forked from codeplaysoftware/syclacademy
-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
423 lines (414 loc) · 15.6 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
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<link rel="stylesheet" href="../common-revealjs/css/reveal.css">
<link rel="stylesheet" href="../common-revealjs/css/theme/white.css">
<link rel="stylesheet" href="../common-revealjs/css/custom.css">
<script>
// This is needed when printing the slides to pdf
var link = document.createElement( 'link' );
link.rel = 'stylesheet';
link.type = 'text/css';
link.href = window.location.search.match( /print-pdf/gi ) ? '../common-revealjs/css/print/pdf.css' : '../common-revealjs/css/print/paper.css';
document.getElementsByTagName( 'head' )[0].appendChild( link );
</script>
<script>
// This is used to display the static images on each slide,
// See global-images in this html file and custom.css
(function() {
if(window.addEventListener) {
window.addEventListener('load', () => {
let slides = document.getElementsByClassName("slide-background");
if (slides.length === 0) {
slides = document.getElementsByClassName("pdf-page")
}
// Insert global images on each slide
for(let i = 0, max = slides.length; i < max; i++) {
let cln = document.getElementById("global-images").cloneNode(true);
cln.removeAttribute("id");
slides[i].appendChild(cln);
}
// Remove top level global images
let elem = document.getElementById("global-images");
elem.parentElement.removeChild(elem);
}, false);
}
})();
</script>
</head>
<body>
<div class="reveal">
<div class="slides">
<div id="global-images" class="global-images">
<img src="../common-revealjs/images/sycl_academy.png" />
<img src="../common-revealjs/images/sycl_logo.png" />
<img src="../common-revealjs/images/trademarks.png" />
</div>
<!--Slide 1-->
<section>
<div class="hbox" data-markdown>
## What is SYCL?
</div>
</section>
<!--Slide 2-->
<section class="hbox" data-markdown>
## Learning Objectives
* Learn about the SYCL specification and its implementations
* Learn about the components of a SYCL implementation
* Learn about how a SYCL source file is compiled
* Learn where to find useful resources for SYCL
</section>
<!--Slide 3-->
<section>
<div class="hbox" data-markdown>
#### What is SYCL?
</div>
<div class="container" data-markdown>
</div>
<div class="hbox" data-markdown>
SYCL is a single source, high-level, standard C++ programming model, that can target a range of heterogeneous platforms
</div>
<aside class="notes">
We'll examine what each of the parts of this statement means.
</aside>
</section>
<!--Slide 4-->
<section>
<div class="hbox" data-markdown>
#### What is SYCL?
</div>
<div class="container" data-markdown>
A first example of SYCL code. Elements will be explained in coming sections!
</div>
<div class="container" data-markdown>
</div>
</section>
<!--Slide 4A-->
<section data-markdown>
## SYCL is...
* SYCL extends C++ in two key ways:
* device discovery (and information)
* device control (kernels of work, memory)
* SYCL is modern C++
* SYCL is open, multivendor, multiarchitecture
</section>
<!--Slide 5-->
<section class="hbox">
<div class="hbox" data-markdown>
#### What is SYCL?
</div>
<div class="hbox" data-markdown>
SYCL is a ***single source***, high-level, standard C++ programming model, that can target a range of heterogeneous platforms
</div>
<div class="container">
<div class="col" data-markdown>
</div>
<div class="col" data-markdown>
* SYCL allows you to write both host CPU and device code in the same C++ source file
* This requires two compilation passes; one for the host code and one for the device code
</div>
</div>
<aside class="notes">
Single source means that SYCL kernel source code can live in the same source file as the other code you are using.
This is not the case with something like OpenCL, another heterogeneous programming interface, where the kernel code lives in a separate file.
Your kernel contains the code you want to execute on parallel hardware, i.e the bit you want to accelerate such as a vector addition
</aside>
</section>
<!--Slide 6-->
<section class="hbox">
<div class="hbox" data-markdown>
#### What is SYCL?
</div>
<div class="hbox" data-markdown>
SYCL is a single source, ***high-level***, standard C++ programming model, that can target a range of heterogeneous platforms
</div>
<div class="container">
<article class="main" data-markdown>
* SYCL provides high-level abstractions over common boilerplate code
* Platform/device selection
* Buffer creation and data movement
* Kernel function compilation
* Dependency management and scheduling
</article>
<aside class="notes">
</aside>
</section>
<!--Slide 7-->
<section class="hbox">
<div class="hbox" data-markdown>
#### What is SYCL?
</div>
<div class="hbox" data-markdown>
SYCL is a single source, high-level ***standard C++*** programming model, that can target a range of heterogeneous platforms
</div>
<div class="container">
<div class="col" data-markdown>
</div>
<div class="col" data-markdown>
* SYCL allows you to write standard C++
* SYCL 2020 is based on C++17
* Unlike the other implementations shown on the left there are:
* No language extensions
* No pragmas
* No attributes
</div>
</div>
<aside class="notes">
SYCL only uses standard C++ code for both the host applications and kernel functions
Other models use language extensions such as pragmas, attributes or keywords to indicate device code
</aside>
</section>
<!--Slide 8-->
<section class="hbox">
<div class="hbox" data-markdown>
#### What is SYCL?
</div>
<div class="hbox" data-markdown>
SYCL is a single source, high-level standard C++ programming model, that can **target a range of heterogeneous platforms**
</div>
<div class="container">
<div class="col" data-markdown>
</div>
<div class="col" data-markdown>
* SYCL can target any device supported by its backend
* SYCL can target a number of different backends
SYCL has been designed to be implemented on top of a variety of backends. Current implementations support backends such as OpenCL, CUDA, HIP, OpenMP and others.
</div>
</div>
<aside class="notes">
SYCL has been designed to enable developers to write C++ code that can be used to target different types of processors that are part of a heterogeneous system.
This means that you can write your code and deploy it to devices that support the SYCL implementation's back-end.
Whilst the current specification of SYCL is bound to OpenCL, it is possible to support non-OpenCL back-ends. For example hipSYCL targets AMD's HIP interface and ComputeCpp targets NVidia's ptx instruction set
</aside>
</section>
<!--Slide 9-->
<section class="hbox">
<div class="hbox" data-markdown>
#### SYCL specification
</div>
<div class="container">
<aside class="aside aside-1" data-markdown>
</aside>
</section>
<!--Slide 10-->
<section class="hbox">
<div class="hbox" data-markdown>
#### SYCL implementations
</div>
<img src="./sycl_implementation_targets.png" alt="SYCL-Implementations">
</section>
<!--Slide 11-->
<section>
<div class="hbox" data-markdown>
#### What a SYCL implementation looks like
</div>
<div class="container">
<div class="col" data-markdown>
</div>
<div class="col" data-markdown>
* The SYCL interface is a C++ template library that developers can use to access the features of SYCL
* The same interface is used for both the host and device code
</div>
</div>
<div class="bottom-bullets" data-markdown>
* The host is generally the CPU and is used to dispatch the parallel execution of kernels
* The device is the parallel unit used to execute the kernels, such as a GPU
</div>
</section>
<!--Slide 12-->
<section>
<div class="hbox" data-markdown>
#### What a SYCL implementation looks like
</div>
<div class="container">
<div class="col" data-markdown>
</div>
<div class="col" data-markdown>
* The SYCL runtime is a library that schedules and executes work
* It loads kernels, tracks data dependencies and schedules commands
</div>
</section>
<!--Slide 13-->
<section>
<div class="hbox" data-markdown>
#### What a SYCL implementation looks like
</div>
<div class="container">
<div class="col" data-markdown>
</div>
<div class="col" data-markdown>
* There is no Host Device in SYCL (as of SYCL 2020)
* SYCL 1.2.1 had a concept of a 'magical' host device - an emulated backend
* SYCL 2020 implementations generally offer a CPU device
* Often, the best debugging on a platform is using a CPU device
* Yet, debugging off the CPU is important to discover offloading issues
</div>
</section>
<!--Slide 14-->
<section>
<div class="hbox" data-markdown>
#### What a SYCL implementation looks like
</div>
<div class="container">
<div class="col" data-markdown>
</div>
<div class="col" data-markdown>
* The back-end interface is where the SYCL runtime calls down into a back-end in order to execute on a particular device
* Many implementations provide OpenCL backends, but some provide additional or different backends.
</div>
</div>
</section>
<!--Slide 15-->
<section>
<div class="hbox" data-markdown>
#### What a SYCL implementation looks like
</div>
<div class="container">
<div class="col" data-markdown>
</div>
<div class="col" data-markdown>
* The SYCL device compiler is a C++ compiler which can identify SYCL kernels and compile them down to an IR or ISA
* This can be SPIR, SPIR-V, GCN, PTX or any proprietary vendor ISA
* Some SYCL implementations are library only in
which case they do not require a device compiler
</div>
</div>
<div class="bottom-bullets" data-markdown>
**IR** = Intermediate Representation **ISA** = Instruction Set Architecture
</div>
</section>
<!--Slide 16-->
<section>
<div class="hbox" data-markdown>
#### Std C++ compilation model
</div>
<div class="container" data-markdown>
</div>
<div class="container" data-markdown>
* This is the typical compilation model for a C++ source file.
</div>
</section>
<!--Slide 17-->
<section>
<div class="hbox" data-markdown>
#### Std C++ compilation model
</div>
<div class="container" data-markdown>
</div>
<div class="container" data-markdown>
* So how do you compile a source file to also target the GPU?
</div>
</section>
<!--Slide 18-->
<section>
<div class="hbox" data-markdown>
#### Std C++ compilation model
</div>
<div class="container" data-markdown>
</div>
<div class="container" data-markdown>
* As SYCL is single source the kernel functions are standard C++ function objects or lambda expressions.
* These are defined by submitting them to specific APIs.
</div>
</section>
<!--Slide 19-->
<section>
<div class="hbox" data-markdown>
#### Std C++ compilation model
</div>
<div class="container" data-markdown>
</div>
<div class="container" data-markdown>
* As well as the standard C++ compiler, the source file is also compiled by a SYCL device compiler.
* This produces a device IR such as SPIR, SPIR-V or PTX or ISA for a specific architecture containing the GPU code.
</div>
</section>
<!--Slide 20-->
<section>
<div class="hbox" data-markdown>
#### Std C++ compilation model
</div>
<div class="container" data-markdown>
</div>
<div class="container" data-markdown>
* The CPU object is then linked with the device IR or ISA to form a single executable with both the CPU and GPU code.
</div>
</section>
<!--Slide 21-->
<section>
<div class="hbox" data-markdown>
#### Std C++ compilation model
</div>
<div class="container" data-markdown>
</div>
<div class="container" data-markdown>
* This is the multi-compiler compilation model.
* This allows the host compiler (MSVC, clang, icx, gcc) to be independent of the SYCL device compiler.
</div>
</section>
<!--Slide 22-->
<section>
<div class="hbox" data-markdown>
#### Std C++ compilation model
</div>
<div class="container" data-markdown>
</div>
<div class="container" data-markdown>
* SYCL also supports a single-compiler compilation model.
* Where both the host compiler and SYCL device compiler are invoked from the same driver.
</div>
</section>
<!--Slide 23-->
<section data-markdown>
## Where to Get Started with SYCL
* Visit https://sycl.tech to find out about all the SYCL book, implementations, tutorials, news, and videos
* Visit https://www.khronos.org/sycl/ to find the latest SYCL specifications
* Checkout the documentation provided with one of the SYCL implementations.
</section>
<!--Slide 24-->
<section data-markdown>
## Questions
</section>
<!--Slide 25-->
<section>
<div class="hbox" data-markdown>
#### Exercise
</div>
<div class="container" data-markdown>
Code_Exercises/Exercise_1_Compiling_with_SYCL/source
</div>
<div class="container" data-markdown>
Configure your environment for using SYCL and compile a source file with the SYCL compiler.
</div>
</section>
</div>
</div>
<script src="../common-revealjs/js/reveal.js"></script>
<script src="../common-revealjs/plugin/markdown/marked.js"></script>
<script src="../common-revealjs/plugin/markdown/markdown.js"></script>
<script src="../common-revealjs/plugin/notes/notes.js"></script>
<script>
Reveal.initialize({mouseWheel: true, defaultNotes: true});
Reveal.configure({ slideNumber: true });
</script>
</body>
</html>