saving current changes

Tests/isa/Cprograms/benchmark/barnes/barnes_hut.c

@@ -0,0 +1,482 @@
+/*
+ * File: barnes_hut.c
+ * --------------
+ * Implements the Barnes Hut algorithm for n-body
+ * simulation with galaxy-like initial conditions.
+ * Uses OpenGl and GLUT for graphics. Written by:
+ * Joel Backsell, Kim Torberntsson & Alexander Bilock.
+ * Source: https://github.com/KimTorberntsson/Barnes-Hut/blob/master/barnes_hut.c
+ */
+
+#include <stdlib.h>
+#include <math.h>
+#include <stdio.h>
+#include <math.h>
+// #include <time.h>
+// #include <GLUT/glut.h>
+
+#include "barnes_hut.h"
+// #include "graphics.h"
+
+void free(void *ptr);
+void *malloc(size_t size);
+
+//Some constants and global variables
+int N = 10;
+const double L = 1, W = 1, dt = 1e-3, alpha = 0.25, V = 50, epsilon = 1e-1, grav = 0.04; //grav should be 100/N
+double *x, *y, *u, *v, *force_x, *force_y, *mass;
+struct node_t *root;
+
+// -------- Modified by Akilan ----------
+
+double my_sqrt(double x) {
+    if (x <= 0) return 0;
+
+    double guess = x;
+    for (int i = 0; i < 10; i++) {
+        guess = 0.5 * (guess + x / guess);
+    }
+    return guess;
+}
+
+#define PI 3.141592653589793
+
+double wrap_angle(double x) {
+    while (x > PI)  x -= 2*PI;
+    while (x < -PI) x += 2*PI;
+    return x;
+}
+
+double my_sin(double x) {
+    x = wrap_angle(x);
+
+    double x2 = x * x;
+    return x
+         - (x * x2) / 6
+         + (x * x2 * x2) / 120
+         - (x * x2 * x2 * x2) / 5040;
+}
+
+double my_cos(double x) {
+    x = wrap_angle(x);
+
+    double x2 = x * x;
+    return 1
+         - x2 / 2
+         + (x2 * x2) / 24
+         - (x2 * x2 * x2) / 720;
+}
+
+// --------------------------------------
+
+// -------- Modified by Akilan ----------
+
+static unsigned int seed = 123456789;  // must be initialized somehow
+
+unsigned int rand_test(void)
+{
+    seed = (1103515245 * seed + 12345);
+    return (seed >> 16) & 0x7FFF; // 15-bit result
+}
+
+// ---------------------------------------
+
+/*
+ * Function for producing a random number between two double values
+ */
+double frand(double xmin, double xmax) {
+    return xmin + (xmax-xmin)*rand_test()/RAND_MAX;
+}
+
+/*
+ * Prints the time between two clocks in seconds
+ */
+// void print_time(clock_t s, clock_t e)
+// {
+//     printf("Time: %f seconds\n", (double)(e-s)/CLOCKS_PER_SEC);
+// }
+
+/*
+ * This function is called every time GLUT refreshes the display.
+ */
+// void display(void) {
+// 	//Do a time step
+// 	time_step();
+
+// 	//Draw points
+// 	drawPoints(x, y, N);
+// }
+
+/*
+ * Updates the positions of the particles of a time step.
+ */
+void time_step(void) {
+	//Allocate memory for root
+	root = malloc(sizeof(struct node_t));
+	set_node(root);
+	root->min_x = 0; root->max_x = 1; root->min_y = 0; root->max_y = 1;
+
+	//Put particles in tree
+	for(int i = 0; i < N; i++) {
+		put_particle_in_tree(i, root);
+	}
+
+	//Calculate mass and center of mass
+	calculate_mass(root);
+	calculate_center_of_mass_x(root);
+	calculate_center_of_mass_y(root);
+
+	//Calculate forces
+	update_forces();
+
+	//Update velocities and positions
+	for(int i = 0; i < N; i++) {
+		double ax = force_x[i]/mass[i];
+		double ay = force_y[i]/mass[i];
+		u[i] += ax*dt;
+		v[i] += ay*dt;
+		x[i] += u[i]*dt;
+		y[i] += v[i]*dt;
+
+		/* This of course doesn't make any sense physically,
+		 * but makes sure that the particles stay within the
+		 * bounds. Normally the particles won't leave the
+		 * area anyway.
+		*/
+		bounce(&x[i], &y[i], &u[i], &v[i]);
+	}
+
+	//Free memory
+	free_node(root);
+	free(root);
+}
+
+/*
+ * If a particle moves beyond any of the boundaries then bounce it back
+ */
+void bounce(double *x, double *y, double *u, double *v) {
+	double W = 1.0f, H = 1.0f;
+	if(*x > W) {
+		*x = 2*W - *x;
+		*u = -*u;
+	}
+
+	if(*x < 0) {
+		*x = -*x;
+		*u = -*u;
+	}
+
+	if(*y > H) {
+		*y = 2*H - *y;
+		*v = -*v;
+	}
+
+	if(*y < 0) {
+		*y = -*y;
+		*v = -*v;
+	}
+}
+
+/*
+ * Puts a particle recursively in the Barnes Hut quad-tree.
+ */
+void put_particle_in_tree(int new_particle, struct node_t *node) {
+	//If no particle is assigned to the node
+	if(!node->has_particle) {
+		node->particle = new_particle;
+		node->has_particle = 1;
+	}
+	//If the node has no children
+	else if(!node->has_children) {
+		//Allocate and initiate children
+		node->children = malloc(4*sizeof(struct node_t));
+		for(int i = 0; i < 4; i++) {
+			set_node(&node->children[i]);
+		}
+
+		//Set boundaries for the children
+		node->children[0].min_x = node->min_x; node->children[0].max_x = (node->min_x + node->max_x)/2;
+		node->children[0].min_y = node->min_y; node->children[0].max_y = (node->min_y + node->max_y)/2;
+
+		node->children[1].min_x = (node->min_x + node->max_x)/2; node->children[1].max_x = node->max_x;
+		node->children[1].min_y = node->min_y; node->children[1].max_y = (node->min_y + node->max_y)/2;
+
+		node->children[2].min_x = node->min_x; node->children[2].max_x = (node->min_x + node->max_x)/2;
+		node->children[2].min_y = (node->min_y + node->max_y)/2; node->children[2].max_y = node->max_y;
+
+		node->children[3].min_x = (node->min_x + node->max_x)/2; node->children[3].max_x = node->max_x;
+		node->children[3].min_y = (node->min_y + node->max_y)/2; node->children[3].max_y = node->max_y;
+
+		//Put old particle into the appropriate child
+		place_particle(node->particle, node);
+
+		//Put new particle into the appropriate child
+		place_particle(new_particle, node);
+
+		//It now has children
+		node->has_children = 1;
+	}
+	//Add the new particle to the appropriate children
+	else {
+		//Put new particle into the appropriate child
+		place_particle(new_particle, node);
+	}
+}
+
+
+/*
+ * Puts a particle in the right child of a node with children.
+ */
+void place_particle(int particle, struct node_t *node) {
+	if(x[particle] <= (node->min_x + node->max_x)/2 && y[particle] <= (node->min_y + node->max_y)/2) {
+		put_particle_in_tree(particle, &node->children[0]);
+	} else if(x[particle] > (node->min_x + node->max_x)/2 && y[particle] < (node->min_y + node->max_y)/2) {
+		put_particle_in_tree(particle, &node->children[1]);
+	} else if(x[particle] < (node->min_x + node->max_x)/2 && y[particle] > (node->min_y + node->max_y)/2) {
+		put_particle_in_tree(particle, &node->children[2]);
+	} else {
+		put_particle_in_tree(particle, &node->children[3]);
+	}
+}
+
+/*
+ * Sets initial values for a new node
+ */
+void set_node(struct node_t *node) {
+	node->has_particle = 0;
+	node->has_children = 0;
+}
+
+/*
+ * Frees memory for a node and its children recursively.
+ */
+void free_node(struct node_t *node){
+	if(node->has_children){
+		free_node(&node->children[0]);
+		free_node(&node->children[1]);
+		free_node(&node->children[2]);
+		free_node(&node->children[3]);
+		free(node->children);
+	}
+}
+
+/*
+ * Displays the boundaries of the tree using OpenGL and GLUT.
+ * Used for debugging purposes.
+ */
+// void display_tree(struct node_t *node) {
+// 	drawRect(node->min_x, node->max_x, node->min_y, node->max_y);
+// 	if(node->has_children == 1) {
+// 		for(int i = 0; i < 4; i++) {
+// 			display_tree(&node->children[i]);
+// 		}
+// 	}
+// }
+
+/*
+ * Calculates the total mass for the node. It recursively updates the mass
+ * of itself and all of its children.
+ */
+double calculate_mass(struct node_t *node) {
+	if(!node->has_particle) {
+		node->total_mass = 0;
+	} else if(!node->has_children) {
+		node->total_mass = mass[node->particle];
+	} else {
+		node->total_mass = 0;
+		for(int i = 0; i < 4; i++) {
+			node->total_mass += calculate_mass(&node->children[i]);
+		}
+	}
+	return node->total_mass;
+}
+
+/*
+ * Calculates the x-position of the centre of mass for the
+ * node. It recursively updates the position of itself and
+ * all of its children.
+ */
+double calculate_center_of_mass_x(struct node_t *node) {
+	if(!node->has_children) {
+		node->c_x = x[node->particle];
+	} else {
+		node->c_x = 0;
+		double m_tot = 0;
+		for(int i = 0; i < 4; i++) {
+			if(node->children[i].has_particle) {
+				node->c_x += node->children[i].total_mass*calculate_center_of_mass_x(&node->children[i]);
+				m_tot += node->children[i].total_mass;
+			}
+		}
+		node->c_x /= m_tot;
+	}
+	return node->c_x;
+}
+
+/*
+ * Calculates the y-position of the centre of mass for the
+ * node. It recursively updates the position of itself and
+ * all of its children.
+ */
+double calculate_center_of_mass_y(struct node_t *node) {
+	if(!node->has_children) {
+		node->c_y = y[node->particle];
+	} else {
+		node->c_y = 0;
+		double m_tot = 0;
+		for(int i = 0; i < 4; i++) {
+			if(node->children[i].has_particle) {
+				node->c_y += node->children[i].total_mass*calculate_center_of_mass_y(&node->children[i]);
+				m_tot += node->children[i].total_mass;
+			}
+		}
+		node->c_y /= m_tot;
+	}
+	return node->c_y;
+}
+
+/*
+ * Calculates the forces in a time step of all particles in
+ * the simulation using the Barnes Hut quad tree.
+ */
+void update_forces(){
+	for(int i = 0; i < N; i++) {
+		force_x[i] = 0;
+		force_y[i] = 0;
+		update_forces_help(i, root);
+	}
+}
+
+/*
+ * Help function for calculating the forces recursively
+ * using the Barnes Hut quad tree.
+ */
+void update_forces_help(int particle, struct node_t *node) {
+    //The node is a leaf node with a particle and not the particle itself
+    if(!node->has_children && node->has_particle && node->particle != particle) {
+        double r = my_sqrt((x[particle] - node->c_x)*(x[particle] - node->c_x) + (y[particle] - node->c_y)*(y[particle] - node->c_y));
+        calculate_force(particle, node, r);
+    }
+    //The node has children
+    else if(node->has_children) {
+        //Calculate r and theta
+        double r = my_sqrt((x[particle] - node->c_x)*(x[particle] - node->c_x) + (y[particle] - node->c_y)*(y[particle] - node->c_y));
+        double theta = (node->max_x - node->min_x)/r;
+
+        /* If the distance to the node's centre of mass is far enough, calculate the force,
+         * otherwise traverse further down the tree
+         */
+        if(theta < 0.5){
+            calculate_force(particle, node, r);
+        } else {
+            update_forces_help(particle, &node->children[0]);
+            update_forces_help(particle, &node->children[1]);
+            update_forces_help(particle, &node->children[2]);
+            update_forces_help(particle, &node->children[3]);
+        }
+    }
+}
+
+/*
+ * Calculates and updates the force of a particle from a node.
+ */
+void calculate_force(int particle, struct node_t *node, double r){
+    double temp = -grav*mass[particle]*node->total_mass/((r + epsilon)*(r + epsilon)*(r + epsilon));
+	force_x[particle] += (x[particle] - node->c_x)*temp;
+	force_y[particle] += (y[particle] - node->c_y)*temp;
+}
+
+/*
+ * Main function.
+ */
+int main(void) {
+
+	//The first arguments sets if the graphics should be used
+	// int graphics = 1;
+	// if (argc > 1) {
+	// 	graphics = atoi(argv[1]);
+	// }
+
+	//The second argument sets the number of time steps
+	int time_steps = 3;
+	// if (argc > 2) {
+	// 	time_steps = atoi(argv[2]);
+	// }
+
+	//Initiate memory for the vectors
+	x = (double *)malloc(N*sizeof(double));
+	y = (double *)malloc(N*sizeof(double));
+	u = (double *)malloc(N*sizeof(double));
+	v = (double *)malloc(N*sizeof(double));
+	// -------- Modified by Akilan ----------
+	force_x = (double *)malloc(N*sizeof(double));
+	force_y = (double *)malloc(N*sizeof(double));
+	// --------------------------------------
+    mass = (double *)malloc(N*sizeof(double));
+
+	//Set the initial values
+	// for(int i = 0; i < N; i++) {
+	// 	mass[i] = 1;
+	// 	double R = frand(0, L/4);
+	// 	double theta = frand(0, 2*M_PI);
+	// 	x[i] = L/2 + R*my_cos(theta);
+	// 	y[i] = W/2 + alpha*R*my_sin(theta);
+	// 	double R_prim = my_sqrt(pow(x[i] - L/2, 2) + pow(y[i] - W/2, 2));
+	// 	u[i] = -V*R_prim*my_sin(theta);
+	// 	v[i] = V*R_prim*my_cos(theta);
+	// }
+	for (int i = 0; i < N; i++) {
+    	mass[i] = 1;
+
+    	double R = frand(0, L/4);
+    	double theta = frand(0, 2*PI);
+
+    	double c = my_cos(theta);
+    	double s = my_sin(theta);
+
+    	x[i] = L/2 + R * c;
+    	y[i] = W/2 + alpha * R * s;
+
+    	double dx = x[i] - L/2;
+    	double dy = y[i] - W/2;
+
+    	double R_prim = my_sqrt(dx*dx + dy*dy);
+
+    	u[i] = -V * R_prim * s;
+    	v[i] =  V * R_prim * c;
+	}
+
+	/* Run the GLUT display function if the graphics mode is on.
+     * Otherwise just run the simulations without graphics
+     */
+	// if (graphics) {
+	// 	// Initialize the graphics routines
+	// 	graphicsInit(&argc, argv, display);
+
+	// 	//Used for the display window
+	// 	glutMainLoop();
+	// } else {
+		//Begin taking time
+		// long start = clock();
+
+		//The main loop
+		for(int i = 0; i < time_steps; i++) {
+			time_step();
+		}
+
+		//Stop taking time and print elapsed time
+		// long stop = clock();
+		// print_time(start, stop);
+	// }
+
+	//Free memory
+	free(x);
+	free(y);
+	free(u);
+	free(v);
+	free(force_x);
+	free(force_y);
+    free(mass);
+
+	return 0;
+}

Tests/isa/Cprograms/benchmark/barnes/barnes_hut.h

@@ -0,0 +1,55 @@
+/*
+ * File: barnes_hut.h
+ * --------------
+ * Header file for barnes_hut.c
+ * Written by:
+ * Joel Backsell, Kim Torberntsson & Alexander Bilock.
+ */
+
+#ifndef Barnes_Hut_barnes_hut_h
+#define Barnes_Hut_barnes_hut_h
+
+/*
+ * Struct that represents a node of the Barnes Hut quad tree.
+ */
+struct node_t {
+    int particle;
+    int has_particle;
+    int has_children;
+    double min_x, max_x, min_y, max_y, total_mass, c_x, c_y;
+    struct node_t *children;
+};
+
+//Function for producing a random number between two double values.
+double frand(double xmin, double xmax);
+
+//Prints the time between two clocks in seconds
+// void print_time(clock_t s, clock_t e);
+
+//OpenGL method for displaying the particles
+// void display(void);
+
+//Calculates the positions of a time step
+// void time_step(void);
+
+//If a particle moves beyond any of the boundaries then bounce it back
+void bounce(double *x, double *y, double *u, double *v);
+
+//Functions for handling the placement of particles in the tree
+void put_particle_in_tree(int new_particle, struct node_t *node);
+void place_particle(int particle, struct node_t *node);
+void set_node(struct node_t *node);
+void free_node(struct node_t *node);
+void display_tree(struct node_t *node);
+
+//Functions for calculating the mass and centre of mass of the tree
+double calculate_mass(struct node_t *node);
+double calculate_center_of_mass_x(struct node_t *node);
+double calculate_center_of_mass_y(struct node_t *node);
+
+//Functions for the force calculations
+void update_forces();
+void update_forces_help(int particle, struct node_t *node);
+void calculate_force(int particle, struct node_t *node, double r);
+
+#endif

Tests/isa/Cprograms/benchmark/barnes/build.sh

@@ -0,0 +1,46 @@
+# scp ../../encoding.h home-1:/home/akilan/
+# scp ../../entry.S home-1:/home/akilan/
+# scp ../../link.ld home-1:/home/akilan/
+# scp ../../riscv_test_p.h home-1:/home/akilan/
+# scp ../../riscv_test.h home-1:/home/akilan/
+# scp ../../string.c home-1:/home/akilan/
+# scp ../../vm.c home-1:/home/akilan/
+# scp ../../test.S home-1:/home/akilan/
+# scp ../../test_macros.h home-1:/home/akilan/
+# scp ../../malloc.c home-1:/home/akilan/
+# scp ../../test.c home-1:/home/akilan/
+scp barnes_hut.c home-1:/home/akilan/
+scp barnes_hut.h home-1:/home/akilan/
+
+# GCC compile
+# ssh home-1 'cd /home/akilan/ && export PATH=/opt/riscv/bin:$PATH && riscv64-unknown-elf-gcc -DENTROPY=0xf21e02b -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -mabi=lp64 -T link.ld vm.c string.c entry.S test.S -o testC'
+
+# Clang compile (architecture Regular RISCV)
+ssh home-1 'clang --target=riscv64-unknown-elf -DENTROPY=0xf21e02b -DDEFINE_MALLOC -DDEFINE_FREE --gcc-toolchain=/opt/riscv -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -fno-builtin-malloc -mabi=lp64 -T link.ld vm.c string.c malloc.c entry.S test.S barnes_hut.c -o testC -lm -lgcc'
+
+# Clang using the cheri clang compiler
+# ssh home-1 'cd cheri/output/sdk/bin/ && ./clang --target=riscv64-unknown-freebsd -DENTROPY=0xf21e02b --sysroot=cheribsd-riscv64-for-purecap-rootfs.cfg -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -march=rv64imafdcxcheri -mabi=lp64d -T link.ld vm.c string.c entry.S test.S -o testC'
+
+scp home-1:/home/akilan/testC ../../../
+
+# scp home-1:/home/akilan/cheri/output/sdk/bin/testC .
+
+# riscv64-unknown-elf-gcc \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostartfiles \
+#   -fno-builtin \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+
+# clang \
+#   --target=riscv64-unknown-elf \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostdlib \
+#   -nostartfiles \
+#   -fuse-ld=lld \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+#   -I. \
+#   -o output.elf%

Tests/isa/Cprograms/benchmark/glibc/build.sh

@@ -0,0 +1,45 @@
+scp ../../encoding.h home-1:/home/akilan/
+scp ../../entry.S home-1:/home/akilan/
+scp ../../link.ld home-1:/home/akilan/
+scp ../../riscv_test_p.h home-1:/home/akilan/
+scp ../../riscv_test.h home-1:/home/akilan/
+scp ../../string.c home-1:/home/akilan/
+scp ../../vm.c home-1:/home/akilan/
+scp ../../test.S home-1:/home/akilan/
+scp ../../test_macros.h home-1:/home/akilan/
+scp ../../malloc.c home-1:/home/akilan/
+# scp ../../test.c home-1:/home/akilan/
+scp glibc.c home-1:/home/akilan/
+
+# GCC compile
+# ssh home-1 'cd /home/akilan/ && export PATH=/opt/riscv/bin:$PATH && riscv64-unknown-elf-gcc -DENTROPY=0xf21e02b -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -mabi=lp64 -T link.ld vm.c string.c entry.S test.S -o testC'
+
+# Clang compile (architecture Regular RISCV)
+ssh home-1 'clang --target=riscv64-unknown-elf -DENTROPY=0xf21e02b -DDEFINE_MALLOC -DDEFINE_FREE --gcc-toolchain=/opt/riscv -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -fno-builtin-malloc -mabi=lp64 -T link.ld vm.c string.c malloc.c entry.S test.S glibc.c -o testC'
+
+# Clang using the cheri clang compiler
+# ssh home-1 'cd cheri/output/sdk/bin/ && ./clang --target=riscv64-unknown-freebsd -DENTROPY=0xf21e02b --sysroot=cheribsd-riscv64-for-purecap-rootfs.cfg -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -march=rv64imafdcxcheri -mabi=lp64d -T link.ld vm.c string.c entry.S test.S -o testC'
+
+scp home-1:/home/akilan/testC ../../../
+
+# scp home-1:/home/akilan/cheri/output/sdk/bin/testC .
+
+# riscv64-unknown-elf-gcc \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostartfiles \
+#   -fno-builtin \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+
+# clang \
+#   --target=riscv64-unknown-elf \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostdlib \
+#   -nostartfiles \
+#   -fuse-ld=lld \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+#   -I. \
+#   -o output.elf%

Tests/isa/Cprograms/benchmark/glibc/glibc.c

@@ -0,0 +1,229 @@
+/* Benchmark malloc and free functions.
+   Copyright (C) 2019-2021 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <https://www.gnu.org/licenses/>.  */
+
+// modified by Daan Leijen to fit the bench suite and add lifo/fifo free order.
+
+// #include <pthread.h>
+// #include <stdio.h>
+// #include <stdlib.h>
+// #include <sys/resource.h>
+// #include "malloc.h"
+
+// #include <cheriintrin.h>
+#include <stdint.h>
+#include <stddef.h>
+
+// #define malloc      MALLOCCHERI
+// #define free        FREECHERI
+
+// #include "bench-timing.h"
+// #include "json-lib.h"
+
+// void free(void * __capability ptr);
+// void * __capability malloc(size_t size);
+
+void* malloc(size_t size);
+void free(void *ptr);
+
+/* Benchmark the malloc/free performance of a varying number of blocks of a
+   given size.  This enables performance tracking of the t-cache and fastbins.
+   It tests 3 different scenarios: single-threaded using main arena,
+   multi-threaded using thread-arena, and main arena with SINGLE_THREAD_P
+   false.  */
+
+// source: https://github.com/daanx/mimalloc-bench/blob/master/bench/glibc-bench/bench-malloc-thread.c
+
+#define NUM_ITERS 20
+#define NUM_ALLOCS 4
+#define MAX_ALLOCS 1600
+
+// Daan: disable timing
+typedef long timing_t;
+#define TIMING_NOW(s) 
+#define TIMING_DIFF(e,start,stop)
+
+
+typedef struct
+{
+  size_t iters;
+  size_t size;
+  int n;
+  timing_t elapsed;
+} malloc_args;
+
+static void
+do_benchmark(malloc_args *args, char **arr)
+{
+  // timing_t start, stop;
+  size_t iters = args->iters;
+  size_t size = args->size;
+  int n = args->n;
+
+//   TIMING_NOW (start);
+
+  for (int j = 0; j < iters; j++)
+    {
+      for (int i = 0; i < n; i++) {
+        arr[i] = malloc (size);
+        for(int g = 0; g < size; g++) { arr[i][g] =(char)g; }
+      }
+
+      // free half in fifo order
+      for (int i = 0; i < n/2; i++) {
+	        free (arr[i]);
+          arr[i] = NULL;  
+      }
+    
+      // and the other half in lifo order
+      for(int i = n-1; i >= n/2; i--) {
+          free(arr[i]);
+          arr[i] = NULL;
+      }
+  }      
+
+//   TIMING_NOW (stop);
+
+//   TIMING_DIFF (args->elapsed, start, stop);
+}
+
+static malloc_args tests[3][NUM_ALLOCS];
+static int allocs[NUM_ALLOCS] = { 25, 100, 400, MAX_ALLOCS };
+
+// static void *
+// thread_test (void *p)
+// {
+//   char **arr = (char**)p;
+
+//   /* Run benchmark multi-threaded.  */
+//   for (int i = 0; i < NUM_ALLOCS; i++)
+//     do_benchmark (&tests[2][i], arr);
+
+//   return p;
+// }
+
+void
+bench (unsigned long size)
+{
+  size_t iters = NUM_ITERS;
+  char**arr = (char**)malloc (MAX_ALLOCS * sizeof (void*));
+  //char * __capability * __capability arr = malloc(MAX_ALLOCS * sizeof(char * __capability));
+
+  for (int t = 0; t < 3; t++)
+    for (int i = 0; i < NUM_ALLOCS; i++)
+      {
+	tests[t][i].n = allocs[i];
+	tests[t][i].size = size;
+	tests[t][i].iters = iters / allocs[i];
+
+	/* Do a quick warmup run.  */
+	if (t == 0)
+	  do_benchmark (&tests[0][i], arr);
+      }
+
+  /* Run benchmark single threaded in main_arena.  */
+  for (int i = 0; i < NUM_ALLOCS; i++)
+    do_benchmark (&tests[0][i], arr);
+
+  /* Run benchmark in a thread_arena.  */
+//   pthread_t t;
+//   pthread_create (&t, NULL, thread_test, (void*)arr);
+//   pthread_join (t, NULL);
+
+  /* Repeat benchmark in main_arena with SINGLE_THREAD_P == false.  */
+  // for (int i = 0; i < NUM_ALLOCS; i++)
+  //   do_benchmark (&tests[1][i], arr);
+
+  free (arr);
+  arr = NULL;
+
+  /*
+  json_ctx_t json_ctx;
+
+  json_init (&json_ctx, 0, stdout);
+
+  json_document_begin (&json_ctx);
+
+  json_attr_string (&json_ctx, "timing_type", TIMING_TYPE);
+
+  json_attr_object_begin (&json_ctx, "functions");
+
+  json_attr_object_begin (&json_ctx, "malloc");
+
+  char s[100];
+  double iters2 = iters;
+
+  json_attr_object_begin (&json_ctx, "");
+  json_attr_double (&json_ctx, "malloc_block_size", size);
+
+  struct rusage usage;
+  getrusage (RUSAGE_SELF, &usage);
+  json_attr_double (&json_ctx, "max_rss", usage.ru_maxrss);
+
+  for (int i = 0; i < NUM_ALLOCS; i++)
+    {
+      sprintf (s, "main_arena_st_allocs_%04d_time", allocs[i]);
+      json_attr_double (&json_ctx, s, tests[0][i].elapsed / iters2);
+    }
+
+  for (int i = 0; i < NUM_ALLOCS; i++)
+    {
+      sprintf (s, "main_arena_mt_allocs_%04d_time", allocs[i]);
+      json_attr_double (&json_ctx, s, tests[1][i].elapsed / iters2);
+    }
+
+  for (int i = 0; i < NUM_ALLOCS; i++)
+    {
+      sprintf (s, "thread_arena__allocs_%04d_time", allocs[i]);
+      json_attr_double (&json_ctx, s, tests[2][i].elapsed / iters2);
+    }
+
+  json_attr_object_end (&json_ctx);
+
+  json_attr_object_end (&json_ctx);
+
+  json_attr_object_end (&json_ctx);
+
+  json_document_end (&json_ctx);
+  */
+}
+
+// static void usage (const char *name)
+// {
+//   fprintf (stderr, "%s: <alloc_size>\n", name);
+//   exit (1);
+// }
+
+int main (void)
+{
+//   INITREGULARALLOC();
+  long size = 3;
+//   if (argc == 2)
+//     size = strtol (argv[1], NULL, 0);
+
+//   if (argc > 2 || size <= 0)
+//     usage (argv[0]);
+
+  bench (size);
+  // bench (2*size);
+  // bench (4*size);
+  // bench (8*size);
+  // bench (16*size);
+  // bench (32*size); 
+
+  return 0;
+}

Tests/isa/Cprograms/benchmark/matrixmul/build.sh

@@ -0,0 +1,45 @@
+scp ../../encoding.h home-1:/home/akilan/
+scp ../../entry.S home-1:/home/akilan/
+scp ../../link.ld home-1:/home/akilan/
+scp ../../riscv_test_p.h home-1:/home/akilan/
+scp ../../riscv_test.h home-1:/home/akilan/
+scp ../../string.c home-1:/home/akilan/
+scp ../../vm.c home-1:/home/akilan/
+scp ../../test.S home-1:/home/akilan/
+scp ../../test_macros.h home-1:/home/akilan/
+scp ../../malloc.c home-1:/home/akilan/
+# scp ../../test.c home-1:/home/akilan/
+scp matrixmul.c home-1:/home/akilan/
+
+# GCC compile
+# ssh home-1 'cd /home/akilan/ && export PATH=/opt/riscv/bin:$PATH && riscv64-unknown-elf-gcc -DENTROPY=0xf21e02b -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -mabi=lp64 -T link.ld vm.c string.c entry.S test.S -o testC'
+
+# Clang compile (architecture Regular RISCV)
+ssh home-1 'clang --target=riscv64-unknown-elf -DENTROPY=0xf21e02b -DDEFINE_MALLOC -DDEFINE_FREE --gcc-toolchain=/opt/riscv -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -fno-builtin-malloc -mabi=lp64 -T link.ld vm.c string.c malloc.c entry.S test.S matrixmul.c -o testC'
+
+# Clang using the cheri clang compiler
+# ssh home-1 'cd cheri/output/sdk/bin/ && ./clang --target=riscv64-unknown-freebsd -DENTROPY=0xf21e02b --sysroot=cheribsd-riscv64-for-purecap-rootfs.cfg -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -march=rv64imafdcxcheri -mabi=lp64d -T link.ld vm.c string.c entry.S test.S -o testC'
+
+scp home-1:/home/akilan/testC ../../../
+
+# scp home-1:/home/akilan/cheri/output/sdk/bin/testC .
+
+# riscv64-unknown-elf-gcc \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostartfiles \
+#   -fno-builtin \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+
+# clang \
+#   --target=riscv64-unknown-elf \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostdlib \
+#   -nostartfiles \
+#   -fuse-ld=lld \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+#   -I. \
+#   -o output.elf%

Tests/isa/Cprograms/benchmark/matrixmul/matrixmul.c

@@ -0,0 +1,205 @@
+// Source: https://github.com/ivanbgd/Matrix-Multiplication-MatMul-C/blob/master/matmul_1d_seq.c
+#define MATMUL_1D
+#ifdef MATMUL_1D
+
+/* Matrices are represented as 1-D arrays in memory.
+ * That means they are contiguous in memory.
+ * Minimum dimension is 1, not 0, and internal dimensions must match. */
+
+#include <math.h>
+// #include <omp.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <time.h>
+
+void free(void *ptr);
+void *malloc(size_t size);
+
+/* Initializes vector or matrix, sequentially, with indices. */
+void init_seq(double *a, const unsigned n_rows_a, const unsigned n_cols_a) {
+    for (size_t i = 0; i < n_rows_a; i++) {
+        for (size_t j = 0; j < n_cols_a; j++) {
+            a[i*n_cols_a + j] = 2.0;
+        }
+    }
+}
+
+// static unsigned int seed = 123456789;  // must be initialized somehow
+
+// unsigned int rand_test(void)
+// {
+//     seed = (1103515245 * seed + 12345);
+//     return (seed >> 16) & 0x7FFF; // 15-bit result
+// }
+
+// /* Initializes vector or matrix, randomly. */
+// void init_rand(double *a, const unsigned n_rows_a, const unsigned n_cols_a) {
+//     for (size_t i = 0; i < n_rows_a; i++) {
+//         for (size_t j = 0; j < n_cols_a; j++) {
+//             a[i*n_cols_a + j] = rand_test() / 32767.0;
+//             // rand_test() / (double)RAND_MAX;
+//         }
+//     }
+// }
+
+/*  Takes and returns a new matrix, t, which is a transpose of the original one, m.
+    It's also flat in memory, i.e., 1-D, but it should be looked at as a transpose
+    of m, meaning, n_rows_t == n_cols_m, and n_cols_t == n_rows_m.
+    The original matrix m stays intact. */
+double *transpose(const double *m, const unsigned n_rows_m, const unsigned n_cols_m, double *t) {
+    for (size_t i = 0; i < n_rows_m; i++) {
+        for (size_t j = 0; j < n_cols_m; j++) {
+            t[j*n_rows_m + i] = m[i*n_cols_m + j];
+        }
+    }
+
+    return t;
+}
+
+/* Dot product of two arrays, or matrix product
+ * Allocates and returns an array.
+ * This variant doesn't transpose matrix b, and it's a lot slower. */
+double *dot_simple(const double *a, const unsigned n_rows_a, const unsigned n_cols_a,\
+                   const double *b, const unsigned n_rows_b, const unsigned n_cols_b) {
+
+    if (n_cols_a != n_rows_b) {
+        // printf("#columns A must be equal to #rows B!\n");
+        // system("pause");
+        // exit(-2);
+    }
+
+    double *c = malloc(n_rows_a * n_cols_b * sizeof(*c));
+    if (c == NULL) {
+        // printf("Couldn't allocate memory!\n");
+        // system("pause");
+        // exit(-1);
+    }
+
+    for (size_t i = 0; i < n_rows_a; i++) {
+        for (size_t k = 0; k < n_cols_b; k++) {
+            double sum = 0.0;
+            for (size_t j = 0; j < n_cols_a; j++) {
+                sum += a[i*n_cols_a + j] * b[j*n_cols_b + k];
+            }
+            c[i*n_cols_b + k] = sum;
+        }
+    }
+
+    return c;
+}
+
+/* Dot product of two arrays, or matrix product
+ * Allocates and returns an array.
+ * This variant transposes matrix b, and it's a lot faster. */
+double *dot(const double *a, const unsigned n_rows_a, const unsigned n_cols_a, \
+            const double *b, const unsigned n_rows_b, const unsigned n_cols_b) {
+
+    if (n_cols_a != n_rows_b) {
+        // printf("#columns A must be equal to #rows B!\n");
+        // system("pause");
+        // exit(-2);
+    }
+
+    double *bt = malloc(n_rows_b * n_cols_b * sizeof(*b));
+
+    double *c = malloc(n_rows_a * n_cols_b * sizeof(*c));
+
+    if ((c == NULL) || (bt == NULL)) {
+        // printf("Couldn't allocate memory!\n");
+        // system("pause");
+        // exit(-1);
+    }
+
+    bt = transpose(b, n_rows_b, n_cols_b, bt);
+
+    for (size_t i = 0; i < n_rows_a; i++) {
+        for (size_t k = 0; k < n_cols_b; k++) {
+            double sum = 0.0;
+            for (size_t j = 0; j < n_cols_a; j++) {
+                sum += a[i*n_cols_a + j] * bt[k*n_rows_b + j];
+            }
+            c[i*n_cols_b + k] = sum;
+        }
+    }
+
+    free(bt);
+
+    return c;
+}
+
+/* Prints vector, or matrix. */
+// void print(const double *a, const unsigned n_rows_a, const unsigned n_cols_a) {
+//     for (size_t i = 0; i < n_rows_a; i++) {
+//         for (size_t j = 0; j < n_cols_a; j++) {
+//             printf("%8.3f ", a[i*n_cols_a + j]);
+//         }
+//         printf("\n");
+//     }
+//     printf("\n");
+// }
+
+int main(void) {
+    /* Intializes random number generator */
+    // time_t t;
+    // srand((unsigned)time(&t));
+    // srand(0);
+
+    /* For measuring time */
+    double t0, t1;
+
+    const unsigned scale = 2;
+    const unsigned n_rows_a = 4 * scale;
+    const unsigned n_cols_a = 3 * scale;
+    const unsigned n_rows_b = 3 * scale;
+    const unsigned n_cols_b = 2 * scale;
+
+    double *a = malloc(n_rows_a * n_cols_a * sizeof(*a));
+    double *b = malloc(n_rows_b * n_cols_b * sizeof(*b));
+    double *c = NULL;
+    double *d = NULL;
+
+    if (!a || !b) {
+        // printf("Couldn't allocate memory!\n");
+        // system("pause");
+        // exit(-1);
+    }
+
+    // init_rand(a, n_rows_a, n_cols_a);
+    // init_rand(b, n_rows_b, n_cols_b);
+
+    init_seq(a, n_rows_a, n_cols_a);
+    init_seq(b, n_rows_b, n_cols_b);
+
+    // t0 = omp_get_wtime();
+    c = dot_simple(a, n_rows_a, n_cols_a, b, n_rows_b, n_cols_b);
+    // t1 = omp_get_wtime();
+    // printf("Dot Simple: Elapsed time %.3f s\n", t1 - t0);
+
+    // t0 = omp_get_wtime();
+    d = dot(a, n_rows_a, n_cols_a, b, n_rows_b, n_cols_b);
+    // t1 = omp_get_wtime();
+    // printf("Dot: Elapsed time %.3f s\n", t1 - t0);
+
+    if (scale == 1) {
+        // printf("Matrix A:\n");
+        // print(a, n_rows_a, n_cols_a);
+        // // printf("Matrix B:\n");
+        // print(b, n_rows_b, n_cols_b);
+        // // printf("Matrix C:\n");
+        // print(c, n_rows_a, n_cols_b);
+        // // printf("Matrix D:\n");
+        // print(d, n_rows_a, n_cols_b);
+    }
+
+    free(a);
+    free(b);
+    free(c);
+    free(d);
+
+    // system("pause");
+
+    return(0);
+}
+
+#endif // MATMUL_1D

Tests/isa/Cprograms/benchmark/memaccess/build.sh

@@ -0,0 +1,45 @@
+scp ../../encoding.h home-1:/home/akilan/
+scp ../../entry.S home-1:/home/akilan/
+scp ../../link.ld home-1:/home/akilan/
+scp ../../riscv_test_p.h home-1:/home/akilan/
+scp ../../riscv_test.h home-1:/home/akilan/
+scp ../../string.c home-1:/home/akilan/
+scp ../../vm.c home-1:/home/akilan/
+scp ../../test.S home-1:/home/akilan/
+scp ../../test_macros.h home-1:/home/akilan/
+scp ../../malloc.c home-1:/home/akilan/
+# scp ../../test.c home-1:/home/akilan/
+scp memaccess.c home-1:/home/akilan/
+
+# GCC compile
+# ssh home-1 'cd /home/akilan/ && export PATH=/opt/riscv/bin:$PATH && riscv64-unknown-elf-gcc -DENTROPY=0xf21e02b -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -mabi=lp64 -T link.ld vm.c string.c entry.S test.S -o testC'
+
+# Clang compile (architecture Regular RISCV)
+ssh home-1 'clang --target=riscv64-unknown-elf -DENTROPY=0xf21e02b -DDEFINE_MALLOC -DDEFINE_FREE --gcc-toolchain=/opt/riscv -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -fno-builtin-malloc -mabi=lp64 -T link.ld vm.c string.c malloc.c entry.S test.S memaccess.c -o testC'
+
+# Clang using the cheri clang compiler
+# ssh home-1 'cd cheri/output/sdk/bin/ && ./clang --target=riscv64-unknown-freebsd -DENTROPY=0xf21e02b --sysroot=cheribsd-riscv64-for-purecap-rootfs.cfg -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -march=rv64imafdcxcheri -mabi=lp64d -T link.ld vm.c string.c entry.S test.S -o testC'
+
+scp home-1:/home/akilan/testC ../../../
+
+# scp home-1:/home/akilan/cheri/output/sdk/bin/testC .
+
+# riscv64-unknown-elf-gcc \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostartfiles \
+#   -fno-builtin \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+
+# clang \
+#   --target=riscv64-unknown-elf \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostdlib \
+#   -nostartfiles \
+#   -fuse-ld=lld \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+#   -I. \
+#   -o output.elf%

Tests/isa/Cprograms/benchmark/memaccess/memaccess.c

@@ -0,0 +1,364 @@
+/*
+memaccesstest.c
+Author Alex Bordei at bigstep
+All rights reserved. 
+*/
+
+// source: https://github.com/bigstepinc/memaccesstest/tree/master
+
+// #include <stdio.h>
+// #include <cheriintrin.h>
+#include <stddef.h>
+#include <stdint.h>
+// #include <stdlib.h>
+// #include <pthread.h>
+// #include "config.h"
+
+// #ifndef __MACH__
+// #include <time.h>
+// #else
+// #include <mach/mach_time.h>
+// #endif
+
+// #include "malloc.h"
+
+// #define malloc      MALLOCCHERI
+// #define free        FREECHERI
+
+#define NPAD 7
+#define MIN_WSS sizeof(struct l)
+#define MAX_WSS 32
+
+void free(void *ptr);
+void *malloc(size_t size);
+
+
+
+// struct l{
+// 	struct l *n;
+// 	struct l *p;
+// 	long int pad[NPAD-1];
+
+// };
+
+struct l{
+    struct l *n;
+    struct l *p;
+    long int pad[NPAD-1];
+};
+
+// uint64_t rand64()
+// {
+// 	// Combin 4 parts of low 16-bit of each rand()
+// 	uint64_t R0 = (uint64_t)random() << 48;
+// 	uint64_t R1 = (uint64_t)random() << 48 >> 16;
+// 	uint64_t R2 = (uint64_t)random() << 48 >> 32;
+// 	uint64_t R3 = (uint64_t)random() << 48 >> 48;
+// 	return R0 | R1 | R2 | R3;
+// };
+
+#ifdef DEBUG
+// void print_ll_lin(struct l *root,unsigned long working_set_size)
+// {
+// 	struct l *current=root->n;
+// 	unsigned long count;
+// 	printf("0");
+// 	count=0;
+// 	while(current!=root)
+// 	{
+// 		// printf("->%ld",current-root);
+// 		current=current->n;
+// 		count++;
+// 		if(count>working_set_size/sizeof(struct l))
+// 		{
+// 			// printf("Error:cycle!");
+// 			break;
+// 		}
+			
+// 	}	
+// 	// printf("\n");
+// }
+// void print_ll_lin_back(struct l *root,unsigned long working_set_size)
+// {
+// 	struct l *current=root->p;
+// 	unsigned long count;
+// 	printf("0");
+// 	count=0;
+// 	while(current!=root)
+// 	{
+// 		// printf("->%ld",current-root);
+// 		current=current->p;
+// 		count++;
+// 		if(count>working_set_size/sizeof(struct l))
+// 		{
+// 			// printf("Error:cycle!");
+// 			break;
+// 		}
+			
+// 	}	
+// 	printf("\n");
+// }
+// void print_ll(struct l *root,unsigned long working_set_size)
+// {
+// 	unsigned long i;
+// 	for(i=0;i<(working_set_size/sizeof(struct l));i++)
+// 		printf("DEBUG: %ld<-%ld->%ld\n",(root+i)->p-root,i,((root+i)->n)-root);		
+// 	printf("DEBUG ");
+// 	print_ll_lin(root, working_set_size);
+// 	printf("DEBUG ");
+// 	print_ll_lin_back(root, working_set_size);
+
+// }
+#endif
+
+// void build_sequencial_ll(struct l * root, long working_set_size)
+void build_sequencial_ll(struct l *root, long working_set_size)
+{
+
+		unsigned long i;
+
+		struct l *current;
+
+
+		for(i=0;i<(working_set_size/sizeof(struct l));i++)
+		{
+			current=root+i;	
+			if(i==(working_set_size/sizeof(struct l)-1))
+				current->n=root;
+			else
+				current->n=current+1;
+			if(i==0)
+				current->p=root+working_set_size/sizeof(struct l)-1;
+			else
+				current->p=current-1;
+		} 
+}
+
+// void build_random_ll(struct l *root, long working_set_size)
+// {
+
+// 		unsigned long i;
+
+
+// 		struct l *a;
+// 		struct l *b;
+// 		struct l *oldan;
+// 		struct l *oldbn;
+// 		struct l *oldap;
+// 		struct l *oldbp;
+
+// 		unsigned long rnd;
+// 		unsigned long tries;
+
+// 		build_sequencial_ll(root, working_set_size);
+
+// 		//construct the array of elements 
+// 		//link the list to the next element in the array
+
+// 		for(i=1;i<(working_set_size/sizeof(struct l)-1);i++)
+// 		{
+
+// 			a=(root+i);
+// 			b=a;
+// 			tries=5;	
+// 			while((b==a || b==root) && tries)
+// 			{
+// 				rnd=rand64() % (working_set_size/sizeof(struct l)-1);	
+// 				b=root+rnd;
+// 				tries--;
+// 			}
+// 			if(!tries)
+// 				continue;
+
+			
+// 			oldan=a->n; //a->n is lost along the way
+// 			oldbn=b->n;  //b->n is lost along the way
+// 			oldap=a->p; //a->p is lost along the way
+// 			oldbp=b->p;  //b->p is lost along the way
+
+// 			a->p->n=b;
+// 			b->p->n=a;
+
+// 			if(b!=oldan)
+// 			{
+// 				b->n=oldan;    		
+// 				oldan->p=b;
+// 			}
+// 			else
+// 			{
+// 				b->n=a;
+// 				a->p=b;
+// 			}
+
+// 			if(a!=oldbn)
+// 			{
+// 				a->n=oldbn;		
+// 				oldbn->p=a;
+// 			}
+// 			else
+// 			{
+// 				a->n=b;
+// 				b->p=a;
+// 			}
+
+// 			if(b!=oldap)
+// 				b->p=oldap; 
+// 			else
+// 				b->p=a;
+
+// 			if(a!=oldbp)
+// 				a->p=oldbp;
+// 			else
+// 				a->p=b;
+
+
+// #ifdef DEBUG
+// 			printf("DEBUG: Round i=%ld\n",i);		
+// 			printf("DEBUG: A=%ld B=%ld\n",a-root,b-root);
+// 			print_ll(root, working_set_size);
+// #endif
+// 		} 
+
+// }
+
+
+
+struct thread_start_data
+{
+	struct l *root;
+	unsigned long working_set_size;
+	int thread_index;		
+};
+
+void * walk(void *param)
+{
+
+		struct thread_start_data *data=param;
+		unsigned long start,end;
+
+
+		struct l *root=data->root;	
+		// struct l* current=root->n;		
+        struct l *current = root->n;  	
+
+		// #ifdef __MACH__
+		// start=mach_absolute_time();
+		// #else
+		// struct timespec ts;
+		// clock_gettime(CLOCK_MONOTONIC_RAW,&ts);
+		// start=ts.tv_nsec; 
+		// #endif
+	
+
+		/////////////////////////////////////
+		//traverse the linked list
+		
+		while(current!=root)	
+		{
+// #ifdef ADDTEST
+// 			current->pad[0]+=current->n->pad[0];
+// #endif
+
+// #ifdef INCTEST
+// 			current->pad[0]++;
+// #endif
+			current=current->n;	
+		}
+
+			
+		/////////////////////////////////////
+// #ifdef __MACH__
+// 		end=mach_absolute_time();
+// #else
+// 		clock_gettime(CLOCK_MONOTONIC_RAW,&ts);
+// 		end=ts.tv_nsec; 
+// #endif
+	
+
+		// printf("%d,%ld,%ld,%ld\n",data->thread_index,data->working_set_size, end-start, data->working_set_size/sizeof(struct l));
+
+		// free(data);
+		return NULL;
+}
+
+int main(void)
+{
+// clock_t tic = clock();
+// INITREGULARALLOC();
+//#ifdef DEBUG
+	// printf("DEBUG:sizeof(struct l)=%ld\n",sizeof(struct l));
+// #endif
+	struct l *root;
+	unsigned long working_set_size;
+	int i;
+	// struct thread_start_data *tdata;
+    struct thread_start_data *tdata;
+
+
+	// pthread_t threads[NTHREADS];
+
+	root=malloc(MAX_WSS/sizeof(struct l) * sizeof(struct l));	
+		
+	for(working_set_size=MIN_WSS;working_set_size<MAX_WSS;working_set_size=working_set_size<<1)
+	{
+
+// #ifdef DEBUG
+// 		printf("DEBUG:Working set size=%ld elements=%ld\n",working_set_size,working_set_size/sizeof(struct l));
+// #endif
+// clock_t toc = clock();
+// printf("Elapsed build get here before ll: %f seconds\n", (double)(toc - tic) / CLOCKS_PER_SEC);
+// #ifdef RANDOM_LIST
+// 		build_random_ll(root, working_set_size);
+// #else
+	// clock_t tic1 = clock();
+	build_sequencial_ll(root, working_set_size);
+	// build_sequencial_ll(root, working_set_size);
+	// clock_t toc1 = clock();
+	// printf("Elapsed build ll: %f seconds\n", (double)(toc1 - tic1) / CLOCKS_PER_SEC);
+// #endif
+
+// #if NTHREADS>0			
+// 		for(i=0;i<NTHREADS;i++)
+// 		{
+// 			tdata= malloc(sizeof(struct thread_start_data));
+// 			tdata->thread_index=i;
+// 			tdata->root=root;
+// 			tdata->working_set_size=working_set_size;
+// 			pthread_create(&threads[i], NULL, walk, (void*) tdata);
+// 		}
+
+// 		for(i=0;i<NTHREADS;i++)
+// 			pthread_join(threads[i],NULL);					
+// #else
+		tdata = malloc(sizeof(struct thread_start_data));
+		tdata->thread_index=0;
+		tdata->root=root;
+		tdata->working_set_size=working_set_size;
+		// clock_t tic2 = clock();
+		for (int i = 0; i < 64; i++) {
+			walk(tdata);	
+		}
+
+        // tdata = malloc(sizeof(struct thread_start_data));
+		// tdata->thread_index=0;
+		// tdata->root=root;
+		// tdata->working_set_size=working_set_size;
+		// // clock_t tic2 = clock();
+		// for (int i = 0; i < 64; i++) {
+		// 	walk(tdata);	
+		// }
+        
+		// clock_t toc2 = clock();
+		// printf("Elapsed build walk: %f seconds\n", (double)(toc2 - tic2) / CLOCKS_PER_SEC);
+// #endif
+	}
+
+	// printf("-----------------------------------------------");
+
+
+	free(root);	
+    root = NULL;
+
+    return 0;
+
+}

Tests/isa/Cprograms/benchmark/richards/build.sh

@@ -0,0 +1,46 @@
+scp ../../encoding.h home-1:/home/akilan/
+scp ../../entry.S home-1:/home/akilan/
+scp ../../link.ld home-1:/home/akilan/
+scp ../../riscv_test_p.h home-1:/home/akilan/
+scp ../../riscv_test.h home-1:/home/akilan/
+scp ../../string.c home-1:/home/akilan/
+scp ../../vm.c home-1:/home/akilan/
+scp ../../test.S home-1:/home/akilan/
+scp ../../test_macros.h home-1:/home/akilan/
+scp ../../malloc.c home-1:/home/akilan/
+
+# scp ../../test.c home-1:/home/akilan/
+scp richards.c home-1:/home/akilan/
+
+# GCC compile
+# ssh home-1 'cd /home/akilan/ && export PATH=/opt/riscv/bin:$PATH && riscv64-unknown-elf-gcc -DENTROPY=0xf21e02b -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -mabi=lp64 -T link.ld vm.c string.c entry.S test.S -o testC'
+
+# Clang compile (architecture Regular RISCV)
+ssh home-1 'clang --target=riscv64-unknown-elf -DENTROPY=0xf21e02b -DDEFINE_MALLOC -DDEFINE_FREE --gcc-toolchain=/opt/riscv -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -fno-builtin-malloc -mabi=lp64 -T link.ld vm.c string.c malloc.c entry.S test.S richards.c -o testC'
+
+# Clang using the cheri clang compiler
+# ssh home-1 'cd cheri/output/sdk/bin/ && ./clang --target=riscv64-unknown-freebsd -DENTROPY=0xf21e02b --sysroot=cheribsd-riscv64-for-purecap-rootfs.cfg -mcmodel=medany -nostdlib -nostartfiles -ffreestanding -fno-builtin -march=rv64imafdcxcheri -mabi=lp64d -T link.ld vm.c string.c entry.S test.S -o testC'
+
+scp home-1:/home/akilan/testC ../../../
+
+# scp home-1:/home/akilan/cheri/output/sdk/bin/testC .
+
+# riscv64-unknown-elf-gcc \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostartfiles \
+#   -fno-builtin \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+
+# clang \
+#   --target=riscv64-unknown-elf \
+#   -march=rv64imac \
+#   -mabi=lp64 \
+#   -nostdlib \
+#   -nostartfiles \
+#   -fuse-ld=lld \
+#   -T link.ld \
+#   entry.S vm.c string.c \
+#   -I. \
+#   -o output.elf%

Tests/isa/Cprograms/benchmark/richards/richards.c

@@ -0,0 +1,455 @@
+// clang-format off
+
+/*  C version of the systems programming language benchmark
+**  Author:  M. J. Jordan  Cambridge Computer Laboratory.
+**
+**  Modified by:  M. Richards, Nov 1996
+**    to be ANSI C and runnable on 64 bit machines + other minor changes
+**  Modified by:  M. Richards, 20 Oct 1998
+**    made minor corrections to improve ANSI compliance (suggested
+**    by David Levine)
+**  Modified by:  Jeremy Singer, 25 May 2022
+**    adapted to use uintptr_t for compatibility with architectures
+**    that do not have word-sized pointers (e.g. CHERI) 
+**    also adapted to use Boehm-Demers-Weiser GC instead of system
+**    malloc, when compiled with -DGC
+**
+**  Compile with, say
+**
+**  gcc -o bench bench.c
+**
+**  or
+**
+**  gcc -o bench100 -Dbench100 bench.c  (for a version that obeys
+**                                       the main loop 100x more often)
+*/
+
+// #include <stdio.h>
+#include <stdint.h>
+#include <stddef.h>
+// #include <stdlib.h>
+
+// #include "harness.h"
+// #include "stddefines.h"
+
+// #ifdef GC
+//   #include "gc.h"
+//   #define MALLOC GC_MALLOC
+// #else
+//   #define MALLOC malloc
+// #endif // GC
+
+// #define malloc      tiny_malloc
+// #define free        tiny_free
+
+void* malloc(size_t);
+void    free(void*);
+
+#ifdef bench100
+#define                Count           10000*100
+#define                Qpktcountval    2326410
+#define                Holdcountval     930563
+#else
+#define                Count           10000
+#define                Qpktcountval    23246
+#define                Holdcountval     9297
+#endif
+
+#define                TRUE            1
+#define                FALSE           0
+#define                MAXINT          32767
+
+#define                BUFSIZE         3
+#define                I_IDLE          1
+#define                I_WORK          2
+#define                I_HANDLERA      3
+#define                I_HANDLERB      4
+#define                I_DEVA          5
+#define                I_DEVB          6
+#define                PKTBIT          1
+#define                WAITBIT         2
+#define                HOLDBIT         4
+#define                NOTPKTBIT       !1
+#define                NOTWAITBIT      !2
+#define                NOTHOLDBIT      0XFFFB
+
+#define                S_RUN           0
+#define                S_RUNPKT        1
+#define                S_WAIT          2
+#define                S_WAITPKT       3
+#define                S_HOLD          4
+#define                S_HOLDPKT       5
+#define                S_HOLDWAIT      6
+#define                S_HOLDWAITPKT   7
+
+#define                K_DEV           1000
+#define                K_WORK          1001
+
+struct packet
+{
+    struct packet  *p_link;
+    int             p_id;
+    int             p_kind;
+    int             p_a1;
+    char            p_a2[BUFSIZE+1];
+};
+
+struct task
+{
+    struct task    *t_link;
+    int             t_id;
+    int             t_pri;
+    struct packet  *t_wkq;
+    int             t_state;
+    struct task    *(*t_fn)(struct packet *);
+    uintptr_t       t_v1;
+    uintptr_t       t_v2;
+};
+
+char  alphabet[28] = "0ABCDEFGHIJKLMNOPQRSTUVWXYZ";
+struct task *tasktab[11]  =  {(struct task *)10,0,0,0,0,0,0,0,0,0,0};
+struct task *tasklist    =  0;
+struct task *tcb;
+long    taskid;
+uintptr_t v1;
+uintptr_t v2;
+int     qpktcount    =  0;
+int     holdcount    =  0;
+int     tracing      =  0;
+int     layout       =  0;
+
+void append(struct packet *pkt, struct packet *ptr);
+
+void createtask(int id,
+                int pri,
+                struct packet *wkq,
+                int state,
+                struct task *(*fn)(struct packet *),
+                uintptr_t v1,
+                uintptr_t v2)
+{
+    struct task *t = (struct task *)malloc(sizeof(struct task));
+
+    tasktab[id] = t;
+    t->t_link   = tasklist;
+    t->t_id     = id;
+    t->t_pri    = pri;
+    t->t_wkq    = wkq;
+    t->t_state  = state;
+    t->t_fn     = fn;
+    t->t_v1     = v1;
+    t->t_v2     = v2;
+    tasklist    = t;
+}
+
+struct packet *pkt(struct packet *link, int id, int kind)
+{
+    int i;
+    struct packet *p = (struct packet *)malloc(sizeof(struct packet));
+
+    for (i=0; i<=BUFSIZE; i++)
+        p->p_a2[i] = 0;
+
+    p->p_link = link;
+    p->p_id = id;
+    p->p_kind = kind;
+    p->p_a1 = 0;
+
+    return (p);
+}
+
+void trace(char a)
+{
+   if ( --layout <= 0 )
+   {
+        // printf("\n");
+        layout = 50;
+    }
+
+    // printf("%c", a);
+}
+
+void schedule()
+{
+    while ( tcb != 0 )
+    {
+        struct packet *pkt;
+        struct task *newtcb;
+
+        pkt=0;
+
+        switch ( tcb->t_state )
+        {
+            case S_WAITPKT:
+                pkt = tcb->t_wkq;
+                tcb->t_wkq = pkt->p_link;
+                tcb->t_state = tcb->t_wkq == 0 ? S_RUN : S_RUNPKT;
+
+            case S_RUN:
+            case S_RUNPKT:
+                taskid = tcb->t_id;
+                v1 = tcb->t_v1;
+                v2 = tcb->t_v2;
+                if (tracing) {
+		  trace(taskid+'0');
+		}
+                newtcb = (*(tcb->t_fn))(pkt);
+                tcb->t_v1 = v1;
+                tcb->t_v2 = v2;
+                tcb = newtcb;
+                break;
+
+            case S_WAIT:
+            case S_HOLD:
+            case S_HOLDPKT:
+            case S_HOLDWAIT:
+            case S_HOLDWAITPKT:
+                tcb = tcb->t_link;
+                break;
+
+            default:
+                return;
+        }
+    }
+}
+
+struct task *wait_task(void)
+{
+    tcb->t_state |= WAITBIT;
+    return (tcb);
+}
+
+struct task *holdself(void)
+{
+    ++holdcount;
+    tcb->t_state |= HOLDBIT;
+    return (tcb->t_link) ;
+}
+
+struct task *findtcb(int id)
+{
+    struct task *t = 0;
+
+    if (1<=id && id<=(long)tasktab[0])
+    t = tasktab[id];
+    // if (t==0) printf("\nBad task id %d\n", id);
+    return(t);
+}
+
+struct task *release(int id)
+{
+    struct task *t;
+
+    t = findtcb(id);
+    if ( t==0 ) return (0);
+
+    t->t_state &= NOTHOLDBIT;
+    if ( t->t_pri > tcb->t_pri ) return (t);
+
+    return (tcb) ;
+}
+
+
+struct task *qpkt(struct packet *pkt)
+{
+    struct task *t;
+
+    t = findtcb(pkt->p_id);
+    if (t==0) return (t);
+
+    qpktcount++;
+
+    pkt->p_link = 0;
+    pkt->p_id = taskid;
+
+   if (t->t_wkq==0)
+    {
+        t->t_wkq = pkt;
+        t->t_state |= PKTBIT;
+        if (t->t_pri > tcb->t_pri) return (t);
+    }
+    else
+    {
+        append(pkt, (struct packet *)&(t->t_wkq));
+    }
+
+    return (tcb);
+}
+
+struct task *idlefn(struct packet *pkt)
+{
+    if ( --v2==0 ) return ( holdself() );
+
+    if ( (v1&1) == 0 )
+    {
+        v1 = ( v1>>1) & MAXINT;
+        return ( release(I_DEVA) );
+    }
+    else
+    {
+        v1 = ( (v1>>1) & MAXINT) ^ 0XD008;
+        return ( release(I_DEVB) );
+    }
+}
+
+struct task *workfn(struct packet *pkt)
+{
+    if ( pkt==0 ) return ( wait_task() );
+    else
+    {
+        int i;
+
+        v1 = I_HANDLERA + I_HANDLERB - v1;
+        pkt->p_id = v1;
+
+        pkt->p_a1 = 0;
+        for (i=0; i<=BUFSIZE; i++)
+        {
+            v2++;
+            if ( v2 > 26 ) v2 = 1;
+            (pkt->p_a2)[i] = alphabet[v2];
+        }
+        return ( qpkt(pkt) );
+    }
+}
+
+struct task *handlerfn(struct packet *pkt)
+{
+  if ( pkt!=0) {
+    append(pkt, (struct packet *)(pkt->p_kind==K_WORK ? &v1 : &v2));
+  }
+
+  if ( v1!=0 ) {
+    int count;
+    struct packet *workpkt = (struct packet *)v1;
+    count = workpkt->p_a1;
+
+    if ( count > BUFSIZE ) {
+      v1 = (uintptr_t)(((struct packet *)v1)->p_link);
+      return ( qpkt(workpkt) );
+    }
+
+    if ( v2!=0 ) {
+      struct packet *devpkt;
+
+      devpkt = (struct packet *)v2;
+      v2 = (uintptr_t)(((struct packet *)v2)->p_link);
+      devpkt->p_a1 = workpkt->p_a2[count];
+      workpkt->p_a1 = count+1;
+      return( qpkt(devpkt) );
+    }
+  }
+  return wait_task();
+}
+
+struct task *devfn(struct packet *pkt)
+{
+    if ( pkt==0 )
+    {
+        if ( v1==0 ) return ( wait_task() );
+        pkt = (struct packet *)v1;
+        v1 = 0;
+        return ( qpkt(pkt) );
+    }
+    else
+    {
+        v1 = (uintptr_t)pkt;
+        if (tracing) trace(pkt->p_a1);
+        return ( holdself() );
+    }
+}
+
+void append(struct packet *pkt, struct packet *ptr)
+{
+    pkt->p_link = 0;
+
+    while ( ptr->p_link ) ptr = ptr->p_link;
+
+    ptr->p_link = pkt;
+}
+
+int bench() {
+    struct packet *wkq = 0;
+
+    // printf("Bench mark starting\n");
+
+    createtask(I_IDLE, 0, wkq, S_RUN, idlefn, 1, Count);
+
+    wkq = pkt(0, 0, K_WORK);
+    wkq = pkt(wkq, 0, K_WORK);
+
+    createtask(I_WORK, 1000, wkq, S_WAITPKT, workfn, I_HANDLERA, 0);
+
+    wkq = pkt(0, I_DEVA, K_DEV);
+    wkq = pkt(wkq, I_DEVA, K_DEV);
+    wkq = pkt(wkq, I_DEVA, K_DEV);
+
+    // createtask(I_HANDLERA, 2000, wkq, S_WAITPKT, handlerfn, 0, 0);
+
+    // wkq = pkt(0, I_DEVB, K_DEV);
+    // wkq = pkt(wkq, I_DEVB, K_DEV);
+    // wkq = pkt(wkq, I_DEVB, K_DEV);
+
+    // createtask(I_HANDLERB, 3000, wkq, S_WAITPKT, handlerfn, 0, 0);
+
+    // wkq = 0;
+    // createtask(I_DEVA, 4000, wkq, S_WAIT, devfn, 0, 0);
+    // createtask(I_DEVB, 5000, wkq, S_WAIT, devfn, 0, 0);
+
+    tcb = tasklist;
+
+    qpktcount = holdcount = 0;
+
+    // printf("Starting\n");
+
+    tracing = FALSE;
+    layout = 0;
+
+    schedule();
+
+    // printf("\nfinished\n");
+
+
+
+
+    // if (!(qpktcount == Qpktcountval && holdcount == Holdcountval)) {
+    //     printf("qpkt count = %d  holdcount = %d\n", qpktcount, holdcount);
+    //     printf("These results are incorrect");
+    //     exit(1);
+    // }
+    // printf("\nend of run\n");
+    return qpktcount;
+}
+
+
+int inner_loop(int inner) {
+    int r = 0;
+    while (inner > 0) {
+        r += bench();
+        inner--;
+    }
+    return r;
+}
+
+int main(void)
+{
+    //INITREGULARALLOC();
+    int iterations = 2;
+    int warmup     = 0;
+    int inner_iterations = 2;
+
+    // parse_argv(argc, argv, &iterations, &warmup, &inner_iterations);
+
+    // while(1);
+
+    int result = 0;
+    while (iterations > 0) {
+        // unsigned long start = microseconds();
+        result += inner_loop(inner_iterations);
+        // unsigned long elapsed = microseconds() - start;
+        // printf("Richards: iterations=1 runtime: %lu%s\n", elapsed, "us");
+        // print(elapsed);
+        iterations--;
+    }
+
+}

Tests/isa/Cprograms/link.ld

@@ -15,5 +15,5 @@ SECTIONS
   _end = .;
 
   __malloc_start = .;
-  . = . + 512;
+  . = . + 0x10000;
 }

Tests/isa/Cprograms/test.c

@@ -4,41 +4,41 @@
 
 void* malloc(size_t size);
 
-int main(void)
-{
-    char* a = malloc(50);
-    char* b = malloc(160);
-
-    // asm volatile("fence rw, rw");   // make sure store completes
-
-    // volatile uint64_t* marker = (uint64_t*)0x80001000;
-
-    // *marker = 0xCAFEBABECAFED00D;   // UNIQUE SIGNATURE
+int main(void) {
+    // char *a = malloc(30);
+    // char *b = malloc(16);
 
 
-    // if (!a || !b) return;
+    // // if (!a || !b) return -1;
 
-    a[0] = 'A';
-
-    // // a[1] = 'C';
-    // // b[0] = 'B';
+    // a[0] = 'A';
+    // a[1] = 'C';
+    // b[0] = 'B';
 
     // // This will fault (out-of-bounds)
     // // a[20] = 'X';
 
-    if (a[0] != 'A') {
-        while (1);
-    }
-    // // char* a = malloc(100);
-    // // char* b = malloc(200);
+    // if (a[1] != 'C') {
+    //     while (1);
+    // }
 
-    // a[0] = 'A';     // triggers page fault if unmapped
-    // b[19] = 'Z';
+    // if (b[0] != 'B') {
+    //     while (1);
+    // }
+
+    // // free(b);
+    // // b = NULL;
+
+    // // if (b[0] == 'B') {
+    // //     while (1);
+    // // }
+
+    // char *c = malloc(16);
+
+    // // if (c[0] != 'B') {
+    // //     while (1);
+    // // }
 
-    // for (int i = 0; i < 1000; i++) {
-    //     a[i * 64] = i;   // each write new cache line
-    // }   
 
-    // while(1);
     return 0;
 }

Tests/isa/Cprograms/vm.c

@@ -333,7 +333,7 @@ void vm_boot(uintptr_t test_addr)
   trapframe_t tf;
   memset(&tf, 0, sizeof(tf));
   tf.epc = test_addr - DRAM_BASE;
-  // call test function
+  // call C benchmark function
   main();
   pop_tf(&tf);
 }

analysis/graph.py

@@ -0,0 +1,92 @@
+import matplotlib.pyplot as plt
+import numpy as np
+
+performance_data = {
+    "Mem": {
+        "regular": {"CPU_cycles": 33940, "L1_DTLB_access": 13202, "L1_DTLB_miss": 36, "L2_DTLB_miss": 10},
+        "modified": {"CPU_cycles": 2457, "L1_DTLB_access": 151, "L1_DTLB_miss": 0, "L2_DTLB_miss": 0},
+    },
+    "Test_C": {
+        "regular": {"CPU_cycles": 34040, "L1_DTLB_access": 13284, "L1_DTLB_miss": 36, "L2_DTLB_miss": 10},
+        "modified": {"CPU_cycles": 2789, "L1_DTLB_access": 396, "L1_DTLB_miss": 0, "L2_DTLB_miss": 0},
+    },
+    "Glibc": {
+        "regular": {"CPU_cycles": 36053, "L1_DTLB_access": 14027, "L1_DTLB_miss": 36, "L2_DTLB_miss": 10},
+        "modified": {"CPU_cycles": 4219, "L1_DTLB_access": 783, "L1_DTLB_miss": 0, "L2_DTLB_miss": 0},
+    },
+    "Richards": {
+        "regular": {"CPU_cycles": 37808, "L1_DTLB_access": 14363, "L1_DTLB_miss": 10, "L2_DTLB_miss": 6},
+        "modified": {"CPU_cycles": 4433, "L1_DTLB_access": 874, "L1_DTLB_miss": 0, "L2_DTLB_miss": 0},
+    },
+    "Matrix_mul": {
+        "regular": {"CPU_cycles": 42383, "L1_DTLB_access": 17020, "L1_DTLB_miss": 14, "L2_DTLB_miss": 10},
+        "modified": {"CPU_cycles": 12149, "L1_DTLB_access": 4411, "L1_DTLB_miss": 0, "L2_DTLB_miss": 0},
+    }
+}
+
+def compute_percent_diff(regular, modified):
+    out = {}
+    for k in regular:
+        if k in modified and regular[k] != 0:
+            out[k] = ((regular[k] - modified[k]) / regular[k]) * 100
+        elif k in modified and regular[k] == 0 and modified[k] == 0:
+            out[k] = 0.0
+        elif k in modified and regular[k] == 0 and modified[k] != 0:
+            out[k] = -100.0 
+    return out
+
+for t, d in performance_data.items():
+    d["percent_diff"] = compute_percent_diff(d["regular"], d["modified"])
+
+tests = list(performance_data.keys())
+metrics = ["CPU_cycles", "L1_DTLB_access", "L1_DTLB_miss", "L2_DTLB_miss"]
+
+# 1. Summary plot: Percent Improvement for all metrics grouped by test
+x = np.arange(len(tests))
+width = 0.2
+fig, ax = plt.subplots(figsize=(12, 7))
+
+for i, metric in enumerate(metrics):
+    values = [performance_data[t]["percent_diff"].get(metric, 0) for t in tests]
+    offset = (i - len(metrics)/2 + 0.5) * width
+    ax.bar(x + offset, values, width, label=metric.replace('_', ' ').title())
+
+ax.set_ylabel('Percent Improvement (%)')
+ax.set_title('Percent Improvement (Regular -> Modified) by Test and Metric')
+ax.set_xticks(x)
+ax.set_xticklabels(tests)
+ax.legend()
+ax.set_ylim(0, 110)
+ax.grid(axis='y', linestyle='--', alpha=0.7)
+fig.tight_layout()
+plt.savefig('percent_improvement_summary.png')
+
+# 2. Individual metric plots: Regular vs Modified grouped by test
+y_label_map = {
+    "CPU_cycles": "Cycles",
+    "L1_DTLB_access": "Access Count",
+    "L1_DTLB_miss": "Miss Count",
+    "L2_DTLB_miss": "Miss Count"
+}
+
+for metric in metrics:
+    fig, ax = plt.subplots(figsize=(10, 6))
+    reg_vals = [performance_data[t]["regular"][metric] for t in tests]
+    mod_vals = [performance_data[t]["modified"][metric] for t in tests]
+    
+    x_local = np.arange(len(tests))
+    w = 0.35
+    
+    ax.bar(x_local - w/2, reg_vals, w, label='Regular', color='steelblue')
+    ax.bar(x_local + w/2, mod_vals, w, label='Modified', color='salmon')
+    
+    # Setting dynamic y-label
+    ax.set_ylabel(y_label_map.get(metric, metric.replace('_', ' ').title()))
+    ax.set_title(f'Comparison: {metric.replace("_", " ").title()}')
+    ax.set_xticks(x_local)
+    ax.set_xticklabels(tests)
+    ax.legend()
+    ax.grid(axis='y', linestyle='--', alpha=0.7)
+    plt.xticks(rotation=45)
+    plt.tight_layout()
+    plt.savefig(f'grouped_comparison_{metric}.png')

analysis/perf_data.py

@@ -0,0 +1,145 @@
+def compute_percent_diff(regular, modified):
+    percent_diff = {}
+    for key in regular:
+        if key in modified and regular[key] != 0:
+            percent_diff[key] = ((regular[key] - modified[key]) / regular[key]) * 100
+    return percent_diff
+
+def compute_percent_diff_mod_vs_pte(modified, after_pte):
+    percent_diff = {}
+    for key in ["CPU_cycles", "L1_DTLB_access"]:
+        if key in modified and key in after_pte and modified[key] != 0:
+            percent_diff[key] = ((modified[key] - after_pte[key]) / modified[key]) * 100
+    return percent_diff
+
+
+performance_data = {
+    "Mem": {
+        "regular": {
+            "CPU_cycles": 33940,
+            "L1_DTLB_access": 13202,
+            "L1_DTLB_miss": 36,
+            "L2_DTLB_miss": 10,
+        },
+        "modified": {
+            "CPU_cycles": 2457,
+            "L1_DTLB_access": 151,
+            "L1_DTLB_miss": 0,
+            "L2_DTLB_miss": 0,
+        },
+        "after_PTE_setup": {
+            "CPU_cycles": 531,
+            "L1_DTLB_access": 25,
+            "L1_DTLB_miss": 0,
+            "L2_DTLB_miss": 0,
+        },
+    },
+
+    "Test_C": {
+        "regular": {
+            "CPU_cycles": 34040,
+            "L1_DTLB_access": 13284,
+            "L1_DTLB_miss": 36,
+            "L2_DTLB_miss": 10,
+        },
+        "modified": {
+            "CPU_cycles": 2789,
+            "L1_DTLB_access": 396,
+            "L1_DTLB_miss": 0,
+            "L2_DTLB_miss": 0,
+        },
+        "after_PTE_setup": {
+            "CPU_cycles": 631,
+            "L1_DTLB_access": 107,
+            "L1_DTLB_miss": 0,
+            "L2_DTLB_miss": 0,
+        },
+    },
+
+    "Glibc": {
+        "regular": {
+            "CPU_cycles": 36053,
+            "L1_DTLB_access": 14027,
+            "L1_DTLB_miss": 36,
+            "L2_DTLB_miss": 10,
+        },
+        "modified": {
+            "CPU_cycles": 4219,
+            "L1_DTLB_access": 783,
+            "L1_DTLB_miss": 0,
+            "L2_DTLB_miss": 0,
+        },
+        "after_PTE_setup": {
+            "CPU_cycles": 2644,
+            "L1_DTLB_access": 850,
+            "L1_DTLB_miss": 0,
+            "L2_DTLB_miss": 0,
+        },
+    },
+
+    "Setup": {
+        "regular": {
+            "CPU_cycles": 33409,
+            "L1_DTLB_access": 13177,
+            "L1_DTLB_miss": 36,
+            "L2_DTLB_miss": 10,
+        }
+    },
+
+    "Richards": {
+        "regular": {
+            "CPU_cycles": 37808,
+            "L1_DTLB_access": 14363,
+            "L1_DTLB_miss": 10,
+            "L2_DTLB_miss": 6,
+        },
+        "modified": {
+            "CPU_cycles": 4433,
+            "L1_DTLB_access": 874,
+            "L1_DTLB_miss": 0,
+            "L2_DTLB_miss": 0,
+        },
+        "after_PTE_setup": {
+            "CPU_cycles": 4399,
+            "L1_DTLB_access": 1186,
+        },
+    },
+
+    "Matrix_mul": {
+        "regular": {
+            "CPU_cycles": 42383,
+            "L1_DTLB_access": 17020,
+            "L1_DTLB_miss": 14,
+            "L2_DTLB_miss": 10,
+        },
+        "modified": {
+            "CPU_cycles": 12149,
+            "L1_DTLB_access": 4411,
+            "L1_DTLB_miss": 0,
+            "L2_DTLB_miss": 0,
+        },
+        "after_PTE_setup": {
+            "CPU_cycles": 8974,
+            "L1_DTLB_access": 3843,
+        },
+    }
+}
+
+# Add percentage differences
+for test, data in performance_data.items():
+    if "regular" in data and "modified" in data:
+        data["percent_diff"] = compute_percent_diff(
+            data["regular"], data["modified"]
+        )
+
+
+# Add modified vs after_PTE_setup percentage differences
+for test, data in performance_data.items():
+    if "modified" in data and "after_PTE_setup" in data:
+        data["mod_vs_pte_percent_diff"] = compute_percent_diff_mod_vs_pte(
+            data["modified"], data["after_PTE_setup"]
+        )
+
+# Example usage:
+print(performance_data["Mem"]["percent_diff"])
+print(performance_data["Mem"]["mod_vs_pte_percent_diff"])

builds/RV64ACDFIMSUxCHERI_Toooba_bluesim/.gitignore

@@ -0,0 +1 @@
+*.txt

src_Core/RISCY_OOO/procs/lib/L2Tlb.bsv

@@ -545,6 +545,7 @@ module mkL2Tlb(L2Tlb::L2Tlb);
             end
             else begin
                 dataSavedPageWalks.incr(saved);
+                $display("L2TlbdataSavedPageWalks = %0d", dataPageWalks);
             end
         end
 // `endif
@@ -742,6 +743,7 @@ module mkL2Tlb(L2Tlb::L2Tlb);
             end
             else begin
                 dataPageWalks.incr(1);
+                $display("L2TlbdataPageWalks = %0d", dataPageWalks);
             end
         end
 // `endif

test.txt

@@ -1,12 +1,5 @@
-Let me know if this message works well on behalf of all of us.
-
-Hi Jan,
-This is a very small gesture from Jose, Mathews, Yukang and Akilan. We have transferred 200 pounds equally contributed by all of us. As you are moving in and settling into your new place, we hope this can help you and Samantha cover certain expenses like buying furniture or for your very efficient interrail travel pass in the near future.
-
-We will definetely miss the german embassy in Slatefort. Happy new year in advance from all of us!
-
-Regards,
-Jose, Mathews, Yukang and Akilan
-(Your UK representation of transportation
-reform)
+- [ ] Benchmarks to port over:
+   - [ ] Barnes (3)
+   - [ ] Matrix multiplication (2)
+   - [ ] Richards (1)