From 7b3d554117219b1300a91b39785e6d981b378364 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=5BQuy=20Anh=5D=20=C2=ABElliot=C2=BB=20Nguyen?= <23906-nguyenquy@users.noreply.gitlab.inf.ethz.ch> Date: Mon, 25 May 2026 00:59:19 +0200 Subject: [PATCH] AVX-512 First Attempt --- .clangd | 12 + Makefile | 25 +- headers/board_reduced.hpp | 3 - headers/constants.hpp | 17 +- variants/variant_avx512_1.cpp | 423 ++++++++++++++++++++++++++++++++++ 5 files changed, 467 insertions(+), 13 deletions(-) create mode 100644 .clangd create mode 100644 variants/variant_avx512_1.cpp diff --git a/.clangd b/.clangd new file mode 100644 index 0000000..76c4b24 --- /dev/null +++ b/.clangd @@ -0,0 +1,12 @@ +CompileFlags: + Add: + [ + '-O3', + '-march=native', + '-std=c++23', + '-Wal', + '-Wextra', + '-flto', + '-lsleef', + '-I/usr/include/x86_64-linux-gnu' + ] diff --git a/Makefile b/Makefile index 92ce00a..d1c0939 100644 --- a/Makefile +++ b/Makefile @@ -47,6 +47,13 @@ CXXFLAGS := -O3 -march=native -std=c++23 -Wall -Wextra -flto VARIANT ?= baseline +# sleef is only needed for the avx512 variant (used by SLEEF intrinsics there) +ifeq ($(ARCH),x86_64) +ifneq ($(filter avx512_%,$(VARIANT)),) +CXXFLAGS += -lsleef +endif +endif + VARIANT_SRC := ./variants/variant_$(VARIANT).cpp ifneq ($(filter float_%,$(VARIANT)),) @@ -95,7 +102,7 @@ ROOFLINE ?= 0 CXXFLAGS += -DROOFLINE=$(ROOFLINE) # Benchmark suite parameters. -BENCH_SIZES ?= 50 100 200 500 1000 2000 5000 10000 +BENCH_SIZES ?= 50 100 200 500 1000 BENCH_ITER ?= 1000 # Verification parameters @@ -116,7 +123,7 @@ all: main # make main : compile the main executable with benchmarking disabled. main: clean - $(CXX) $(CXXFLAGS) $(SRC) -o $(TARGET) + $(CXX) $(SRC) -o $(TARGET) $(CXXFLAGS) # make run : compile the code and run the simulation without generating a video. run: main @@ -137,7 +144,7 @@ verify: main # make bench : compile the main executable with benchmarking enabled. bench: CXXFLAGS += -DBENCHMARK=1 bench: clean - $(SUDO_BENCH) $(CXX) $(CXXFLAGS) $(SRC) -o $(TARGET) + $(SUDO_BENCH) $(CXX) $(SRC) -o $(TARGET) $(CXXFLAGS) # make run-bench : compile and run benchmark measurements. run-bench: bench @@ -160,7 +167,7 @@ video: run # make test : compile and run all unit tests. test: clean - $(CXX) $(CXXFLAGS) $(TEST_SRC) $(TEST_LIBS) -o $(TEST_TARGET) + $(CXX) $(TEST_SRC) $(TEST_LIBS) -o $(TEST_TARGET) $(CXXFLAGS) ./$(TEST_TARGET) # make asm : compile and generate assembly files. @@ -172,19 +179,19 @@ ASM_SRCS := $(ASM_CPP_SRCS) $(ASM_HPP_SRCS) ASM_TARGETS := $(patsubst %.cpp,%.s,$(patsubst %.hpp,%.s,$(ASM_SRCS))) asm: clean - $(foreach src,$(ASM_CPP_SRCS),$(CXX) -O3 -march=native -std=c++23 -Wall -Wextra -DRANDOM_SEED=$(RANDOM_SEED) -S $(src) -o $(src:.cpp=.s); ) - $(foreach src,$(ASM_HPP_SRCS),$(CXX) -O3 -march=native -std=c++23 -Wall -Wextra -DRANDOM_SEED=$(RANDOM_SEED) -x c++ -S $(src) -o $(src:.hpp=.s); ) + $(foreach src,$(ASM_CPP_SRCS),$(CXX) -S $(src) -o $(src:.cpp=.s) $(CXXFLAGS); ) + $(foreach src,$(ASM_HPP_SRCS),$(CXX) -x c++ -S $(src) -o $(src:.hpp=.s) $(CXXFLAGS); ) %.s: %.cpp $(HEADERS) - $(CXX) -O3 -march=native -std=c++23 -Wall -Wextra -DRANDOM_SEED=$(RANDOM_SEED) -S $< -o $@ + $(CXX) -S $< -o $@ $(CXXFLAGS) # make clean : delete generated files from previous builds. clean: - rm -f main_* $(TEST_TARGET) *.o *.s + rm -f main_* $(SIZES_TARGET) $(TEST_TARGET) *.o *.s variants/*.s variants/*.o headers/*.s headers/*.o # make clean-all : delete all generated files from previous builds and runs. # WARNING: this deletes all *.csv and *.mp4 files in this directory. # Move any output files you want to keep to another directory before running it. clean-all: clean - rm -f *.csv *.mp4 + rm -f *.csv *.mp4 \ No newline at end of file diff --git a/headers/board_reduced.hpp b/headers/board_reduced.hpp index 2ca4ed9..e2b068d 100644 --- a/headers/board_reduced.hpp +++ b/headers/board_reduced.hpp @@ -1,6 +1,3 @@ -// # Replaced Vector2d with double -// # Predecessor: board.hpp - #pragma once struct Obstacle { diff --git a/headers/constants.hpp b/headers/constants.hpp index 37b7986..129d02f 100644 --- a/headers/constants.hpp +++ b/headers/constants.hpp @@ -2,6 +2,10 @@ #include #include +#ifdef __x86_64__ +#include "immintrin.h" +#endif + // // $\frac{1}{\tau_{\alpha}}$; $\tau_{\alpha} = 0.5s$ // @@ -96,10 +100,21 @@ const double EPS = 1e-9; // Squared numerical tolerance const double EPS_SQ = EPS * EPS; +#ifdef __AVX512F__ +const __m512d LN2_VEC = _mm512_set1_pd(log(2)); +const __m512d ZERO_VEC = _mm512_setzero_pd(); +const __m512d NEG_TWO_SIGMA_INV_VEC = _mm512_set1_pd(NEG_TWO_SIGMA_INV); +const __m512d NEG_V0_AB_DIV_TWO_SIGMA_VEC = _mm512_set1_pd(NEG_V0_AB_div_TWO_SIGMA); +const __m512d c_neg_times_dt_vec = _mm512_set1_pd(c_neg_times_dt); +const __m512d cos_phi_inv_vec = _mm512_set1_pd(cos_phi_inv); +const __m512d dt_neg_vec = _mm512_set1_pd(dt_neg); +#endif + // Cutoff for pedestrian-pedestrian repulsion const double CUTOFF = 12.0; const double CUTOFF_SQ = CUTOFF * CUTOFF; + const double cos_phi_sq = cos_phi * cos_phi; -const size_t CACHE_BLOCK_SIZE = std::hardware_destructive_interference_size; \ No newline at end of file +const size_t CACHE_BLOCK_SIZE = std::hardware_destructive_interference_size; diff --git a/variants/variant_avx512_1.cpp b/variants/variant_avx512_1.cpp new file mode 100644 index 0000000..8262058 --- /dev/null +++ b/variants/variant_avx512_1.cpp @@ -0,0 +1,423 @@ +// # This file attempts to use AVX-512 intrinsics to speed up computation. +// # Predecessor: variant_inlining_7.cpp +#ifdef __AVX512F__ + +#include "../headers/board_reduced.hpp" +#include "../headers/constants.hpp" +#include "../headers/pedestrian_reduced.hpp" + +#include "immintrin.h" +#include "sleef.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define SEED 42 +#define BOARD_WIDTH 10.0 +#define BOARD_LENGTH 50.0 +#define WALL_EPS 1e-3 +#define BOARD_WIDTH_MINUS_WALL_EPS 9.999 + +#ifndef RANDOM_SEED +#define RANDOM_SEED 0 +#endif + +#ifndef BENCHMARK +#define BENCHMARK 0 +#endif + +using namespace std; + +Pedestrian *init_pedestrians(int n_p) { +#if RANDOM_SEED + random_device rd; + mt19937 gen(rd()); +#else + mt19937 gen(SEED); +#endif + + normal_distribution velocity_gen(v0_mean, v0_std_dev); + uniform_real_distribution y_gen(0.3, BOARD_WIDTH - 0.3); + uniform_real_distribution x_gen_left(2.0, BOARD_LENGTH / 2.0); + uniform_real_distribution x_gen_right(BOARD_LENGTH / 2.0, BOARD_LENGTH - 2.0); + + Pedestrian *P = new Pedestrian[n_p]; + + int n_p_half = n_p / 2; + + for (int i = 0; i < n_p_half; i++) { + double x = x_gen_left(gen); + double y = y_gen(gen); + + double start_pos_x = x; + double start_pos_y = y; + + double destination_x = BOARD_LENGTH + 30.0; + double destination_y = y; + + double init_e_x = destination_x - start_pos_x; + double init_e_y = destination_y - start_pos_y; + double init_e_norm = sqrt(init_e_x * init_e_x + init_e_y * init_e_y); + double init_e_unit_x = init_e_x / init_e_norm; + double init_e_unit_y = init_e_y / init_e_norm; + + double desired_v0; + do { + desired_v0 = velocity_gen(gen); + } while (desired_v0 <= 0.1); + + double v_max = 1.3 * desired_v0; + + P[i] = Pedestrian(0, 0, 0, start_pos_x, start_pos_y, init_e_unit_x, init_e_unit_y, v_max, desired_v0, destination_x, destination_y); + } + + for (int i = n_p_half; i < n_p; i++) { + double x = x_gen_right(gen); + double y = y_gen(gen); + + double start_pos_x = x; + double start_pos_y = y; + double destination_x = -30; + double destination_y = y; + + double init_e_x = destination_x - start_pos_x; + double init_e_y = destination_y - start_pos_y; + double init_e_norm = sqrt(init_e_x * init_e_x + init_e_y * init_e_y); + double init_e_unit_x = init_e_x / init_e_norm; + double init_e_unit_y = init_e_y / init_e_norm; + + double desired_v0; + do { + desired_v0 = velocity_gen(gen); + } while (desired_v0 <= 0.1); + + double v_max = 1.3 * desired_v0; + + P[i] = Pedestrian(1, 0, 0, start_pos_x, start_pos_y, init_e_unit_x, init_e_unit_y, v_max, desired_v0, destination_x, destination_y); + } + + return P; +} + +Board init_board_without_obstacles() { return Board(BOARD_LENGTH, BOARD_WIDTH, 0, nullptr); } + +int run_simulation(Pedestrian *pedestrians, Board &board, int n_p, int n_iter, const char *dump_path) { + +#if BENCHMARK + (void)dump_path; +#endif + +#if !BENCHMARK + const int dump_interval = 1; + + vector> samples; + + if (dump_path) { + samples.reserve(static_cast(n_iter / dump_interval + 1) * n_p); + + for (int i = 0; i < n_p; i++) { + Pedestrian &p = pedestrians[i]; + samples.emplace_back(0, i, p.curr_pos_x, p.curr_pos_y, p.e_x, p.e_y, p.color); + } + } +#endif + + for (int t = 0; t < n_iter; t++) { + for (int i = 0; i < n_p; i++) { + Pedestrian &p = pedestrians[i]; + + double desired_velocity_x = p.e_x * p.desired_v0; + double desired_velocity_y = p.e_y * p.desired_v0; + double f_alpha_dt_x = (desired_velocity_x - p.curr_vel_x) * dt_times_tau_alpha_inv; + double f_alpha_dt_y = (desired_velocity_y - p.curr_vel_y) * dt_times_tau_alpha_inv; + + __m512d desired_velocity_x_vec = _mm512_set1_pd(desired_velocity_x); + __m512d desired_velocity_y_vec = _mm512_set1_pd(desired_velocity_y); + __m512d f_alpha_dt_x_vec = _mm512_set1_pd(f_alpha_dt_x); + __m512d f_alpha_dt_y_vec = _mm512_set1_pd(f_alpha_dt_y); + + __m512d alpha_curr_pos_x_vec = _mm512_set1_pd(p.curr_pos_x); + __m512d alpha_curr_pos_y_vec = _mm512_set1_pd(p.curr_pos_y); + __m512d alpha_e_x_vec = _mm512_set1_pd(p.e_x); + __m512d alpha_e_y_vec = _mm512_set1_pd(p.e_y); + + int j; + + // # ############################################################################################################################################################################################ + + for (j = 0; j < n_p - 7; j += 8) { + + if (j <= i && i <= j + 7) { + + for (int jj = j; jj < j + 8; jj++) { + + if (jj == i) { + continue; + } + + double r_ab_x = p.curr_pos_x - pedestrians[jj].curr_pos_x; + double r_ab_y = p.curr_pos_y - pedestrians[jj].curr_pos_y; + + double r_ab_norm = sqrt(r_ab_x * r_ab_x + r_ab_y * r_ab_y); + + double r_ab2_x = r_ab_x - pedestrians[jj].s_b_e_x; + double r_ab2_y = r_ab_y - pedestrians[jj].s_b_e_y; + + double r_ab2_norm = sqrt(r_ab2_x * r_ab2_x + r_ab2_y * r_ab2_y); + double r_ab_norm_sum = r_ab_norm + r_ab2_norm; + double two_b_sq = r_ab_norm_sum * r_ab_norm_sum - pedestrians[jj].s_b_sq; + + double two_b = sqrt(two_b_sq); + double scalar_mult_common = (r_ab_norm_sum / (two_b * r_ab_norm * r_ab2_norm)) * (NEG_V0_AB_div_TWO_SIGMA * exp(two_b * NEG_TWO_SIGMA_INV)); + double scalar_mult_r_ab = scalar_mult_common * r_ab2_norm; + double scalar_mult_r_ab2 = scalar_mult_common * r_ab_norm; + + double f_alpha_beta_neg_x = r_ab_x * scalar_mult_r_ab + r_ab2_x * scalar_mult_r_ab2; + double f_alpha_beta_neg_y = r_ab_y * scalar_mult_r_ab + r_ab2_y * scalar_mult_r_ab2; + + double w_val; + double a = (p.e_x * f_alpha_beta_neg_x + p.e_y * f_alpha_beta_neg_y) * cos_phi_inv; + double b_sq = f_alpha_beta_neg_x * f_alpha_beta_neg_x + f_alpha_beta_neg_y * f_alpha_beta_neg_y; + + w_val = dt_neg; + if (a >= 0 && (a * a > b_sq)) { + w_val = c_neg_times_dt; + } + + f_alpha_dt_x = f_alpha_dt_x + f_alpha_beta_neg_x * w_val; + f_alpha_dt_y = f_alpha_dt_y + f_alpha_beta_neg_y * w_val; + } + + } else { + __m512d beta_curr_pos_x_vec = _mm512_setr_pd(pedestrians[j].curr_pos_x, pedestrians[j + 1].curr_pos_x, pedestrians[j + 2].curr_pos_x, pedestrians[j + 3].curr_pos_x, pedestrians[j + 4].curr_pos_x, pedestrians[j + 5].curr_pos_x, pedestrians[j + 6].curr_pos_x, pedestrians[j + 7].curr_pos_x); + + __m512d beta_curr_pos_y_vec = _mm512_setr_pd(pedestrians[j].curr_pos_y, pedestrians[j + 1].curr_pos_y, pedestrians[j + 2].curr_pos_y, pedestrians[j + 3].curr_pos_y, pedestrians[j + 4].curr_pos_y, pedestrians[j + 5].curr_pos_y, pedestrians[j + 6].curr_pos_y, pedestrians[j + 7].curr_pos_y); + + __m512d r_ab_x_vec = _mm512_sub_pd(alpha_curr_pos_x_vec, beta_curr_pos_x_vec); + __m512d r_ab_y_vec = _mm512_sub_pd(alpha_curr_pos_y_vec, beta_curr_pos_y_vec); + __m512d r_ab_y_mul_r_ab_y_vec = _mm512_mul_pd(r_ab_y_vec, r_ab_y_vec); + + __m512d r_ab_norm_sq_vec = _mm512_fmadd_pd(r_ab_x_vec, r_ab_x_vec, r_ab_y_mul_r_ab_y_vec); + __m512d r_ab_norm_vec = _mm512_sqrt_pd(r_ab_norm_sq_vec); + + __m512d beta_s_b_e_x_vec = _mm512_setr_pd(pedestrians[j].s_b_e_x, pedestrians[j + 1].s_b_e_x, pedestrians[j + 2].s_b_e_x, pedestrians[j + 3].s_b_e_x, pedestrians[j + 4].s_b_e_x, pedestrians[j + 5].s_b_e_x, pedestrians[j + 6].s_b_e_x, pedestrians[j + 7].s_b_e_x); + __m512d r_ab2_x_vec = _mm512_sub_pd(r_ab_x_vec, beta_s_b_e_x_vec); + + __m512d beta_s_b_e_y_vec = _mm512_setr_pd(pedestrians[j].s_b_e_y, pedestrians[j + 1].s_b_e_y, pedestrians[j + 2].s_b_e_y, pedestrians[j + 3].s_b_e_y, pedestrians[j + 4].s_b_e_y, pedestrians[j + 5].s_b_e_y, pedestrians[j + 6].s_b_e_y, pedestrians[j + 7].s_b_e_y); + __m512d r_ab2_y_vec = _mm512_sub_pd(r_ab_y_vec, beta_s_b_e_y_vec); + + __m512d r_ab2_y_mul_r_ab2_y_vec = _mm512_mul_pd(r_ab2_y_vec, r_ab2_y_vec); + __m512d r_ab2_norm_sq_vec = _mm512_fmadd_pd(r_ab2_x_vec, r_ab2_x_vec, r_ab2_y_mul_r_ab2_y_vec); + __m512d r_ab2_norm_vec = _mm512_sqrt_pd(r_ab2_norm_sq_vec); + + __m512d r_ab_norm_sum_vec = _mm512_add_pd(r_ab_norm_vec, r_ab2_norm_vec); + + __m512d beta_s_b_sq_vec = _mm512_setr_pd(pedestrians[j].s_b_sq, pedestrians[j + 1].s_b_sq, pedestrians[j + 2].s_b_sq, pedestrians[j + 3].s_b_sq, pedestrians[j + 4].s_b_sq, pedestrians[j + 5].s_b_sq, pedestrians[j + 6].s_b_sq, pedestrians[j + 7].s_b_sq); + __m512d two_b_sq_vec = _mm512_fmsub_pd(r_ab_norm_sum_vec, r_ab_norm_sum_vec, beta_s_b_sq_vec); + + __m512d two_b_vec = _mm512_sqrt_pd(two_b_sq_vec); + + __m512d two_b_mul_r_ab_norm_vec = _mm512_mul_pd(two_b_vec, r_ab_norm_vec); + __m512d denom_vec = _mm512_mul_pd(two_b_mul_r_ab_norm_vec, r_ab2_norm_vec); + __m512d scalar_mult_common_left_vec = _mm512_div_pd(r_ab_norm_sum_vec, denom_vec); + __m512d two_b_mul_NEG_TWO_SIGMA_INV_vec = _mm512_mul_pd(two_b_vec, NEG_TWO_SIGMA_INV_VEC); + + __m512d e_vec = Sleef_expd8_u10(two_b_mul_NEG_TWO_SIGMA_INV_vec); + + __m512d scalar_mult_common_right_vec = _mm512_mul_pd(NEG_V0_AB_DIV_TWO_SIGMA_VEC, e_vec); + __m512d scalar_mult_common_vec = _mm512_mul_pd(scalar_mult_common_left_vec, scalar_mult_common_right_vec); + + __m512d scalar_mult_r_ab_vec = _mm512_mul_pd(scalar_mult_common_vec, r_ab2_norm_vec); + __m512d scalar_mult_r_ab2_vec = _mm512_mul_pd(scalar_mult_common_vec, r_ab_norm_vec); + + __m512d r_ab2_x_mul_scalar_mult_r_ab2_vec = _mm512_mul_pd(r_ab2_x_vec, scalar_mult_r_ab2_vec); + __m512d r_ab2_y_mul_scalar_mult_r_ab2_vec = _mm512_mul_pd(r_ab2_y_vec, scalar_mult_r_ab2_vec); + + __m512d f_alpha_beta_neg_x_vec = _mm512_fmadd_pd(r_ab_x_vec, scalar_mult_r_ab_vec, r_ab2_x_mul_scalar_mult_r_ab2_vec); + __m512d f_alpha_beta_neg_y_vec = _mm512_fmadd_pd(r_ab_y_vec, scalar_mult_r_ab_vec, r_ab2_y_mul_scalar_mult_r_ab2_vec); + + __m512d rgt_vec = _mm512_mul_pd(alpha_e_y_vec, f_alpha_beta_neg_y_vec); + __m512d lft_add_rgt_vec = _mm512_fmadd_pd(alpha_e_x_vec, f_alpha_beta_neg_x_vec, rgt_vec); + __m512d a_vec = _mm512_mul_pd(lft_add_rgt_vec, cos_phi_inv_vec); + + __m512d f_alpha_beta_neg_y_sq_vec = _mm512_mul_pd(f_alpha_beta_neg_y_vec, f_alpha_beta_neg_y_vec); + __m512d b_sq_vec = _mm512_fmadd_pd(f_alpha_beta_neg_x_vec, f_alpha_beta_neg_x_vec, f_alpha_beta_neg_y_sq_vec); + + __m512d w_val_vec = dt_neg_vec; + __mmask8 a_ge_zero_mask = _mm512_cmp_pd_mask(a_vec, ZERO_VEC, _CMP_GE_OQ); + __m512d a_sq_vec = _mm512_mul_pd(a_vec, a_vec); + __mmask8 a_sq_gt_b_sq_mask = _mm512_cmp_pd_mask(a_sq_vec, b_sq_vec, _CMP_GT_OQ); + __mmask8 cond_mask = _kand_mask8(a_ge_zero_mask, a_sq_gt_b_sq_mask); + w_val_vec = _mm512_mask_mov_pd(w_val_vec, cond_mask, c_neg_times_dt_vec); + + __m512d x_term_vec = _mm512_mul_pd(f_alpha_beta_neg_x_vec, w_val_vec); + __m512d y_term_vec = _mm512_mul_pd(f_alpha_beta_neg_y_vec, w_val_vec); + + f_alpha_dt_x += _mm512_reduce_add_pd(x_term_vec); + f_alpha_dt_y += _mm512_reduce_add_pd(y_term_vec); + } + } + + // # ############################################################################################################################################################################################ + + for (; j < n_p; j++) { + + if (j == i) { + continue; + } + + double r_ab_x = p.curr_pos_x - pedestrians[j].curr_pos_x; + double r_ab_y = p.curr_pos_y - pedestrians[j].curr_pos_y; + + double r_ab_norm = sqrt(r_ab_x * r_ab_x + r_ab_y * r_ab_y); + + double r_ab2_x = r_ab_x - pedestrians[j].s_b_e_x; + double r_ab2_y = r_ab_y - pedestrians[j].s_b_e_y; + + double r_ab2_norm = sqrt(r_ab2_x * r_ab2_x + r_ab2_y * r_ab2_y); + double r_ab_norm_sum = r_ab_norm + r_ab2_norm; + double two_b_sq = r_ab_norm_sum * r_ab_norm_sum - pedestrians[j].s_b_sq; + double two_b = sqrt(two_b_sq); + double scalar_mult_common = (r_ab_norm_sum / (two_b * r_ab_norm * r_ab2_norm)) * (NEG_V0_AB_div_TWO_SIGMA * exp(two_b * NEG_TWO_SIGMA_INV)); + double scalar_mult_r_ab = scalar_mult_common * r_ab2_norm; + double scalar_mult_r_ab2 = scalar_mult_common * r_ab_norm; + + double f_alpha_beta_neg_x = r_ab_x * scalar_mult_r_ab + r_ab2_x * scalar_mult_r_ab2; + double f_alpha_beta_neg_y = r_ab_y * scalar_mult_r_ab + r_ab2_y * scalar_mult_r_ab2; + + double w_val; + double a = (p.e_x * f_alpha_beta_neg_x + p.e_y * f_alpha_beta_neg_y) * cos_phi_inv; + double b_sq = f_alpha_beta_neg_x * f_alpha_beta_neg_x + f_alpha_beta_neg_y * f_alpha_beta_neg_y; + + w_val = dt_neg; + if (a >= 0 && (a * a > b_sq)) { + w_val = c_neg_times_dt; + } + + f_alpha_dt_x = f_alpha_dt_x + f_alpha_beta_neg_x * w_val; + f_alpha_dt_y = f_alpha_dt_y + f_alpha_beta_neg_y * w_val; + } + + // # ############################################################################################################################################################################################ + + for (int k = 0; k < board.n_obstacles; k++) { + + double v1_x = p.curr_pos_x - board.obstacles[k].start_x; + double v1_y = p.curr_pos_y - board.obstacles[k].start_y; + + double projection = (v1_x * board.obstacles[k].obstacle_v_x + v1_y * board.obstacles[k].obstacle_v_y) * board.obstacles[k].length_sq_inv; + + if (projection < 0.0) + projection = 0.0; + else if (projection > 1.0) + projection = 1.0; + + double r_aB_x = p.curr_pos_x - (board.obstacles[k].start_x + board.obstacles[k].obstacle_v_x * projection); + double r_aB_y = p.curr_pos_y - (board.obstacles[k].start_y + board.obstacles[k].obstacle_v_y * projection); + + double r_aB_norm = sqrt(r_aB_x * r_aB_x + r_aB_y * r_aB_y); + double scalar_mult = (NEG_U0_AB_div_R * exp(r_aB_norm * NEG_R_INV)) / r_aB_norm; + double f_alpha_B_neg_x = r_aB_x * scalar_mult; + double f_alpha_B_neg_y = r_aB_y * scalar_mult; + + double w_val; + double a = (p.e_x * f_alpha_B_neg_x + p.e_y * f_alpha_B_neg_y) * cos_phi_inv; + double b_sq = f_alpha_B_neg_x * f_alpha_B_neg_x + f_alpha_B_neg_y * f_alpha_B_neg_y; + + w_val = dt_neg; + if (a >= 0 && (a * a > b_sq)) { + w_val = c_neg_times_dt; + } + + double f_alpha_B_dt_x = f_alpha_B_neg_x * w_val; + double f_alpha_B_dt_y = f_alpha_B_neg_y * w_val; + + f_alpha_dt_x += f_alpha_B_dt_x; + f_alpha_dt_y += f_alpha_B_dt_y; + } + + p.w_x += f_alpha_dt_x; + p.w_y += f_alpha_dt_y; + } + + // # ############################################################################################################################################################################################ + + for (int i = 0; i < n_p; i++) { + Pedestrian &p = pedestrians[i]; + + double w_norm = sqrt(p.w_x * p.w_x + p.w_y * p.w_y); + + double correction_factor = min(1.0, p.v_max / w_norm); + + p.curr_vel_x = p.w_x * correction_factor; + p.curr_vel_y = p.w_y * correction_factor; + + p.curr_pos_x = p.curr_pos_x + p.curr_vel_x * dt; + p.curr_pos_y = p.curr_pos_y + p.curr_vel_y * dt; + + if (p.color == 0 && p.curr_pos_x > (BOARD_LENGTH - 1.5)) { + p.curr_pos_x = 1.0; + p.destination_y = p.curr_pos_y; + p.w_x = p.desired_v0; + p.w_y = 0.0; + p.curr_vel_x = p.w_x; + p.curr_vel_y = p.w_y; + } + if (p.color == 1 && p.curr_pos_x < 1.5) { + p.curr_pos_x = BOARD_LENGTH - 1.0; + p.destination_y = p.curr_pos_y; + p.w_x = -p.desired_v0; + p.w_y = 0.0; + p.curr_vel_x = p.w_x; + p.curr_vel_y = p.w_y; + } + + p.curr_pos_y = clamp(p.curr_pos_y, WALL_EPS, BOARD_WIDTH_MINUS_WALL_EPS); + + double e_x = p.destination_x - p.curr_pos_x; + double e_y = p.destination_y - p.curr_pos_y; + + double e_norm_inv = 1.0 / sqrt(e_x * e_x + e_y * e_y); + + p.e_x = e_x * e_norm_inv; + p.e_y = e_y * e_norm_inv; + + double velocity_norm_sq = p.curr_vel_x * p.curr_vel_x + p.curr_vel_y * p.curr_vel_y; + + double velocity_norm = sqrt(velocity_norm_sq); + + p.s_b_sq = velocity_norm_sq * delta_t_sq; + + p.s_b_e_x = p.e_x * (velocity_norm * delta_t); + p.s_b_e_y = p.e_y * (velocity_norm * delta_t); + +#if !BENCHMARK + if (dump_path && (t % dump_interval == 0)) { + samples.emplace_back(t + 1, i, p.curr_pos_x, p.curr_pos_y, p.e_x, p.e_y, p.color); + } +#endif + } + } + +#if !BENCHMARK + if (dump_path) { + ofstream out(dump_path); + out << "time,id,x,y,p.e.x,p.e.y,color\n"; + + for (auto &[t, id, x, y, e_x, e_y, c] : samples) { + out << t << ',' << id << ',' << x << ',' << y << ',' << e_x << ',' << e_y << ',' << c << '\n'; + } + } +#endif + + return 0; +} + +#endif