project/lift"><i class="fa fa-github fa-2x" aria-hidden="true"></i> https://github.com/lift-project/lift</a>.<br>
Lift is open source software released under the permissive MIT license.
</p>
</div>
</div>
<!-- Main -->
<div id="pubs">
<div id="content" class="container">
<header>
<h2>Publications</h2>
</header>
<ul class="10u -1u">
<li>
<a href="https://naumsmogers.me/assets/other/CC22_liftpar.pdf">
Mapping Parallelism in a Functional IR through Constraint Satisfaction: A Case Study on Convolution for Mobile GPUs
</a><br>
Naums Mogers, Lu Li, Valentin Radu, and Christophe Dubach<br>
<a href="https://conf.researchr.org/home/CC-2022">
ACM SIGPLAN 2022 International Conference on Compiler Construction
</a><br>
<a href="https://naumsmogers.me/assets/other/CC22_liftpar.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper
</a>
<a href="https://gitlab.com/naummo/liftpar-cc-2022-artifact" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-bar-chart fa-stack-1x fa-flip-horizontal" style="left:-.1em;top:-.1em"></i>
<span class="fa fa-stack-2x">
<i class="fa fa-search" aria-hidden="true"></i>
</span>
</span>
Artifact</a>
<a href="https://naumsmogers.me/assets/other/CC22_talk.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-picture-o fa-stack-1x fa-inverse"></i>
</span>
Talk slides</a>
<a href="https://www.youtube.com/watch?v=2LsQZoWTDQk" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-youtube fa-stack-1x fa-inverse"></i>
</span>
Video</a>
</li>
<li>
<a href="publications/2020/mogers2020directconvonmobile.pdf">
Automatic Generation of Specialized Direct Convolutions for Mobile GPUs
</a><br>
Naums Mogers, Valentin Radu, Lu Li, Jack Turner, Michael O'Boyle, and Christophe Dubach<br>
<a href="https://insight-archlab.github.io/gpgpu.html">
General Purpose Processing Using GPU (GPGPU) 2020
</a><br>
<a href="publications/2020/mogers2020directconvonmobile.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper
</a>
</li>
<li>
<a href="publications/2019/pizzuti19positiondependentarrays.pdf">
Position-Dependent Arrays and Their Application for High Performance Code Generation
</a><br>
Federico Pizzuti, Michel Steuwer, and Christophe Dubach<br>
<a href="https://icfp19.sigplan.org/home/FHPNC-2019">
Functional High-Performance and Numerical Computing (FHPNC) 2019
</a><br>
<a href="publications/2019/pizzuti19positiondependentarrays.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper
</a>
</li>
<li>
<a href="publications/2019/kristien2019hlsfunctionalpatterns.pdf">
High-Level Synthesis of Functional Patterns with Lift
</a><br>
Martin Kristien, Bruno Bodin, Michel Steuwer, and Christophe Dubach<br>
<a href="https://pldi19.sigplan.org/home/ARRAY-2019#program">
6th ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming (ARRAY) 2019
</a><br>
<a href="publications/2019/kristien2019hlsfunctionalpatterns.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper
</a>
</li>
<li>
<a href="publications/2019/mogers2019dnnaccelerators.pdf">
Towards Mapping Lift to Deep Neural Network Accelerators
</a><br>
Naums Mogers, Aaron Smith, Dimitrios Vytiniotis, Michel Steuwer, Christophe Dubach, Ryota Tomioka<br>
<a href="https://www.hipeac.net/2019/valencia/#/schedule/sessions/7627/">
Workshop on Emerging Deep Learning Accelerators (EDLA) 2019 @ HiPEAC
</a><br>
<a href="publications/2019/mogers2019dnnaccelerators.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper</a>
<a href="presentations/2019/EDLA-2019.pdf" style="margin-right: 1em;">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-picture-o fa-stack-1x fa-inverse"></i>
</span>
Slides of the talk</a>
<a href="posters/2019/MogersHiPEAC2019Poster.pdf" style="margin-right: 1em;">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-picture-o fa-stack-1x fa-inverse"></i>
</span>
Poster</a>
</li>
<li>
<a href="publications/2018/hagedorn18Stencils.pdf">
High Performance Stencil Code Generation with Lift
</a><br>
Bastian Hagedorn, Larisa Stoltzfus, Michel Steuwer, Sergei Gorlatch, Christophe Dubach<br>
<a href="http://cgo.org/cgo2018/">
International Symposium on Code Generation and Optimization (CGO) 2018
</a><br>
<a href="publications/2018/hagedorn18Stencils.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper</a>
<a href="presentations/2018/CGO-2018.pdf" style="margin-right: 1em;">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-picture-o fa-stack-1x fa-inverse"></i>
</span>
Slides of the talk</a>
<a href="https://gitlab.com/larisa.stoltzfus/liftstencil-cgo2018-artifact" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-bar-chart fa-stack-1x fa-flip-horizontal" style="left:-.1em;top:-.1em"></i>
<span class="fa fa-stack-2x">
<i class="fa fa-search" aria-hidden="true"></i>
</span>
</span>
Software Artifact for independent evaluation of the performance results</a>
</li>
<li>
<a href="publications/2017/steuwer17LiftIR.pdf">
Lift: A Functional Data-Parallel IR for High-Performance GPU Code Generation
</a><br>
Michel Steuwer, Toomas Remmelg, and Christophe Dubach<br>
<a href="http://cgo.org/cgo2017/">
International Symposium on Code Generation and Optimization (CGO) 2017
</a><br>
<a href="publications/2017/steuwer17LiftIR.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper</a>
<a href="presentations/2017/CGO-2017.pdf" style="margin-right: 1em;">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-picture-o fa-stack-1x fa-inverse"></i>
</span>
Slides of the talk</a>
<a href="https://gitlab.com/michel-steuwer/cgo_2017_artifact" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-bar-chart fa-stack-1x fa-flip-horizontal" style="left:-.1em;top:-.1em"></i>
<span class="fa fa-stack-2x">
<i class="fa fa-search" aria-hidden="true"></i>
</span>
</span>
Software Artifact for independent evaluation of the performance results</a>
</li>
<li>
<a href="publications/2016/steuwer16beyondAutoTuning.pdf">
Matrix Multiplication Beyond Auto-Tuning: Rewrite-based GPU Code Generation
</a><br>
Michel Steuwer, Toomas Remmelg, and Christophe Dubach<br>
<a href="http://www.esweek.org/cases">
International Conference on Compilers, Architecture and Synthesis for Embedded Systems (CASES) 2016
</a><br>
<a href="publications/2016/steuwer16beyondAutoTuning.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper</a>
<a href="presentations/2016/CASES-2016.pdf" style="margin-right: 1em;">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-picture-o fa-stack-1x fa-inverse"></i>
</span>
Slides of the talk</a>
</li>
<li>
<a href="publications/2016/remmelg16perfport.pdf">
Performance Portable GPU Code Generation for Matrix Multiplication
</a><br>
Toomas Remmelg, Thibaut Lutz, Michel Steuwer, and Christophe Dubach<br>
<a href="http://conf.researchr.org/track/PPoPP-2016/GPGPU-2016-papers">
Workshop on General Purpose Processing using GPUs (GPGPU) 2016 @ PPoPP
</a><br>
<a href="publications/2016/remmelg16perfport.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper</a>
<a href="presentations/2016/GPGPU-2016.pdf" style="margin-right: 1em;">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-picture-o fa-stack-1x fa-inverse"></i>
</span>
Slides of the talk</a>
</li>
<li>
<a href="publications/2016/harries16sparse.pdf">
Compositional Compilation for Sparse, Irregular Data Parallelism
</a><br>
Adam Harries, Michel Steuwer, Murray Cole, Alan Gray, and Christophe Dubach<br>
<a href="https://sites.google.com/site/hlpgpu2016/">
Workshop on High-Level Programming for Heterogeneous and Hierarchical Parallel Systems (HLPGPU) 2016 @ HiPEAC
</a><br>
<a href="publications/2016/harries16sparse.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper</a>
</li>
<li>
<a href="publications/2015/steuwer15generating.pdf">
Generating Performance Portable Code using Rewrite Rules: From High-Level Functional Expressions to High-Performance OpenCL Code
</a><br>
Michel Steuwer, Christian Fensch, Sam Lindley, and Christophe Dubach<br>
<a href="http://www.icfpconference.org/icfp2015/index.html">
International Conference on Functional Programming (ICFP) 2015
</a><br>
<a href="publications/2015/steuwer15generating.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Paper</a>
<a href="presentations/2015/ICFP-2015.pdf" style="margin-right: 1em;">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<i style="margin-top:5px"
class="fa fa-picture-o fa-stack-1x fa-inverse"></i>
</span>
Slides of the talk</a>
<a href="https://www.youtube.com/watch?v=xNYMIsGTXC4" style="margin-right:1em">
<i style="vertical-align: middle;" class="fa fa-youtube-play fa-2x"></i>
Video of the talk</a>
</li>
<li>
<a href="publications/2015/steuwer15arxiv.pdf">
Patterns and Rewrite Rules for Systematic Code Generation (From High-Level Functional Patterns to High-Performance Opencl Code)
</a><br>
Michel Steuwer, Christian Fensch, and Christophe Dubach<br>
Technical Report <a href="http://arxiv.org/abs/1502.02389">arXiv:1502.02389</a><br>
<a href="publications/2015/steuwer15arxiv.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
Technical Report</a>
</li>
<li>
<a href="publications/2015/steuwer15phdthesis.pdf">
Improving Programmability and Performance Portability
</a><br>
Michel Steuwer<br>
PhD Thesis, University of Münster<br>
<a href="publications/2015/steuwer15phdthesis.pdf" style="margin-right:1em">
<span class="fa-stack fa-1x">
<i class="fa fa-file fa-stack-2x fa-flip-horizontal"></i>
<span class="fa fa-stack-1x" style="color:white;">
<span style="font-size:12px; margin-top:5px; display:block;">
PDF
</span>
</span>
</span>
PhD Thesis</a>
</li>
</ul>
</div>
</div>
<div id="research">
<div id="content" class="container">
<header>
<h2>Research directions</h2>
</header>
<div class="row">
<section class="4u">
<span class="pennant"><a href="algebra"><span class="fa fa-clone"></span></a></span>
<h3>Linear Algebra</h3>
<p>Starting from a single high-level program, our compiler automatically generates highly optimized
and specialized matrix multiplication implementations.
We group simple rewrite rules into more complex macro-rules, each describing a well-known optimization
like tiling and register blocking in a composable way. Using
an exploration strategy our compiler automatically generates
50,000 OpenCL kernels, each providing a differently
optimized — but provably correct — implementation of matrix
multiplication.</p>
<div class="button_wrapper">
<a href="algebra" class="button button-style1">Read More</a>
</div>
</section>
<section class="4u">
<span class="pennant"><a href="ml"><span class="fa fa-connectdevelop"></span></a></span>
<h3>Machine Learning</h3>
<p>Making neural networks (NNs) performance-portable using <span class="lift-bold">Lift</span>:
implementing generic and NN-specific optimizations as rewrite rules for efficient hardware utilization;
introducing traditional NN building blocks such as <i>conv</i>, <i>norm</i> and <i>fully_connected</i>
for seamless integration of <span class="lift-bold">Lift</span> with popular machine learning libraries.</p>
<div class="button_wrapper">
<a href="ml" class="button button-style1">Read More</a>
</div>
</section>
<section class="4u">
<span class="pennant"><a href="reducts_scans"><span class="fa fa-cubes"></span></a></span>
<h3>Optimising Reductions and Scans</h3>
<p>Using <span class="lift-bold">Lift</span> to create a high-level programming environment for
Heterogeneous Computation, freeing the programmer from the burden of having to write complex
device-specific code. This makes it possible to automatically identify parallelism in
non-associative reductions and scans, hence enabling generation of efficient parallel implementations for
GPUs.</p>
<div class="button_wrapper">
<a href="reducts_scans" class="button button-style1">Read More</a>
</div>
</section>
</div>
<div class="row">
<section class="4u">
<span class="pennant"><a href="sparse_par"><span class="fa fa-barcode"></span></a></span>
<h3>Sparse Data Parallelism</h3>
<p>We demonstrate that
high level programming and high performance GPU execution for
sparse, irregular problems are not mutually exclusive. Our insight
is that this can be achieved by capturing sparsity and irregularity
friendly implementations within the target space of a pattern-oriented,
high-level compilation and transformation system. By
working in a language rather than a library, we benefit from the
ability to generate implementations by program-specific composition
of building blocks which capture detailed, low-level implementation
choices.</p>
<div class="button_wrapper">
<a href="sparse_par" class="button button-style1">Read More</a>
</div>
</section>
<section class="4u">
<span class="pennant"><a href="waves"><span class="fa fa-rss"></span></a></span>
<h3>3D Wave Modelling</h3>
<p>Simplified room acoustics simulations have been thoroughly investigated in
<span class="lift-bold">Lift</span> and a number of other 2D and 3D benchmarks have also been implemented.
Current work also involves developing and formalising stencil optimisations for 3D codes,
in particular 2.5D tiling, as well as ground penetrating radar algorithms.
How to best abstract
out absorbing boundary conditions needs to be investigated in more detail and primitives to
accommodate these conditions need to be designed and added. Finally, a stencil-based DSL needs to
be extended to compile into the <span class="lift-bold">Lift</span> language.</p>
<div class="button_wrapper">
<a href="waves" class="button button-style1">Read More</a>
</div>
</section>
<section class="4u">
<span class="pennant"><a href="stencils"><span class="fa fa-arrows"></span></a></span>
<h3>Stencil Computations</h3>
<p>Stencil computations are used in a wide range of
applications from physical simulations to machine learning.
Optimizing and tuning them for parallel
hardware remains challenging.
<span class="lift-bold">Lift</span> is a new approach to achieving performance
portability based on a small set of reusable
parallel primitives. Its key novelty is encoding of
optimization as a system of rewrite rules which are
used to explore the optimization space.
We extend <span class="lift-bold">Lift</span> with support for stencil computations
by adding a small number of primitives together
with a few rewrite rules to achieve performance
portability for stencil computations.</p>
<div class="button_wrapper">
<a href="stencils" class="button button-style1">Read More</a>
</div>
</section>
</div>
<div class="row">
<section class="4u">
<span class="pennant"><a href="fpga"><span class="fa fa-microchip"></span></a></span>
<h3>High-Level Synthesis<br/>For FPGAs</h3>
<p>FPGAs are highly energy-efficient and offer a lot of flexibility to implement an application.
With <span class="lift-bold">Lift</span>, we want to exploit the FPGA's characteristics to lift the performance and energy-efficiency of neural network tasks.
Starting from a high-level functional specification, the <span class="lift-bold">Lift</span> compiler applies rewrite rules to explore different implementations and finally generates a low-level, optimised hardware design for FPGAs.
</p>
<a href="fpga" class="button button-style1">Read More</a>
</section>
</div>
<!--
<div class="row">
<section class="4u">
NOTE: find your icon here: http://fontawesome.io/icons/
<span class="pennant"><a href="YOUR_PROJECT_PAGE"><span class="fa YOUR_ICON_CODE"></span></a></span>
<h3>YOUR_PROJECT_TITLE_SHORT</h3>
<p>YOUR_PROJECT_DESCRIPTION</p>
<a href="YOUR_PROJECT_PAGE.html" class="button button-style1">Read More</a>
</section>
<section class="4u">
NOTE: find your icon here: http://fontawesome.io/icons/
<span class="pennant"><a href="YOUR_PROJECT_PAGE"><span class="fa YOUR_ICON_CODE"></span></a></span>
<h3>YOUR_PROJECT_TITLE_SHORT</h3>
<p>YOUR_PROJECT_DESCRIPTION</p>
<a href="YOUR_PROJECT_PAGE.html" class="button button-style1">Read More</a>
</section>
<section class="4u">
NOTE: find your icon here: http://fontawesome.io/icons/
<span class="pennant"><a href="YOUR_PROJECT_PAGE"><span class="fa YOUR_ICON_CODE"></span></a></span>
<h3>YOUR_PROJECT_TITLE_SHORT</h3>
<p>YOUR_PROJECT_DESCRIPTION</p>
<a href="YOUR_PROJECT_PAGE.html" class="button button-style1">Read More</a>
</section>
</div>
-->
</div>
</div>
<div id="team">
<div id="content" class="container">
<header>
<h2>Meet the Lifters</h2>
</header>
<div class="row">
<section class="2u" style="margin-left: 8%">
<a href="http://homepages.inf.ed.ac.uk/cdubach/" class="image full"><img src="images/cdubach.png" alt=""></a>
<a href="http://homepages.inf.ed.ac.uk/cdubach/">Christophe Dubach</a>
<br>Reader<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
<section class="2u">
<a href="https://michel.steuwer.info" class="image full"><img src="images/msteuwer.jpg" alt=""></a>
<a href="https://michel.steuwer.info">Michel Steuwer</a>
<br>Lecturer<br><a href="http://www.gla.ac.uk/schools/computing/">University of Glasgow</a>
</section>
<section class="2u">
<a href="http://www.inf.ed.ac.uk/people/staff/Lu_Li.html" class="image full"><img src="images/lu.jpg" alt=""></a>
<a href="http://www.inf.ed.ac.uk/people/staff/Lu_Li.html">Lu Li</a>
<br><nobr>Postdoctoral Researcher</nobr><br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
<section class="2u">
<a href="https://www.inf.ed.ac.uk/people/students/Toomas_Remmelg.html" class="image full"><img src="images/toomas.jpg" alt=""></a>
<a href="https://www.inf.ed.ac.uk/people/students/Toomas_Remmelg.html">Toomas Remmelg</a>
<br>PhD Student<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
<section class="2u">
<a href="https://bastianhagedorn.github.io/" class="image full"><img src="images/bastian.jpg" alt=""></a>
<a href="https://bastianhagedorn.github.io/">Bastian Hagedorn</a>
<br>PhD Student<br><a href="https://www.uni-muenster.de/Informatik/en/index.shtml">University of Münster</a>
</section>
</div>
<div class="row" style="margin-top:2em;">
<section class="2u" style="margin-left: 8%">
<a href="http://homepages.inf.ed.ac.uk/s1147290/index.html" class="image full"><img src="images/larisa.jpg" alt=""></a>
<a href="http://homepages.inf.ed.ac.uk/s1147290/index.html">Larisa Stoltzfus</a>
<br>PhD Student<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
<section class="2u">
<a href="http://www.inf.ed.ac.uk/people/students/Federico_Pizzuti.html" class="image full"><img src="images/federico.jpg" alt=""></a>
<a href="http://www.inf.ed.ac.uk/people/students/Federico_Pizzuti.html">Federico Pizzuti</a>
<br>PhD Student<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
<section class="2u">
<a href="https://naumsmogers.me/" class="image full"><img src="images/naums.jpg" alt=""></a>
<a href="https://naumsmogers.me/">Naums Mogers</a>
<br>PhD Student<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
<section class="2u">
<a href="http://homepages.inf.ed.ac.uk/s1210443/" class="image full"><img src="images/andrej.jpg" alt=""></a>
<a href="http://homepages.inf.ed.ac.uk/s1210443/">Andrej Ivanis</a>
<br>PhD Student<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
<section class="2u">
<a href="http://homepages.inf.ed.ac.uk/s1894023/" class="image full"><img src="images/christof.jpg" alt=""></a>
<a href="http://homepages.inf.ed.ac.uk/s1894023/">Christof Schlaak</a>
<br>PhD Student<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
</div>
<div class="row" style="margin-top:2em;">
<section class="2u" style="margin-left: 8%">
<a href="https://www.uni-muenster.de/PVS/mitarbeiter/koepcke.html" class="image full"><img src="images/bastian_koepcke.png" alt="Picture of Bastian Koepcke"></a>
<a href="https://www.uni-muenster.de/PVS/mitarbeiter/koepcke.html">Bastian Köpcke</a>
<br>PhD Student<br><a href="https://www.uni-muenster.de/Informatik/en/index.shtml">University of Münster</a>
</section>
<section class="2u">
<a href="https://www.gla.ac.uk/schools/computing/researchstudents/thomaskoehler/" class="image full"><img src="images/thomas.jpg" alt="Picture of Thomas Koehler"></a>
<a href="https://www.gla.ac.uk/schools/computing/researchstudents/thomaskoehler/">Thomas Koehler</a>
<br>PhD Student<br><a href="http://www.gla.ac.uk/schools/computing/">University of Glasgow</a>
</section>
</div>
<header>
<h2>Lift Alumni</h2>
</header>
<div class="row" style="margin-top:2em;">
<section class="2u" style="margin-left: 8%">
<a href="http://homepages.inf.ed.ac.uk/s1467120/" class="image full"><img src="images/adam.jpg" alt=""></a>
<a href="http://homepages.inf.ed.ac.uk/s1467120/">Adam Harries</a>
<br>PhD Student<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
<section class="2u">
<a href="http://homepages.inf.ed.ac.uk/s1343145/" class="image full"><img src="images/martin.jpg" alt=""></a>
<a href="http://homepages.inf.ed.ac.uk/s1343145/">Martin Kristien</a>
<br>PhD Student<br><a href="http://www.ed.ac.uk/informatics/">University of Edinburgh</a>
</section>
</div>
</div>
</div>
<div id="support">
<div id="content" class="container">
<header>
<h2>Support</h2>
</header>
This project is partially supported by:<br/>
<ul>
<li><a href="http://research.google.com/research-outreach.html#/research-outreach/faculty-engagement/faculty-research-award-recipients"><b>Google</b> Faculty Research Award</a></li>
<li><a href="http://www.huawei.com/en/hirp"><b>Huawei</b> Innovation Research Program FLAGSHIP (HIRP FLAGSHIP)</a></li>
</ul>
</div>
</div>
<!--
<div id="tweet">
<div class="container">
<section>
<blockquote>“In posuere eleifend odio. Quisque semper augue mattis wisi. Maecenas ligula. Pellentesque viverra vulputate enim. Aliquam erat volutpat.”</blockquote>
</section>
</div>
</div>-->
<!-- Footer -->
<!--
<div id="footer">
<div class="container">
<section>
<header>
<h2>Get in touch</h2>
<span class="byline">Integer sit amet pede vel arcu aliquet pretium</span>
</header>
<ul class="contact">
<li><a href="#" class="fa fa-twitter"><span>Twitter</span></a></li>
<li><a href="" class="fa fa-at"><span>Email</span></a></li>
</ul>
</section>
</div>
</div>
-->
{% include footer.html %}
</body>
</html>