Loading...
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 | /* SPDX-License-Identifier: MIT */ /* * Copyright 2023 Advanced Micro Devices, Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. * * Authors: AMD * */ #include "dml2_policy.h" static void get_optimal_ntuple( const struct soc_bounding_box_st *socbb, struct soc_state_bounding_box_st *entry) { if (entry->dcfclk_mhz > 0) { float bw_on_sdp = (float)(entry->dcfclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); entry->fabricclk_mhz = bw_on_sdp / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); entry->dram_speed_mts = bw_on_sdp / (socbb->num_chans * socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); } else if (entry->fabricclk_mhz > 0) { float bw_on_fabric = (float)(entry->fabricclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); entry->dcfclk_mhz = bw_on_fabric / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); entry->dram_speed_mts = bw_on_fabric / (socbb->num_chans * socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); } else if (entry->dram_speed_mts > 0) { float bw_on_dram = (float)(entry->dram_speed_mts * socbb->num_chans * socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); entry->fabricclk_mhz = bw_on_dram / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); entry->dcfclk_mhz = bw_on_dram / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); } } static float calculate_net_bw_in_mbytes_sec(const struct soc_bounding_box_st *socbb, struct soc_state_bounding_box_st *entry) { float memory_bw_mbytes_sec = (float)(entry->dram_speed_mts * socbb->num_chans * socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); float fabric_bw_mbytes_sec = (float)(entry->fabricclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); float sdp_bw_mbytes_sec = (float)(entry->dcfclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); float limiting_bw_mbytes_sec = memory_bw_mbytes_sec; if (fabric_bw_mbytes_sec < limiting_bw_mbytes_sec) limiting_bw_mbytes_sec = fabric_bw_mbytes_sec; if (sdp_bw_mbytes_sec < limiting_bw_mbytes_sec) limiting_bw_mbytes_sec = sdp_bw_mbytes_sec; return limiting_bw_mbytes_sec; } static void insert_entry_into_table_sorted(const struct soc_bounding_box_st *socbb, struct soc_states_st *table, struct soc_state_bounding_box_st *entry) { int index = 0; int i = 0; float net_bw_of_new_state = 0; get_optimal_ntuple(socbb, entry); if (table->num_states == 0) { index = 0; } else { net_bw_of_new_state = calculate_net_bw_in_mbytes_sec(socbb, entry); while (net_bw_of_new_state > calculate_net_bw_in_mbytes_sec(socbb, &table->state_array[index])) { index++; if (index >= (int) table->num_states) break; } for (i = table->num_states; i > index; i--) { table->state_array[i] = table->state_array[i - 1]; } //ASSERT(index < MAX_CLK_TABLE_SIZE); } table->state_array[index] = *entry; table->state_array[index].dcfclk_mhz = (int)entry->dcfclk_mhz; table->state_array[index].fabricclk_mhz = (int)entry->fabricclk_mhz; table->state_array[index].dram_speed_mts = (int)entry->dram_speed_mts; table->num_states++; } static void remove_entry_from_table_at_index(struct soc_states_st *table, unsigned int index) { int i; if (table->num_states == 0) return; for (i = index; i < (int) table->num_states - 1; i++) { table->state_array[i] = table->state_array[i + 1]; } memset(&table->state_array[--table->num_states], 0, sizeof(struct soc_state_bounding_box_st)); } int dml2_policy_build_synthetic_soc_states(struct dml2_policy_build_synthetic_soc_states_scratch *s, struct dml2_policy_build_synthetic_soc_states_params *p) { int i, j; unsigned int min_fclk_mhz = p->in_states->state_array[0].fabricclk_mhz; unsigned int min_dcfclk_mhz = p->in_states->state_array[0].dcfclk_mhz; unsigned int min_socclk_mhz = p->in_states->state_array[0].socclk_mhz; int max_dcfclk_mhz = 0, max_dispclk_mhz = 0, max_dppclk_mhz = 0, max_phyclk_mhz = 0, max_dtbclk_mhz = 0, max_fclk_mhz = 0, max_uclk_mhz = 0, max_socclk_mhz = 0; int num_uclk_dpms = 0, num_fclk_dpms = 0; for (i = 0; i < __DML_MAX_STATE_ARRAY_SIZE__; i++) { if (p->in_states->state_array[i].dcfclk_mhz > max_dcfclk_mhz) max_dcfclk_mhz = (int) p->in_states->state_array[i].dcfclk_mhz; if (p->in_states->state_array[i].fabricclk_mhz > max_fclk_mhz) max_fclk_mhz = (int) p->in_states->state_array[i].fabricclk_mhz; if (p->in_states->state_array[i].socclk_mhz > max_socclk_mhz) max_socclk_mhz = (int) p->in_states->state_array[i].socclk_mhz; if (p->in_states->state_array[i].dram_speed_mts > max_uclk_mhz) max_uclk_mhz = (int) p->in_states->state_array[i].dram_speed_mts; if (p->in_states->state_array[i].dispclk_mhz > max_dispclk_mhz) max_dispclk_mhz = (int) p->in_states->state_array[i].dispclk_mhz; if (p->in_states->state_array[i].dppclk_mhz > max_dppclk_mhz) max_dppclk_mhz = (int) p->in_states->state_array[i].dppclk_mhz; if (p->in_states->state_array[i].phyclk_mhz > max_phyclk_mhz) max_phyclk_mhz = (int)p->in_states->state_array[i].phyclk_mhz; if (p->in_states->state_array[i].dtbclk_mhz > max_dtbclk_mhz) max_dtbclk_mhz = (int)p->in_states->state_array[i].dtbclk_mhz; if (p->in_states->state_array[i].fabricclk_mhz > 0) num_fclk_dpms++; if (p->in_states->state_array[i].dram_speed_mts > 0) num_uclk_dpms++; } if (!max_dcfclk_mhz || !max_dispclk_mhz || !max_dppclk_mhz || !max_phyclk_mhz || !max_dtbclk_mhz) return -1; p->out_states->num_states = 0; s->entry = p->in_states->state_array[0]; s->entry.dispclk_mhz = max_dispclk_mhz; s->entry.dppclk_mhz = max_dppclk_mhz; s->entry.dtbclk_mhz = max_dtbclk_mhz; s->entry.phyclk_mhz = max_phyclk_mhz; s->entry.dscclk_mhz = max_dispclk_mhz / 3; s->entry.phyclk_mhz = max_phyclk_mhz; s->entry.dtbclk_mhz = max_dtbclk_mhz; // Insert all the DCFCLK STAs first for (i = 0; i < p->num_dcfclk_stas; i++) { s->entry.dcfclk_mhz = p->dcfclk_stas_mhz[i]; s->entry.fabricclk_mhz = 0; s->entry.dram_speed_mts = 0; if (i > 0) s->entry.socclk_mhz = max_socclk_mhz; insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); } // Insert the UCLK DPMS for (i = 0; i < num_uclk_dpms; i++) { s->entry.dcfclk_mhz = 0; s->entry.fabricclk_mhz = 0; s->entry.dram_speed_mts = p->in_states->state_array[i].dram_speed_mts; if (i == 0) { s->entry.socclk_mhz = min_socclk_mhz; } else { s->entry.socclk_mhz = max_socclk_mhz; } insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); } // Insert FCLK DPMs (if present) if (num_fclk_dpms > 2) { for (i = 0; i < num_fclk_dpms; i++) { s->entry.dcfclk_mhz = 0; s->entry.fabricclk_mhz = p->in_states->state_array[i].fabricclk_mhz; s->entry.dram_speed_mts = 0; insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); } } // Add max FCLK else { s->entry.dcfclk_mhz = 0; s->entry.fabricclk_mhz = p->in_states->state_array[num_fclk_dpms - 1].fabricclk_mhz; s->entry.dram_speed_mts = 0; insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); } // Remove states that require higher clocks than are supported for (i = p->out_states->num_states - 1; i >= 0; i--) { if (p->out_states->state_array[i].dcfclk_mhz > max_dcfclk_mhz || p->out_states->state_array[i].fabricclk_mhz > max_fclk_mhz || p->out_states->state_array[i].dram_speed_mts > max_uclk_mhz) remove_entry_from_table_at_index(p->out_states, i); } // At this point, the table contains all "points of interest" based on // DPMs from PMFW, and STAs. Table is sorted by BW, and all clock // ratios (by derate, are exact). // Round up UCLK to DPMs for (i = p->out_states->num_states - 1; i >= 0; i--) { for (j = 0; j < num_uclk_dpms; j++) { if (p->in_states->state_array[j].dram_speed_mts >= p->out_states->state_array[i].dram_speed_mts) { p->out_states->state_array[i].dram_speed_mts = p->in_states->state_array[j].dram_speed_mts; break; } } } // If FCLK is coarse grained, round up to next DPMs if (num_fclk_dpms > 2) { for (i = p->out_states->num_states - 1; i >= 0; i--) { for (j = 0; j < num_fclk_dpms; j++) { if (p->in_states->state_array[j].fabricclk_mhz >= p->out_states->state_array[i].fabricclk_mhz) { p->out_states->state_array[i].fabricclk_mhz = p->in_states->state_array[j].fabricclk_mhz; break; } } } } // Clamp to min FCLK/DCFCLK for (i = p->out_states->num_states - 1; i >= 0; i--) { if (p->out_states->state_array[i].fabricclk_mhz < min_fclk_mhz) { p->out_states->state_array[i].fabricclk_mhz = min_fclk_mhz; } if (p->out_states->state_array[i].dcfclk_mhz < min_dcfclk_mhz) { p->out_states->state_array[i].dcfclk_mhz = min_dcfclk_mhz; } } // Remove duplicate states, note duplicate states are always neighbouring since table is sorted. i = 0; while (i < (int) p->out_states->num_states - 1) { if (p->out_states->state_array[i].dcfclk_mhz == p->out_states->state_array[i + 1].dcfclk_mhz && p->out_states->state_array[i].fabricclk_mhz == p->out_states->state_array[i + 1].fabricclk_mhz && p->out_states->state_array[i].dram_speed_mts == p->out_states->state_array[i + 1].dram_speed_mts) remove_entry_from_table_at_index(p->out_states, i); else i++; } return 0; } void build_unoptimized_policy_settings(enum dml_project_id project, struct dml_mode_eval_policy_st *policy) { for (int i = 0; i < __DML_NUM_PLANES__; i++) { policy->MPCCombineUse[i] = dml_mpc_as_needed_for_voltage; // TOREVIEW: Is this still needed? When is MPCC useful for pstate given CRB? policy->ODMUse[i] = dml_odm_use_policy_combine_as_needed; policy->ImmediateFlipRequirement[i] = dml_immediate_flip_required; policy->AllowForPStateChangeOrStutterInVBlank[i] = dml_prefetch_support_uclk_fclk_and_stutter_if_possible; } /* Change the default policy initializations as per spreadsheet. We might need to * review and change them later on as per Jun's earlier comments. */ policy->UseUnboundedRequesting = dml_unbounded_requesting_enable; policy->UseMinimumRequiredDCFCLK = false; policy->DRAMClockChangeRequirementFinal = true; // TOREVIEW: What does this mean? policy->FCLKChangeRequirementFinal = true; // TOREVIEW: What does this mean? policy->USRRetrainingRequiredFinal = true; policy->EnhancedPrefetchScheduleAccelerationFinal = true; // TOREVIEW: What does this mean? policy->NomDETInKByteOverrideEnable = false; policy->NomDETInKByteOverrideValue = 0; policy->DCCProgrammingAssumesScanDirectionUnknownFinal = true; policy->SynchronizeTimingsFinal = true; policy->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = true; policy->AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported = true; // TOREVIEW: What does this mean? policy->AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported = true; // TOREVIEW: What does this mean? if (project == dml_project_dcn35 || project == dml_project_dcn351) { policy->DCCProgrammingAssumesScanDirectionUnknownFinal = false; policy->EnhancedPrefetchScheduleAccelerationFinal = 0; policy->AllowForPStateChangeOrStutterInVBlankFinal = dml_prefetch_support_uclk_fclk_and_stutter_if_possible; /*new*/ policy->UseOnlyMaxPrefetchModes = 1; } } |