-
Notifications
You must be signed in to change notification settings - Fork 19
/
sysfs-gpu-controls.c
739 lines (646 loc) · 22.8 KB
/
sysfs-gpu-controls.c
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
#include <stdint.h>
#include <stdbool.h>
#include <dirent.h>
#include "miner.h"
#include "pptable_v1_0.h"
#include "sysfs-gpu-controls.h"
bool initialized = false;
bool has_sysfs_hwcontrols = false;
bool opt_reorder = false;
int opt_hysteresis = 3;
int opt_targettemp = 80;
int opt_overheattemp = 85;
float sysfs_gpu_vddc(int gpu) { return -1; }
int sysfs_gpu_activity(int gpu) { return -1; }
int sysfs_set_vddc(int gpu, float fVddc) { return 1; }
#ifndef __linux__
static void sysfs_init(gpu_sysfs_info *info, int gpu_idx)
{
memset(info, 0, sizeof(gpu_sysfs_info));
info->fd_pptable = info->fd_fan = info->fd_temp = info->fd_pwm = info->fd_sclk = info->fd_mclk = -1;
}
#else
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
static void sysfs_init(gpu_sysfs_info *info, int gpu_idx)
{
char path[256];
struct dirent *inner_hwmon;
info->fd_pptable = info->fd_fan = info->fd_temp = info->fd_pwm = info->fd_sclk = info->fd_mclk = -1;
snprintf(path, sizeof(path), "/sys/bus/pci/devices/0000:%.2x:%.2x.%.1x/",
info->pcie_index[0], info->pcie_index[1], info->pcie_index[2]);
size_t len = strlen(path);
snprintf(path + len, sizeof(path) - len, "hwmon");
DIR *hwmon = opendir(path);
if (hwmon == NULL) {
applog(LOG_DEBUG, "Failed to open hwmon directory %s for GPU%d", path, gpu_idx);
snprintf(path, sizeof(path), "/sys/class/drm/card%d/device/hwmon", gpu_idx);
len = strlen(path) - 5;
hwmon = opendir(path);
if (hwmon == NULL) {
applog(LOG_DEBUG, "Failed to open hwmon directory %s for GPU%d", path, gpu_idx);
return;
}
}
snprintf(path + len, sizeof(path) - len, "pp_table");
int fd = open(path, O_RDONLY | O_RSYNC);
bool success = false;
if (fd != -1) {
int size = lseek(fd, 0, SEEK_END);
if (size > 0) {
lseek(fd, 0, SEEK_SET);
info->pptable = (uint8_t*) malloc(size);
info->default_pptable = (uint8_t*) malloc(size);
info->pptable_size = read(fd, info->pptable, size);
memcpy(info->default_pptable, info->pptable, info->pptable_size);
info->fd_pptable = open(path, O_WRONLY | O_DSYNC);
success = (info->fd_pptable != -1 && info->pptable_size > 0);
}
close(fd);
}
if (!success) {
if (info->fd_pptable != -1) {
close(info->fd_pptable);
info->fd_pptable = -1;
}
applog(LOG_DEBUG, "Failed to open/read %s", path);
}
else
pthread_mutex_init(&info->rw_lock, NULL);
snprintf(path + len, sizeof(path) - len, "pp_dpm_sclk");
info->fd_sclk = open(path, O_RDONLY | O_RSYNC);
if (info->fd_sclk == -1)
applog(LOG_DEBUG, "Failed to open %s", path);
snprintf(path + len, sizeof(path) - len, "pp_dpm_mclk");
info->fd_mclk = open(path, O_RDONLY | O_RSYNC);
if (info->fd_mclk == -1)
applog(LOG_DEBUG, "Failed to open %s", path);
while (true) {
inner_hwmon = readdir(hwmon);
if (inner_hwmon == NULL) {
closedir(hwmon);
applog(LOG_DEBUG, "Failed to parse hwmon directory for GPU%d", gpu_idx);
return;
}
if (inner_hwmon->d_type != DT_DIR)
continue;
if (!memcmp(inner_hwmon->d_name, "hwmon", 5))
break;
}
snprintf(path + len, sizeof(path) - len, "hwmon/%s/", inner_hwmon->d_name);
len = strlen(path);
closedir(hwmon);
snprintf(path + len, sizeof(path) - len, "pwm1");
info->fd_fan = open(path, O_RDWR | O_RSYNC | O_DSYNC);
if (info->fd_fan == -1)
applog(LOG_DEBUG, "Failed to open %s", path);
else
pthread_mutex_init(&info->rw_lock, NULL);
char buf[17] = {0};
snprintf(path + len, sizeof(path) - len, "pwm1_min");
fd = open(path, O_RDONLY | O_RSYNC);
info->min_fanspeed = 0;
if (fd != -1) {
int bytes_read = read(fd, &buf, sizeof(buf) - 1);
if (bytes_read > 0)
info->min_fanspeed = strtoul(buf, NULL, 10);
else
applog(LOG_DEBUG, "Failed to read %s", path);
close(fd);
}
else
applog(LOG_DEBUG, "Failed to open %s", path);
snprintf(path + len, sizeof(path) - len, "pwm1_max");
fd = open(path, O_RDONLY | O_RSYNC);
info->max_fanspeed = 255;
if (fd != -1) {
int bytes_read = read(fd, &buf, sizeof(buf) - 1);
if (bytes_read > 0) {
buf[bytes_read] = '\0';
info->max_fanspeed = strtoul(buf, NULL, 10);
}
else
applog(LOG_DEBUG, "Failed to read %s", path);
close(fd);
}
else
applog(LOG_DEBUG, "Failed to open %s", path);
snprintf(path + len, sizeof(path) - len, "temp1_input");
info->fd_temp = open(path, O_RDONLY | O_RSYNC);
if (info->fd_temp == -1)
applog(LOG_DEBUG, "Failed to open %s", path);
snprintf(path + len, sizeof(path) - len, "fan1_input");
info->fd_pwm = open(path, O_RDONLY | O_RSYNC);
if (info->fd_pwm == -1)
applog(LOG_DEBUG, "Failed to open %s", path);
}
#endif
static int parse_dpm_clk_table(int fd)
{
char buf[1024];
int ret = -1;
#ifdef __linux__
if (has_sysfs_hwcontrols && fd != -1) {
lseek(fd, 0, SEEK_SET);
size_t len = read(fd, buf, sizeof(buf) - 1);
if (len > 0) {
buf[len] = '\0';
char *ptr = strtok(buf, "\n");
while (ptr != NULL) {
int index;
int freq;
char active_str[3] = {0};
sscanf(ptr, "%d: %dMhz%s", &index, &freq, active_str);
if (active_str[0] == '*' || active_str[1] == '*') {
ret = freq;
break;
}
ptr = strtok(NULL, "\n");
}
}
}
#endif
return ret;
}
int sysfs_gpu_engineclock(int gpu)
{
return parse_dpm_clk_table(gpus[gpu].sysfs_info.fd_sclk);
}
int sysfs_gpu_memclock(int gpu)
{
return parse_dpm_clk_table(gpus[gpu].sysfs_info.fd_mclk);
}
float sysfs_gpu_temp(int gpu)
{
gpu_sysfs_info *info = &gpus[gpu].sysfs_info;
char temp_str[17];
float ret = -1.f;
#ifdef __linux__
if (has_sysfs_hwcontrols && info->fd_temp != -1) {
lseek(info->fd_temp, 0, SEEK_SET);
int bytes_read = read(info->fd_temp, &temp_str, sizeof(temp_str) - 1);
if (bytes_read > 0) {
temp_str[bytes_read] = '\0';
ret = strtol(temp_str, NULL, 10) / 1000.f;
}
}
#endif
return ret;
}
static int fanpercent_to_speed(gpu_sysfs_info *info, float fanpercent)
{
int min = info->min_fanspeed;
int max = info->max_fanspeed;
float range = max - min;
float speed = fanpercent / 100.f * range + min + 0.5f;
return speed;
}
static float fanspeed_to_percent(gpu_sysfs_info *info, int speed)
{
int min = info->min_fanspeed;
int max = info->max_fanspeed;
float range = max - min;
float fanpercent = (speed - min) / range * 100.f;
return fanpercent;
}
int sysfs_gpu_fanspeed(int gpu)
{
gpu_sysfs_info *info = &gpus[gpu].sysfs_info;
char speed_str[17];
int ret = -1;
#ifdef __linux__
if (has_sysfs_hwcontrols && info->fd_pwm != -1) {
lseek(info->fd_pwm, 0, SEEK_SET);
int bytes_read = read(info->fd_pwm, speed_str, sizeof(speed_str) - 1);
if (bytes_read > 0) {
speed_str[bytes_read] = '\0';
ret = strtoul(speed_str, NULL, 10);
}
}
#endif
return ret;
}
float sysfs_gpu_fanpercent(int gpu)
{
gpu_sysfs_info *info = &gpus[gpu].sysfs_info;
char speed_str[17];
float ret = -1.f;
#ifdef __linux__
if (has_sysfs_hwcontrols && info->fd_fan != -1) {
pthread_mutex_lock(&info->rw_lock);
lseek(info->fd_fan, 0, SEEK_SET);
int bytes_read = read(info->fd_fan, speed_str, sizeof(speed_str) - 1);
pthread_mutex_unlock(&info->rw_lock);
if (bytes_read > 0) {
speed_str[bytes_read] = '\0';
unsigned long speed = strtoul(speed_str, NULL, 10);
ret = fanspeed_to_percent(info, speed);
}
}
#endif
return ret;
}
static int __set_fanspeed(gpu_sysfs_info *info, float fanpercent)
{
char speed_str[17];
int speed = fanpercent_to_speed(info, fanpercent);
int ret = 1;
snprintf(speed_str, sizeof(speed_str), "%d", speed);
#ifdef __linux__
if (info->fd_fan != -1 && fanpercent >= 0.f) {
lseek(info->fd_fan, 0, SEEK_SET);
ret = (write(info->fd_fan, speed_str, strlen(speed_str)) <= 0);
if (!ret)
info->target_fanpercent = fanpercent;
}
#endif
return ret;
}
int sysfs_set_fanspeed(int gpu, float fanpercent)
{
int ret = 1;
if (has_sysfs_hwcontrols && gpus[gpu].has_sysfs_hwcontrols) {
gpu_sysfs_info *info = &gpus[gpu].sysfs_info;
applog(LOG_DEBUG, "GPU%d: set fanpercent to %.3g%%", gpu, fanpercent);
pthread_mutex_lock(&info->rw_lock);
ret = __set_fanspeed(info, fanpercent);
pthread_mutex_unlock(&info->rw_lock);
}
return ret;
}
static Tonga_State_Array *get_state_array(uint8_t *pptable)
{
Tonga_POWERPLAYTABLE *header = (Tonga_POWERPLAYTABLE*) pptable;
return (Tonga_State_Array*) (pptable + header->usStateArrayOffset);
}
static Tonga_MCLK_Dependency_Table *get_mclk_table(uint8_t *pptable)
{
Tonga_POWERPLAYTABLE *header = (Tonga_POWERPLAYTABLE*) pptable;
return (Tonga_MCLK_Dependency_Table*) (pptable + header->usMclkDependencyTableOffset);
}
static Tonga_SCLK_Dependency_Table *get_sclk_table(uint8_t *pptable)
{
Tonga_POWERPLAYTABLE *header = (Tonga_POWERPLAYTABLE*) pptable;
return (Tonga_SCLK_Dependency_Table*) (pptable + header->usSclkDependencyTableOffset);
}
static Tonga_SCLK_Dependency_Record *get_sclk_record(Tonga_SCLK_Dependency_Table *sclk_tbl, int idx)
{
Tonga_SCLK_Dependency_Record *record = NULL;
if (/*idx < sclk_tbl->ucNumEntries &&*/ sclk_tbl->ucRevId <= 1) {
int entry_size = (sclk_tbl->ucRevId == 0 ? sizeof(Tonga_SCLK_Dependency_Record) : sizeof(Polaris_SCLK_Dependency_Record));
record = (Tonga_SCLK_Dependency_Record*) ((uint8_t*) sclk_tbl->entries + idx * entry_size);
}
return record;
}
static int __apply_pptable(gpu_sysfs_info *info)
{
#ifdef __linux__
lseek(info->fd_pptable, 0, SEEK_SET);
int bytes_written = write(info->fd_pptable, info->pptable, info->pptable_size);
if (opt_autofan && info->target_fanpercent >= 0)
__set_fanspeed(info, info->target_fanpercent); // restore fan settings
return (bytes_written != info->pptable_size);
#else
return 0;
#endif
}
int sysfs_set_powertune(int gpu, int power_limit)
{
gpu_sysfs_info *info = &gpus[gpu].sysfs_info;
int ret = 1;
if (has_sysfs_hwcontrols && power_limit > 0 && info->fd_pptable != -1) {
applog(LOG_DEBUG, "GPU%d: set power limit to %dW", gpu, power_limit);
pthread_mutex_lock(&info->rw_lock);
Tonga_POWERPLAYTABLE *header = (Tonga_POWERPLAYTABLE*) info->pptable;
Tonga_PowerTune_Table *power_tbl = (Tonga_PowerTune_Table*) (info->pptable + header->usPowerTuneTableOffset);
Tonga_PowerTune_Table *default_power_tbl = (Tonga_PowerTune_Table*) (info->default_pptable + header->usPowerTuneTableOffset);
power_tbl->usTDC = (float) default_power_tbl->usTDC * power_limit / default_power_tbl->usMaximumPowerDeliveryLimit;
power_tbl->usMaximumPowerDeliveryLimit = power_limit;
ret = __apply_pptable(info);
pthread_mutex_unlock(&info->rw_lock);
}
return ret;
}
static int gpu_powertune(int gpu)
{
int ret = 0;
gpu_sysfs_info *info = &gpus[gpu].sysfs_info;
if (has_sysfs_hwcontrols && info->fd_pptable != -1) {
Tonga_POWERPLAYTABLE *header = (Tonga_POWERPLAYTABLE*) info->pptable;
Tonga_PowerTune_Table *power_tbl = (Tonga_PowerTune_Table*) (info->pptable + header->usPowerTuneTableOffset);
ret = power_tbl->usMaximumPowerDeliveryLimit;
}
return ret;
}
static bool __set_engineclock(struct cgpu_info *cgpu, int iEngineClock)
{
gpu_sysfs_info *info = &cgpu->sysfs_info;
int engineclock = info->engineclock;
if (iEngineClock <= 0)
return false;
if (cgpu->min_engine > 0)
iEngineClock = MAX(iEngineClock, cgpu->min_engine);
if (cgpu->gpu_engine > 0)
iEngineClock = MIN(iEngineClock, cgpu->gpu_engine);
applog(LOG_DEBUG, "GPU%d: set engineclock to %dMHz", cgpu->device_id, iEngineClock);
iEngineClock *= 100;
Tonga_SCLK_Dependency_Table *sclk_tbl = get_sclk_table(info->pptable);
//replace current sclk table with default table
size_t sclk_tbl_size = (uintptr_t) get_sclk_record(sclk_tbl, sclk_tbl->ucNumEntries) - (uintptr_t) sclk_tbl;
memcpy(sclk_tbl, get_sclk_table(info->default_pptable), sclk_tbl_size);
//parse pp_table
Tonga_SCLK_Dependency_Record *prev_record = NULL, *record;
int idx = sclk_tbl->ucNumEntries - 1;
for (; idx >= 0; idx--) {
record = get_sclk_record(sclk_tbl, idx);
if (iEngineClock > record->ulSclk)
break;
prev_record = record;
}
if (prev_record != NULL)
record = prev_record;
//set new values
if (idx >= 0) {
record->ulSclk = iEngineClock;
info->engineclock = iEngineClock / 100;
}
else
info->engineclock = record->ulSclk / 100;
Tonga_State_Array *state_array = get_state_array(info->pptable);
idx = MIN(idx + 1, sclk_tbl->ucNumEntries - 1);
state_array->entries[state_array->ucNumEntries - 1].ucEngineClockIndexHigh = idx;
info->sclk_ind = idx;
return (engineclock != info->engineclock);
}
int sysfs_set_engineclock(int gpu, int iEngineClock)
{
struct cgpu_info *cgpu = &gpus[gpu];
gpu_sysfs_info *info = &cgpu->sysfs_info;
int ret = !has_sysfs_hwcontrols || info->fd_pptable == -1;
if (!ret) {
pthread_mutex_lock(&info->rw_lock);
bool updated = __set_engineclock(cgpu, iEngineClock);
if (updated)
ret = __apply_pptable(info);
pthread_mutex_unlock(&info->rw_lock);
}
return ret;
}
static bool __change_engineclock(struct cgpu_info *cgpu, bool increase)
{
gpu_sysfs_info *info = &cgpu->sysfs_info;
Tonga_SCLK_Dependency_Table *sclk_tbl = get_sclk_table(info->default_pptable);
Tonga_SCLK_Dependency_Record *record = get_sclk_record(sclk_tbl, info->sclk_ind);
bool updated = false;
if (increase) {
//check whether the power state can be increased
if (info->sclk_ind + 1 < sclk_tbl->ucNumEntries) {
record = get_sclk_record(sclk_tbl, ++info->sclk_ind);
updated = __set_engineclock(cgpu, record->ulSclk / 100);
}
else if (cgpu->gpu_engine > 0)
updated = __set_engineclock(cgpu, cgpu->gpu_engine);
}
else {
//check whether the power state can be decreased
if (info->sclk_ind - 1 >= 0) {
record = get_sclk_record(sclk_tbl, --info->sclk_ind);
updated = __set_engineclock(cgpu, record->ulSclk / 100);
}
else if (cgpu->min_engine > 0)
updated = __set_engineclock(cgpu, cgpu->min_engine);
}
return updated;
}
static bool __set_memoryclock(struct cgpu_info *cgpu, int iMemoryClock)
{
if (iMemoryClock <= 0)
return false;
gpu_sysfs_info *info = &cgpu->sysfs_info;
applog(LOG_DEBUG, "GPU%d: set memoryclock %dMHz", cgpu->device_id, iMemoryClock);
int memclock = info->memclock;
Tonga_MCLK_Dependency_Table* mclk_tbl = get_mclk_table(info->pptable);
Tonga_MCLK_Dependency_Record *record = &mclk_tbl->entries[mclk_tbl->ucNumEntries - 1];
record->ulMclk = iMemoryClock * 100;
info->memclock = iMemoryClock;
return (memclock != info->memclock);
}
int sysfs_set_memoryclock(int gpu, int iMemoryClock)
{
struct cgpu_info *cgpu = &gpus[gpu];
gpu_sysfs_info *info = &cgpu->sysfs_info;
int ret = !has_sysfs_hwcontrols || info->fd_pptable == -1;
if (!ret) {
pthread_mutex_lock(&info->rw_lock);
bool updated = __set_memoryclock(cgpu, iMemoryClock);
if (updated)
ret = __apply_pptable(info);
pthread_mutex_unlock(&info->rw_lock);
}
return ret;
}
/* Returns whether the fanspeed is optimal already or not. The fan_window bool
* tells us whether the current fanspeed is in the target range for fanspeeds.
*/
static bool sysfs_fan_autotune(int gpu, int temp, float fanpercent, int lasttemp, bool *fan_window)
{
struct cgpu_info *cgpu = &gpus[gpu];
gpu_sysfs_info *info = &gpus[gpu].sysfs_info;
int tdiff = temp - lasttemp;
int top = gpus[gpu].gpu_fan;
int bot = gpus[gpu].min_fan;
float newpercent = info->target_fanpercent;//fanpercent;
float iMin = 0, iMax = 100;
if (!opt_autoengine && temp > info->overheat_temp && fanpercent < iMax) {
applog(LOG_WARNING, "Overheat detected on GPU %d, increasing fan to 100%% (temp was %d, overtemp is %d)\n", gpu, temp, info->overheat_temp);
newpercent = iMax;
dev_error(cgpu, REASON_DEV_OVER_HEAT);
}
else if (temp > info->target_temp && fanpercent < top && tdiff >= 0) {
applog(LOG_DEBUG, "Temperature over target, increasing fanspeed");
if (temp > info->target_temp + opt_hysteresis)
newpercent = info->target_fanpercent + 10;
else
newpercent = info->target_fanpercent + 5;
if (newpercent > top)
newpercent = top;
}
else if (fanpercent > bot && temp < info->target_temp - opt_hysteresis) {
/* Detect large swings of 5 degrees or more and change fan by
* a proportion more */
if (tdiff <= 0) {
applog(LOG_DEBUG, "Temperature %d degrees below target, decreasing fanspeed", opt_hysteresis);
newpercent = info->target_fanpercent - 1 + tdiff / 5;
}
else if (tdiff >= 5) {
applog(LOG_DEBUG, "Temperature climbed %d while below target, increasing fanspeed", tdiff);
newpercent = info->target_fanpercent + tdiff / 5;
}
}
else {
/* We're in the optimal range, make minor adjustments if the
* temp is still drifting */
if (fanpercent > bot && tdiff < 0 && lasttemp < info->target_temp) {
applog(LOG_DEBUG, "Temperature dropping while in target range, decreasing fanspeed");
newpercent = info->target_fanpercent + tdiff;
}
else if (fanpercent < top && tdiff > 0 && temp > info->target_temp - opt_hysteresis) {
applog(LOG_DEBUG, "Temperature rising while in target range, increasing fanspeed");
newpercent = info->target_fanpercent + tdiff;
}
}
if (newpercent > iMax)
newpercent = iMax;
else if (newpercent < iMin)
newpercent = iMin;
if (newpercent < top)
*fan_window = true;
else
*fan_window = false;
if (newpercent != fanpercent) {
applog(LOG_INFO, "Setting GPU %d fan percentage to %g", gpu, newpercent);
set_fanspeed(gpu, newpercent);
/* If the fanspeed is going down and we're below the top speed,
* consider the fan optimal to prevent minute changes in
* fanspeed delaying GPU engine speed changes */
if (newpercent < fanpercent && *fan_window)
return true;
return false;
}
return true;
}
void sysfs_gpu_autotune(int gpu, enum dev_enable *denable)
{
struct cgpu_info *cgpu = &gpus[gpu];
if (!has_sysfs_hwcontrols || !cgpu->has_sysfs_hwcontrols)
return;
gpu_sysfs_info *info = &cgpu->sysfs_info;
bool fan_window = true;
int temp = sysfs_gpu_temp(gpu);
int fanpercent = sysfs_gpu_fanpercent(gpu) + 0.5f;
if (temp && fanpercent >= 0 && opt_autofan)
sysfs_fan_autotune(gpu, temp, info->target_fanpercent, info->last_temp, &fan_window);
info->last_temp = temp;
uint32_t ctr_diff = info->ctr++ - info->last_ctr;
if (opt_autoengine && info->fd_pptable != 1) {
bool updated = false;
pthread_mutex_lock(&info->rw_lock);
if (temp > cgpu->cutofftemp && *denable == DEV_ENABLED) {
applog(LOG_WARNING, "Hit thermal cutoff limit on GPU %d, disabling!", gpu);
updated = __set_engineclock(cgpu, cgpu->min_engine);
*denable = DEV_RECOVER;
dev_error(cgpu, REASON_DEV_THERMAL_CUTOFF);
}
else if (temp > info->overheat_temp && *denable == DEV_ENABLED) {
applog(LOG_WARNING, "Overheat detected, decreasing GPU %d clock speed", gpu);
updated = __change_engineclock(cgpu, false);
dev_error(cgpu, REASON_DEV_OVER_HEAT);
/* Only try to tune engine speed up if this GPU is not disabled */
}
else if (temp < info->overheat_temp - opt_hysteresis && fan_window && ctr_diff >= 6 && *denable == DEV_ENABLED) {
applog(LOG_DEBUG, "Temperature below overheat, increasing clock speed");
updated = __change_engineclock(cgpu, true);
info->last_ctr = info->ctr;
}
else if (temp < info->target_temp && *denable == DEV_RECOVER && opt_restart) {
applog(LOG_NOTICE, "Device recovered to temperature below target, re-enabling");
*denable = DEV_ENABLED;
for (int i = 0; i < cgpu->threads; i++)
cgsem_post(&cgpu->thr[i]->sem);
}
if (updated)
__apply_pptable(info);
pthread_mutex_unlock(&info->rw_lock);
}
}
bool sysfs_gpu_stats(int gpu, float *temp, int *engineclock, int *memclock, float *vddc,
int *activity, int *fanspeed, int *fanpercent, int *powertune)
{
if (!has_sysfs_hwcontrols || !gpus[gpu].has_sysfs_hwcontrols)
return false;
*temp = gpu_temp(gpu);
*fanspeed = gpu_fanspeed(gpu);
*fanpercent = gpu_fanpercent(gpu) + 0.5f;
*engineclock = gpu_engineclock(gpu);
*memclock = gpu_memclock(gpu);
*vddc = 0;
*activity = 0;
*powertune = gpu_powertune(gpu);
return true;
}
void sysfs_cleanup(int nDevs)
{
#ifdef __linux__
if (!has_sysfs_hwcontrols)
return;
has_sysfs_hwcontrols = false;
for (int i = 0; i < nDevs; i++) {
gpus[i].has_sysfs_hwcontrols = false;
gpu_sysfs_info *info = &gpus[i].sysfs_info;
// only overwrite pptable to default -> fd and memory are not released
if (info->pptable != NULL && info->fd_pptable != -1) {
pthread_mutex_lock(&info->rw_lock);
memcpy(info->pptable, info->default_pptable, info->pptable_size);
lseek(info->fd_pptable, 0, SEEK_SET);
write(info->fd_pptable, info->pptable, info->pptable_size);
pthread_mutex_unlock(&info->rw_lock);
sync();
}
}
#endif
}
bool init_sysfs_hwcontrols(int nDevs)
{
gpu_temp = &sysfs_gpu_temp;
gpu_engineclock = &sysfs_gpu_engineclock;
gpu_memclock = &sysfs_gpu_memclock;
gpu_vddc = &sysfs_gpu_vddc;
gpu_activity = &sysfs_gpu_activity;
gpu_fanspeed = &sysfs_gpu_fanspeed;
gpu_fanpercent = &sysfs_gpu_fanpercent;
set_powertune = &sysfs_set_powertune;
set_vddc = &sysfs_set_vddc;
set_fanspeed = &sysfs_set_fanspeed;
set_engineclock = &sysfs_set_engineclock;
set_memoryclock = &sysfs_set_memoryclock;
gpu_stats = &sysfs_gpu_stats;
gpu_autotune = &sysfs_gpu_autotune;
extern bool opt_noadl;
if (opt_noadl) {
extern bool adl_active;
adl_active = false;
has_sysfs_hwcontrols = false;
for (int i = 0; i < nDevs; i++) {
gpus[i].has_adl = false;
gpus[i].has_sysfs_hwcontrols = false;
}
return false;
}
for (int i = 0; i < nDevs; ++i) {
gpu_sysfs_info *info = &gpus[i].sysfs_info;
if (!initialized)
sysfs_init(info, i);
info->target_fanpercent = sysfs_gpu_fanpercent(i);
info->last_temp = sysfs_gpu_temp(i);
if (!info->overheat_temp)
info->overheat_temp = opt_overheattemp;
if (!info->target_temp)
info->target_temp = opt_targettemp;
gpus[i].has_sysfs_hwcontrols = (info->fd_temp != -1) & (info->fd_fan != -1);
Tonga_POWERPLAYTABLE *header = (Tonga_POWERPLAYTABLE*) info->pptable;
COMMON_TABLE_HEADER *common_hdr = (COMMON_TABLE_HEADER*) &header->sHeader;
//enable pptable support only for Tonga, Polaris and Fiji (?)
if (info->fd_pptable != -1 &&
(common_hdr->ucTableFormatRevision != 7 || common_hdr->ucTableContentRevision != 1 || header->ucTableRevision != 0)) {
close(info->fd_pptable);
info->fd_pptable = -1;
applog(LOG_WARNING, "No sysfs pptable support for GPU%d (%s)", i, gpus[i].name);
}
if (gpus[i].has_sysfs_hwcontrols)
has_sysfs_hwcontrols = true;
}
initialized = true;
return has_sysfs_hwcontrols;
}