github-mirrors/ibex
1
0
Fork 0
mirror of https://github.com/lowRISC/ibex.git synced 2025-06-27 17:00:41 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

[test] Performance evaluation

Browse source
This commit is contained in:
Hailin 2022-07-08 14:30:16 +02:00 • committed by Pirmin Vogel
parent 13bbaf72d9
commit 1ac25b8089
5 changed files with 233 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

2
examples/simple_system/rtl/ibex_simple_system.sv
View file

@ -306,7 +306,7 @@ module ibex_simple_system (
fpu_ss #(
.PULP_ZFINX ( 0 ),
.INPUT_BUFFER_DEPTH ( 2 ),
.INPUT_BUFFER_DEPTH ( 1 ),
.OUT_OF_ORDER ( 1 ),
.FORWARDING ( 1 ),
.FPU_FEATURES ( ),

8
examples/sw/simple_system/common/common.mk
View file

@ -27,8 +27,12 @@ OBJDUMP ?= $(CROSS_COMPILE)objdump
LINKER_SCRIPT ?= $(COMMON_DIR)/link.ld
CRT ?= $(COMMON_DIR)/crt0.S
CFLAGS ?= -march=$(ARCH) -mabi=ilp32 -static -mcmodel=medany -Wall -g -Os\
-fvisibility=hidden -nostdlib -nostartfiles -ffreestanding $(PROGRAM_CFLAGS)
# CFLAGS ?= -march=$(ARCH) -mabi=ilp32 -static -mcmodel=medany -Wall -g -Os\
# -fvisibility=hidden -nostdlib -nostartfiles -ffreestanding $(PROGRAM_CFLAGS)
# CFLAGS ?= -march=$(ARCH) -mabi=ilp32 -static -mcmodel=medany -Wall -O3 -ffast-math\
# -fvisibility=hidden -nostdlib -nostartfiles -ffreestanding $(PROGRAM_CFLAGS)
CFLAGS ?= -march=$(ARCH) -mabi=ilp32 -static -mcmodel=medlow -mtune=sifive-3-series \
-O3 -falign-functions=16 -funroll-all-loops -finline-functions -falign-jumps=4 -nostdlib -nostartfiles -ffreestanding -mstrict-align
OBJS := ${C_SRCS:.c=.o} ${ASM_SRCS:.S=.o} ${CRT:.S=.o}
DEPS = $(OBJS:%.o=%.d)

14
examples/sw/simple_system/matrix/Makefile Normal file
View file

@ -0,0 +1,14 @@
# Copyright lowRISC contributors.
# Licensed under the Apache License, Version 2.0, see LICENSE for details.
# SPDX-License-Identifier: Apache-2.0
#
# Generate a baremetal application
# Name of the program $(PROGRAM).c will be added as a source file
PROGRAM = matrix
PROGRAM_DIR := $(shell dirname $(realpath $(lastword $(MAKEFILE_LIST))))
# Any extra source files to include in the build. Use the upper case .S
# extension for assembly files
EXTRA_SRCS := matmulNxN.c
include ${PROGRAM_DIR}/../common/common.mk

30
examples/sw/simple_system/matrix/matmulNxN.c Normal file
View file

@ -0,0 +1,30 @@
/*
* Copyright 2020 ETH Zurich
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
*/
void matmulNxN(float* matA, float* matB , float* matC, const int N)
{
float tot;
for (int i = 0; i < N; ++i) {
for (int j = 0; j < N; ++j) {
tot = 0;
for (int k = 0; k < N; ++k) {
tot = tot + matA[i*N+k] * matB[k*N+j];
}
matC[i*N+j] = tot;
}
}
}

182
examples/sw/simple_system/matrix/matrix.c Normal file
View file

@ -0,0 +1,182 @@
/*
* Copyright 2020 ETH Zurich
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include <stdio.h>
#include <stdlib.h>
#include "simple_system_common.h"
void matmulNxN(float* matA, float* matB, float* matC, const int N);
#define N 25
float matA[N*N], matB[N*N];
float matC[N*N], matC_ref[N*N];
void activate_random_stall(void)
{
// Address vector for rnd_stall_reg, to control memory stalls/interrupt
volatile unsigned int *rnd_stall_reg[16];
// Setup the address vector
rnd_stall_reg[0] = 0x16000000;
for (int i = 1; i < 16; i++) {
rnd_stall_reg[i] = rnd_stall_reg[i-1] + 1; // It is a pointer to int ("+ 1" means "the next int")
}
/* The interposition of the stall generator between CPU and MEM should happen BEFORE the stall generetor is active */
// Interpose the stall generator between CPU and D-MEM (rnd_stall_reg[1])
*rnd_stall_reg[1] = 0x01;
// Interpose the stall generator between CPU and I-MEM (rnd_stall_reg[0])
*rnd_stall_reg[0] = 0x01;
// DATA MEMORY
// Set max n. stalls on both GNT and VALID for RANDOM mode (rnd_stall_reg[5])
*rnd_stall_reg[5] = 0x05;
// Set n. stalls on GNT (rnd_stall_reg[7])
*rnd_stall_reg[7] = 0x05;
// Set n. stalls on VALID (rnd_stall_reg[9])
*rnd_stall_reg[9] = 0x05;
// INSTRUCTION MEMORY
// Set max n. stalls on both GNT and VALID for RANDOM mode (rnd_stall_reg[4])
*rnd_stall_reg[4] = 0x05;
// Set n. stalls on GNT (rnd_stall_reg[6])
*rnd_stall_reg[6] = 0x05;
// Set n. stalls on VALID (rnd_stall_reg[8])
*rnd_stall_reg[8] = 0x05;
/* Activating stalls on D and I Mem has to be done as last operation. Do not change the order. */
// Set stall mode on D-MEM (off=0, standard=1, random=2) (rnd_stall_reg[3])
*rnd_stall_reg[3] = 0x02;
// Set stall mode on I-MEM (off=0, standard=1, random=2) (rnd_stall_reg[2])
*rnd_stall_reg[2] = 0x02;
}
int main(int argc, char *argv[])
{
#ifdef RANDOM_MEM_STALL
activate_random_stall();
#endif
for (int i = 0; i < N; ++i) {
matC_ref[i] = 9469.599609375f;
}
for (int i = N; i < 2*N; ++i) {
matC_ref[i] = 9402.7197265625f;
}
for (int i = 2*N; i < 3*N; ++i) {
matC_ref[i] = 9335.8408203125f;
}
for (int i = 3*N; i < 4*N; ++i) {
matC_ref[i] = 9268.9609375f;
}
for (int i = 4*N; i < 5*N; ++i) {
matC_ref[i] = 9202.0810546875f;
}
for (int i = 5*N; i < 6*N; ++i) {
matC_ref[i] = 9135.201171875f;
}
for (int i = 6*N; i < 7*N; ++i) {
matC_ref[i] = 9068.3203125f;
}
for (int i = 7*N; i < 8*N; ++i) {
matC_ref[i] = 9001.44140625f;
}
for (int i = 8*N; i < 9*N; ++i) {
matC_ref[i] = 8934.5625f;
}
for (int i = 9*N; i < 10*N; ++i){
matC_ref[i] = 8867.681640625f;
}
for (int i = 10*N; i < 11*N; ++i){
matC_ref[i] = 8800.802734375f;
}
for (int i = 11*N; i < 12*N; ++i){
matC_ref[i] = 8733.921875f;
}
for (int i = 12*N; i < 13*N; ++i){
matC_ref[i] = 8667.0419921875f;
}
for (int i = 13*N; i < 14*N; ++i){
matC_ref[i] = 8600.162109375f;
}
for (int i = 14*N; i < 15*N; ++i){
matC_ref[i] = 8533.283203125f;
}
for (int i = 15*N; i < 16*N; ++i){
matC_ref[i] = 8466.4033203125f;
}
for (int i = 16*N; i < 17*N; ++i){
matC_ref[i] = 8399.52246094f;
}
for (int i = 17*N; i < 18*N; ++i){
matC_ref[i] = 8332.642578125f;
}
for (int i = 18*N; i < 19*N; ++i){
matC_ref[i] = 8265.7646484375f;
}
for (int i = 19*N; i < 20*N; ++i){
matC_ref[i] = 8198.8837890625f;
}
for (int i = 20*N; i < 21*N; ++i){
matC_ref[i] = 8132.00341796875f;
}
for (int i = 21*N; i < 22*N; ++i){
matC_ref[i] = 8065.12451171875f;
}
for (int i = 22*N; i < 23*N; ++i){
matC_ref[i] = 7998.2451171875f;
}
for (int i = 23*N; i < 24*N; ++i){
matC_ref[i] = 7931.36474609375f;
}
for (int i = 24*N; i < 25*N; ++i){
matC_ref[i] = 7864.484375f;
}
float tmpA, tmpB, tmpC;
int error = 0;
tmpA = 62.299999237060546875f;
tmpB = 0.800000011920928955078125f;
tmpC = 0.439999997615814208984375f;
for (int i = 0; i < N; ++i) {
for (int j = 0; j < N; ++j) {
matA[i*N+j] = tmpA;
matB[i*N+j] = tmpB;
}
tmpA = tmpA - tmpC;
tmpB = tmpB + tmpC;
}
matmulNxN(matA, matB, matC, N);
for (int i = 0; i < N*N; ++i) {
if (matC_ref[i] != matC[i]) {
error++;
puts("Error!");
// printf("Error at index %d, expected %x, got %x\n", i, (*(int*)&matC_ref[i]), (*(int*)&matC[i]));
}
}
if (error)
return error;
else
return EXIT_SUCCESS;
}