1# Copyright (C) 2014-2018 Intel Corporation.   All Rights Reserved.
2#
3# Permission is hereby granted, free of charge, to any person obtaining a
4# copy of this software and associated documentation files (the "Software"),
5# to deal in the Software without restriction, including without limitation
6# the rights to use, copy, modify, merge, publish, distribute, sublicense,
7# and/or sell copies of the Software, and to permit persons to whom the
8# Software is furnished to do so, subject to the following conditions:
9#
10# The above copyright notice and this permission notice (including the next
11# paragraph) shall be included in all copies or substantial portions of the
12# Software.
13#
14# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
17# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
18# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
19# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
20# IN THE SOFTWARE.
21import sys
22
23# Python source
24KNOBS = [
25
26    ['ENABLE_ASSERT_DIALOGS', {
27        'type'      : 'bool',
28        'default'   : 'true',
29        'desc'      : ['Use dialogs when asserts fire.',
30                       'Asserts are only enabled in debug builds'],
31        'category'  : 'debug',
32    }],
33
34    ['SINGLE_THREADED', {
35        'type'      : 'bool',
36        'default'   : 'false',
37        'desc'      : ['If enabled will perform all rendering on the API thread.',
38                       'This is useful mainly for debugging purposes.'],
39        'category'  : 'debug',
40    }],
41
42    ['DUMP_SHADER_IR', {
43        'type'      : 'bool',
44        'default'   : 'false',
45        'desc'      : ['Dumps shader LLVM IR at various stages of jit compilation.'],
46        'category'  : 'debug',
47    }],
48
49    ['USE_GENERIC_STORETILE', {
50        'type'      : 'bool',
51        'default'   : 'false',
52        'desc'      : ['Always use generic function for performing StoreTile.',
53                       'Will be slightly slower than using optimized (jitted) path'],
54        'category'  : 'debug_adv',
55    }],
56
57    ['FAST_CLEAR', {
58        'type'      : 'bool',
59        'default'   : 'true',
60        'desc'      : ['Replace 3D primitive execute with a SWRClearRT operation and',
61                       'defer clear execution to first backend op on hottile, or hottile store'],
62        'category'  : 'perf_adv',
63    }],
64
65    ['MAX_NUMA_NODES', {
66        'type'      : 'uint32_t',
67        'default'   : '1' if sys.platform == 'win32' else '0',
68        'desc'      : ['Maximum # of NUMA-nodes per system used for worker threads',
69                       '  0 == ALL NUMA-nodes in the system',
70                       '  N == Use at most N NUMA-nodes for rendering'],
71        'category'  : 'perf',
72    }],
73
74    ['MAX_CORES_PER_NUMA_NODE', {
75        'type'      : 'uint32_t',
76        'default'   : '0',
77        'desc'      : ['Maximum # of cores per NUMA-node used for worker threads.',
78                       '  0 == ALL non-API thread cores per NUMA-node',
79                       '  N == Use at most N cores per NUMA-node'],
80        'category'  : 'perf',
81    }],
82
83    ['MAX_THREADS_PER_CORE', {
84        'type'      : 'uint32_t',
85        'default'   : '1',
86        'desc'      : ['Maximum # of (hyper)threads per physical core used for worker threads.',
87                       '  0 == ALL hyper-threads per core',
88                       '  N == Use at most N hyper-threads per physical core'],
89        'category'  : 'perf',
90    }],
91
92    ['MAX_WORKER_THREADS', {
93        'type'      : 'uint32_t',
94        'default'   : '0',
95        'desc'      : ['Maximum worker threads to spawn.',
96                       '',
97                       'IMPORTANT: If this is non-zero, no worker threads will be bound to',
98                       'specific HW threads.  They will all be "floating" SW threads.',
99                       'In this case, the above 3 KNOBS will be ignored.'],
100        'category'  : 'perf',
101    }],
102
103    ['BASE_NUMA_NODE', {
104        'type'      : 'uint32_t',
105        'default'   : '0',
106        'desc'      : ['Starting NUMA node index to use when allocating compute resources.',
107                       'Setting this to a non-zero value will reduce the maximum # of NUMA nodes used.'],
108        'category'  : 'perf',
109    }],
110
111    ['BASE_CORE', {
112        'type'      : 'uint32_t',
113        'default'   : '0',
114        'desc'      : ['Starting core index to use when allocating compute resources.',
115                       'Setting this to a non-zero value will reduce the maximum # of cores used.'],
116        'category'  : 'perf',
117    }],
118
119    ['BASE_THREAD', {
120        'type'      : 'uint32_t',
121        'default'   : '0',
122        'desc'      : ['Starting thread index to use when allocating compute resources.',
123                       'Setting this to a non-zero value will reduce the maximum # of threads used.'],
124        'category'  : 'perf',
125    }],
126
127    ['BUCKETS_START_FRAME', {
128        'type'      : 'uint32_t',
129        'default'   : '1200',
130        'desc'      : ['Frame from when to start saving buckets data.',
131                       '',
132                       'NOTE: KNOB_ENABLE_RDTSC must be enabled in core/knobs.h',
133                       'for this to have an effect.'],
134        'category'  : 'perf_adv',
135    }],
136
137    ['BUCKETS_END_FRAME', {
138        'type'      : 'uint32_t',
139        'default'   : '1400',
140        'desc'      : ['Frame at which to stop saving buckets data.',
141                       '',
142                       'NOTE: KNOB_ENABLE_RDTSC must be enabled in core/knobs.h',
143                       'for this to have an effect.'],
144        'category'  : 'perf_adv',
145    }],
146
147    ['WORKER_SPIN_LOOP_COUNT', {
148        'type'      : 'uint32_t',
149        'default'   : '5000',
150        'desc'      : ['Number of spin-loop iterations worker threads will perform',
151                       'before going to sleep when waiting for work'],
152        'category'  : 'perf_adv',
153    }],
154
155    ['MAX_DRAWS_IN_FLIGHT', {
156        'type'      : 'uint32_t',
157        'default'   : '256',
158        'desc'      : ['Maximum number of draws outstanding before API thread blocks.',
159                       'This value MUST be evenly divisible into 2^32'],
160        'category'  : 'perf_adv',
161    }],
162
163    ['MAX_PRIMS_PER_DRAW', {
164        'type'      : 'uint32_t',
165        'default'   : '49152',
166        'desc'      : ['Maximum primitives in a single Draw().',
167                       'Larger primitives are split into smaller Draw calls.',
168                       'Should be a multiple of (3 * vectorWidth).'],
169        'category'  : 'perf_adv',
170    }],
171
172    ['MAX_TESS_PRIMS_PER_DRAW', {
173        'type'      : 'uint32_t',
174        'default'   : '16',
175        'desc'      : ['Maximum primitives in a single Draw() with tessellation enabled.',
176                       'Larger primitives are split into smaller Draw calls.',
177                       'Should be a multiple of (vectorWidth).'],
178        'category'  : 'perf_adv',
179    }],
180
181
182    ['DEBUG_OUTPUT_DIR', {
183        'type'      : 'std::string',
184        'default'   : r'%TEMP%\Rast\DebugOutput' if sys.platform == 'win32' else '/tmp/Rast/DebugOutput',
185        'desc'      : ['Output directory for debug data.'],
186        'category'  : 'debug',
187    }],
188
189    ['JIT_ENABLE_CACHE', {
190        'type'      : 'bool',
191        'default'   : 'false',
192        'desc'      : ['Enables caching of compiled shaders'],
193        'category'  : 'debug_adv',
194    }],
195
196    ['JIT_OPTIMIZATION_LEVEL', {
197        'type'      : 'int',
198        'default'   : '-1',
199        'desc'      : ['JIT compile optimization level:',],
200        'category'  : 'debug',
201        'control'   : 'dropdown',
202        'choices' : [
203            {
204                'name'  : 'Automatic',
205                'desc'  : 'Automatic based on other KNOB and build settings',
206                'value' : -1,
207            },
208            {
209                'name'  : 'Debug',
210                'desc'  : 'No optimization: -O0',
211                'value' : 0,
212            },
213            {
214                'name'  : 'Less',
215                'desc'  : 'Some optimization: -O1',
216                'value' : 1,
217            },
218            {
219                'name'  : 'Optimize',
220                'desc'  : 'Default Clang / LLVM optimizations: -O2',
221                'value' : 2,
222            },
223            {
224                'name'  : 'Aggressive',
225                'desc'  : 'Maximum optimization: -O3',
226                'value' : 3,
227            },
228        ],
229    }],
230
231    ['JIT_CACHE_DIR', {
232        'type'      : 'std::string',
233        'default'   : r'%TEMP%\SWR\JitCache' if sys.platform == 'win32' else '${HOME}/.swr/jitcache',
234        'desc'      : ['Cache directory for compiled shaders.'],
235        'category'  : 'debug',
236    }],
237
238    ['TOSS_DRAW', {
239        'type'      : 'bool',
240        'default'   : 'false',
241        'desc'      : ['Disable per-draw/dispatch execution'],
242        'category'  : 'perf',
243    }],
244
245    ['TOSS_QUEUE_FE', {
246        'type'      : 'bool',
247        'default'   : 'false',
248        'desc'      : ['Stop per-draw execution at worker FE',
249                       '',
250                       'NOTE: Requires KNOB_ENABLE_TOSS_POINTS to be enabled in core/knobs.h'],
251        'category'  : 'perf_adv',
252    }],
253
254    ['TOSS_FETCH', {
255        'type'      : 'bool',
256        'default'   : 'false',
257        'desc'      : ['Stop per-draw execution at vertex fetch',
258                       '',
259                       'NOTE: Requires KNOB_ENABLE_TOSS_POINTS to be enabled in core/knobs.h'],
260        'category'  : 'perf_adv',
261    }],
262
263    ['TOSS_IA', {
264        'type'      : 'bool',
265        'default'   : 'false',
266        'desc'      : ['Stop per-draw execution at input assembler',
267                       '',
268                       'NOTE: Requires KNOB_ENABLE_TOSS_POINTS to be enabled in core/knobs.h'],
269        'category'  : 'perf_adv',
270    }],
271
272    ['TOSS_VS', {
273        'type'      : 'bool',
274        'default'   : 'false',
275        'desc'      : ['Stop per-draw execution at vertex shader',
276                       '',
277                       'NOTE: Requires KNOB_ENABLE_TOSS_POINTS to be enabled in core/knobs.h'],
278        'category'  : 'perf_adv',
279    }],
280
281    ['TOSS_SETUP_TRIS', {
282        'type'      : 'bool',
283        'default'   : 'false',
284        'desc'      : ['Stop per-draw execution at primitive setup',
285                       '',
286                       'NOTE: Requires KNOB_ENABLE_TOSS_POINTS to be enabled in core/knobs.h'],
287        'category'  : 'perf_adv',
288    }],
289
290    ['TOSS_BIN_TRIS', {
291        'type'      : 'bool',
292        'default'   : 'false',
293        'desc'      : ['Stop per-draw execution at primitive binning',
294                       '',
295                       'NOTE: Requires KNOB_ENABLE_TOSS_POINTS to be enabled in core/knobs.h'],
296        'category'  : 'perf_adv',
297    }],
298
299    ['TOSS_RS', {
300        'type'      : 'bool',
301        'default'   : 'false',
302        'desc'      : ['Stop per-draw execution at rasterizer',
303                       '',
304                       'NOTE: Requires KNOB_ENABLE_TOSS_POINTS to be enabled in core/knobs.h'],
305        'category'  : 'perf_adv',
306    }],
307
308    ['DISABLE_SPLIT_DRAW', {
309        'type'      : 'bool',
310        'default'   : 'false',
311        'desc'      : ['Don\'t split large draws into smaller draws.,',
312                       'MAX_PRIMS_PER_DRAW and MAX_TESS_PRIMS_PER_DRAW can be used to control split size.',
313                       '',
314                       'Useful to disable split draws for gathering archrast stats.'],
315        'category'  : 'perf_adv',
316    }],
317
318    ['AR_ENABLE_PIPELINE_STATS', {
319        'type'      : 'bool',
320        'default'   : 'true',
321        'desc'      : ['Enable pipeline stats when using Archrast'],
322        'category'  : 'archrast',
323    }],
324
325    ['AR_ENABLE_SHADER_STATS', {
326        'type'      : 'bool',
327        'default'   : 'true',
328        'desc'      : ['Enable shader stats when using Archrast'],
329        'category'  : 'archrast',
330    }],
331
332    ['AR_ENABLE_SWTAG_DATA', {
333        'type'      : 'bool',
334        'default'   : 'false',
335        'desc'      : ['Enable SWTag data when using Archrast'],
336        'category'  : 'archrast',
337    }],
338
339    ['AR_ENABLE_SWR_EVENTS', {
340        'type'      : 'bool',
341        'default'   : 'true',
342        'desc'      : ['Enable internal SWR events when using Archrast'],
343        'category'  : 'archrast',
344    }],
345
346    ['AR_ENABLE_PIPELINE_EVENTS', {
347        'type'      : 'bool',
348        'default'   : 'true',
349        'desc'      : ['Enable pipeline events when using Archrast'],
350        'category'  : 'archrast',
351    }],
352
353    ['AR_ENABLE_SHADER_EVENTS', {
354        'type'      : 'bool',
355        'default'   : 'true',
356        'desc'      : ['Enable shader events when using Archrast'],
357        'category'  : 'archrast',
358    }],
359
360    ['AR_ENABLE_SWTAG_EVENTS', {
361        'type'      : 'bool',
362        'default'   : 'false',
363        'desc'      : ['Enable SWTag events when using Archrast'],
364        'category'  : 'archrast',
365    }],
366
367    ['AR_ENABLE_MEMORY_EVENTS', {
368        'type'      : 'bool',
369        'default'   : 'false',
370        'desc'      : ['Enable memory events when using Archrast'],
371        'category'  : 'archrast',
372    }],
373
374    ['AR_MEM_SET_BYTE_GRANULARITY', {
375        'type'      : 'uint32_t',
376        'default'   : '64',
377        'desc'      : ['Granularity and alignment of tracking of memory accesses',
378                       'ONLY ACTIVE UNDER ArchRast.'],
379        'category'  : 'archrast',
380    }],
381
382
383    ]
384