Index: soft/giet_vm/Makefile =================================================================== --- soft/giet_vm/Makefile (revision 752) +++ soft/giet_vm/Makefile (revision 753) @@ -37,5 +37,6 @@ applications/shell/shell.py \ applications/sort/sort.py \ - applications/transpose/transpose.py + applications/transpose/transpose.py \ + applications/coremark/coremark.py # build the list of applications to be executed (used in the all rule) @@ -131,5 +132,5 @@ CFLAGS = -Wall -ffreestanding -mno-gpopt -mips32 -g -O2 \ - -fno-delete-null-pointer-checks + -fno-delete-null-pointer-checks $(XCFLAGS) GIET_INCLUDE = -Igiet_boot \ @@ -197,9 +198,10 @@ cd applications/sort && $(MAKE) clean && cd ../.. cd applications/transpose && $(MAKE) clean && cd ../.. + cd applications/coremark && $(MAKE) clean && cd ../.. ######################################## ### delete disk image clean-disk: - rm -rf hdd/ + rm -rf $(DISK_IMAGE) ######################################## @@ -223,4 +225,5 @@ mmd -o -i $< ::/bin/sort || true mmd -o -i $< ::/bin/transpose || true + mmd -o -i $< ::/bin/coremark || true mmd -o -i $< ::/misc || true mmd -o -i $< ::/home || true @@ -240,4 +243,5 @@ mcopy -o -i $< applications/sort/appli.elf ::/bin/sort || true mcopy -o -i $< applications/transpose/appli.elf ::/bin/transpose || true + mcopy -o -i $< applications/coremark/appli.elf ::/bin/coremark || true mcopy -o -i $< images/images_128.raw ::/misc mcopy -o -i $< images/philips_1024.raw ::/misc @@ -432,2 +436,7 @@ $(MAKE) -C applications/transpose +######################################## +### coremark compilation +applications/coremark/appli.elf: build/libs/libuser.a + $(MAKE) -C applications/coremark + Index: soft/giet_vm/applications/coremark/Makefile =================================================================== --- soft/giet_vm/applications/coremark/Makefile (revision 753) +++ soft/giet_vm/applications/coremark/Makefile (revision 753) @@ -0,0 +1,46 @@ + +APP_NAME = coremark + +OBJS= core_main.o \ + core_list_join.o \ + core_matrix.o \ + core_state.o \ + core_util.o \ + mach/core_portme.o + +LIBS= -L../../build/libs -luser + +INCLUDES = -I. -I./mach -I../.. -I../../giet_libs -I../../giet_xml + +DEFS := -DFLAGS_STR="\"${CFLAGS}\"" -DTOTAL_DATA_SIZE=2000 + +COREMARK ?= RUN1 +ifeq (${COREMARK},RUN1) + $(info Coremark Performance Run) + DEFS := ${DEFS} -DPERFORMANCE_RUN=1 +endif +ifeq (${COREMARK},RUN2) + $(info Coremark Validation Run) + DEFS := ${DEFS} -DVALIDATION_RUN=1 +endif +ifeq (${COREMARK},RUN3) + $(info Coremark Profile Run) + DEFS := ${DEFS} -DPROFILE_RUN=1 +endif + +#DEFS := ${DEFS} -DCORE_DEBUG=1 + +LIB_DEPS = ../../build/libs/libuser.a + +appli.elf: $(OBJS) $(APP_NAME).ld $(LIBS_DEPS) + $(LD) -o $@ -T $(APP_NAME).ld $(OBJS) $(LIBS) + $(DU) -D $@ > $@.txt + +check_coremark: coremark.md5 + md5sum -c $< + +%.o: %.c + $(CC) $(INCLUDES) $(CFLAGS) ${DEFS} -c -o $@ $< + +clean: + rm -f *.o *.elf appli.elf.txt *.pyc core *~ Index: soft/giet_vm/applications/coremark/core_list_join.c =================================================================== --- soft/giet_vm/applications/coremark/core_list_join.c (revision 753) +++ soft/giet_vm/applications/coremark/core_list_join.c (revision 753) @@ -0,0 +1,496 @@ +/* +Author : Shay Gal-On, EEMBC + +This file is part of EEMBC(R) and CoreMark(TM), which are Copyright (C) 2009 +All rights reserved. + +EEMBC CoreMark Software is a product of EEMBC and is provided under the terms of the +CoreMark License that is distributed with the official EEMBC COREMARK Software release. +If you received this EEMBC CoreMark Software without the accompanying CoreMark License, +you must discontinue use and download the official release from www.coremark.org. + +Also, if you are publicly displaying scores generated from the EEMBC CoreMark software, +make sure that you are in compliance with Run and Reporting rules specified in the accompanying readme.txt file. + +EEMBC +4354 Town Center Blvd. Suite 114-200 +El Dorado Hills, CA, 95762 +*/ + +#include "coremark.h" +/* +Topic: Description + Benchmark using a linked list. + + Linked list is a common data structure used in many applications. + + For our purposes, this will excercise the memory units of the processor. + In particular, usage of the list pointers to find and alter data. + + We are not using Malloc since some platforms do not support this library. + + Instead, the memory block being passed in is used to create a list, + and the benchmark takes care not to add more items then can be + accomodated by the memory block. The porting layer will make sure + that we have a valid memory block. + + All operations are done in place, without using any extra memory. + + The list itself contains list pointers and pointers to data items. + Data items contain the following: + + idx - An index that captures the initial order of the list. + data - Variable data initialized based on the input parameters. The 16b are divided as follows: + o Upper 8b are backup of original data. + o Bit 7 indicates if the lower 7 bits are to be used as is or calculated. + o Bits 0-2 indicate type of operation to perform to get a 7b value. + o Bits 3-6 provide input for the operation. + +*/ + +/* local functions */ + +list_head *core_list_find(list_head *list,list_data *info); +list_head *core_list_reverse(list_head *list); +list_head *core_list_remove(list_head *item); +list_head *core_list_undo_remove(list_head *item_removed, list_head *item_modified); +list_head *core_list_insert_new(list_head *insert_point + , list_data *info, list_head **memblock, list_data **datablock + , list_head *memblock_end, list_data *datablock_end); +typedef ee_s32(*list_cmp)(list_data *a, list_data *b, core_results *res); +list_head *core_list_mergesort(list_head *list, list_cmp cmp, core_results *res); + +ee_s16 calc_func(ee_s16 *pdata, core_results *res) { + ee_s16 data=*pdata; + ee_s16 retval; + ee_u8 optype=(data>>7) & 1; /* bit 7 indicates if the function result has been cached */ + if (optype) /* if cached, use cache */ + return (data & 0x007f); + else { /* otherwise calculate and cache the result */ + ee_s16 flag=data & 0x7; /* bits 0-2 is type of function to perform */ + ee_s16 dtype=((data>>3) & 0xf); /* bits 3-6 is specific data for the operation */ + dtype |= dtype << 4; /* replicate the lower 4 bits to get an 8b value */ + switch (flag) { + case 0: + if (dtype<0x22) /* set min period for bit corruption */ + dtype=0x22; + retval=core_bench_state(res->size,res->memblock[3],res->seed1,res->seed2,dtype,res->crc); + if (res->crcstate==0) + res->crcstate=retval; + break; + case 1: + retval=core_bench_matrix(&(res->mat),dtype,res->crc); + if (res->crcmatrix==0) + res->crcmatrix=retval; + break; + default: + retval=data; + break; + } + res->crc=crcu16(retval,res->crc); + retval &= 0x007f; + *pdata = (data & 0xff00) | 0x0080 | retval; /* cache the result */ + return retval; + } +} +/* Function: cmp_complex + Compare the data item in a list cell. + + Can be used by mergesort. +*/ +ee_s32 cmp_complex(list_data *a, list_data *b, core_results *res) { + ee_s16 val1=calc_func(&(a->data16),res); + ee_s16 val2=calc_func(&(b->data16),res); + return val1 - val2; +} + +/* Function: cmp_idx + Compare the idx item in a list cell, and regen the data. + + Can be used by mergesort. +*/ +ee_s32 cmp_idx(list_data *a, list_data *b, core_results *res) { + if (res==NULL) { + a->data16 = (a->data16 & 0xff00) | (0x00ff & (a->data16>>8)); + b->data16 = (b->data16 & 0xff00) | (0x00ff & (b->data16>>8)); + } + return a->idx - b->idx; +} + +void copy_info(list_data *to,list_data *from) { + to->data16=from->data16; + to->idx=from->idx; +} + +/* Benchmark for linked list: + - Try to find multiple data items. + - List sort + - Operate on data from list (crc) + - Single remove/reinsert + * At the end of this function, the list is back to original state +*/ +ee_u16 core_bench_list(core_results *res, ee_s16 finder_idx) { + ee_u16 retval=0; + ee_u16 found=0,missed=0; + list_head *list=res->list; + ee_s16 find_num=res->seed3; + list_head *this_find; + list_head *finder, *remover; + list_data info; + ee_s16 i; + + info.idx=finder_idx; + /* find values in the list, and change the list each time (reverse and cache if value found) */ + for (i=0; inext->info->data16 >> 8) & 1; + } + else { + found++; + if (this_find->info->data16 & 0x1) /* use found value */ + retval+=(this_find->info->data16 >> 9) & 1; + /* and cache next item at the head of the list (if any) */ + if (this_find->next != NULL) { + finder = this_find->next; + this_find->next = finder->next; + finder->next=list->next; + list->next=finder; + } + } + if (info.idx>=0) + info.idx++; +#if CORE_DEBUG + ee_printf("List find %d: [%d,%d,%d]\n",i,retval,missed,found); +#endif + } + retval+=found*4-missed; + /* sort the list by data content and remove one item*/ + if (finder_idx>0) + list=core_list_mergesort(list,cmp_complex,res); + remover=core_list_remove(list->next); + /* CRC data content of list from location of index N forward, and then undo remove */ + finder=core_list_find(list,&info); + if (!finder) + finder=list->next; + while (finder) { + retval=crc16(list->info->data16,retval); + finder=finder->next; + } +#if CORE_DEBUG + ee_printf("List sort 1: %04x\n",retval); +#endif + remover=core_list_undo_remove(remover,list->next); + /* sort the list by index, in effect returning the list to original state */ + list=core_list_mergesort(list,cmp_idx,NULL); + /* CRC data content of list */ + finder=list->next; + while (finder) { + retval=crc16(list->info->data16,retval); + finder=finder->next; + } +#if CORE_DEBUG + ee_printf("List sort 2: %04x\n",retval); +#endif + return retval; +} +/* Function: core_list_init + Initialize list with data. + + Parameters: + blksize - Size of memory to be initialized. + memblock - Pointer to memory block. + seed - Actual values chosen depend on the seed parameter. + The seed parameter MUST be supplied from a source that cannot be determined at compile time + + Returns: + Pointer to the head of the list. + +*/ +list_head *core_list_init(ee_u32 blksize, list_head *memblock, ee_s16 seed) { + /* calculated pointers for the list */ + ee_u32 per_item=16+sizeof(struct list_data_s); + ee_u32 size=(blksize/per_item)-2; /* to accomodate systems with 64b pointers, and make sure same code is executed, set max list elements */ + list_head *memblock_end=memblock+size; + list_data *datablock=(list_data *)(memblock_end); + list_data *datablock_end=datablock+size; + /* some useful variables */ + ee_u32 i; + list_head *finder,*list=memblock; + list_data info; + + /* create a fake items for the list head and tail */ + list->next=NULL; + list->info=datablock; + list->info->idx=0x0000; + list->info->data16=(ee_s16)0x8080; + memblock++; + datablock++; + info.idx=0x7fff; + info.data16=(ee_s16)0xffff; + core_list_insert_new(list,&info,&memblock,&datablock,memblock_end,datablock_end); + + /* then insert size items */ + for (i=0; inext; + i=1; + while (finder->next!=NULL) { + if (iinfo->idx=i++; + else { + ee_u16 pat=(ee_u16)(i++ ^ seed); /* get a pseudo random number */ + finder->info->idx=0x3fff & (((i & 0x07) << 8) | pat); /* make sure the mixed items end up after the ones in sequence */ + } + finder=finder->next; + } + list = core_list_mergesort(list,cmp_idx,NULL); +#if CORE_DEBUG + ee_printf("Initialized list:\n"); + finder=list; + while (finder) { + ee_printf("[%04x,%04x]",finder->info->idx,(ee_u16)finder->info->data16); + finder=finder->next; + } + ee_printf("\n"); +#endif + return list; +} + +/* Function: core_list_insert + Insert an item to the list + + Parameters: + insert_point - where to insert the item. + info - data for the cell. + memblock - pointer for the list header + datablock - pointer for the list data + memblock_end - end of region for list headers + datablock_end - end of region for list data + + Returns: + Pointer to new item. +*/ +list_head *core_list_insert_new(list_head *insert_point, list_data *info, list_head **memblock, list_data **datablock + , list_head *memblock_end, list_data *datablock_end) { + list_head *newitem; + + if ((*memblock+1) >= memblock_end) + return NULL; + if ((*datablock+1) >= datablock_end) + return NULL; + + newitem=*memblock; + (*memblock)++; + newitem->next=insert_point->next; + insert_point->next=newitem; + + newitem->info=*datablock; + (*datablock)++; + copy_info(newitem->info,info); + + return newitem; +} + +/* Function: core_list_remove + Remove an item from the list. + + Operation: + For a singly linked list, remove by copying the data from the next item + over to the current cell, and unlinking the next item. + + Note: + since there is always a fake item at the end of the list, no need to check for NULL. + + Returns: + Removed item. +*/ +list_head *core_list_remove(list_head *item) { + list_data *tmp; + list_head *ret=item->next; + /* swap data pointers */ + tmp=item->info; + item->info=ret->info; + ret->info=tmp; + /* and eliminate item */ + item->next=item->next->next; + ret->next=NULL; + return ret; +} + +/* Function: core_list_undo_remove + Undo a remove operation. + + Operation: + Since we want each iteration of the benchmark to be exactly the same, + we need to be able to undo a remove. + Link the removed item back into the list, and switch the info items. + + Parameters: + item_removed - Return value from the + item_modified - List item that was modified during + + Returns: + The item that was linked back to the list. + +*/ +list_head *core_list_undo_remove(list_head *item_removed, list_head *item_modified) { + list_data *tmp; + /* swap data pointers */ + tmp=item_removed->info; + item_removed->info=item_modified->info; + item_modified->info=tmp; + /* and insert item */ + item_removed->next=item_modified->next; + item_modified->next=item_removed; + return item_removed; +} + +/* Function: core_list_find + Find an item in the list + + Operation: + Find an item by idx (if not 0) or specific data value + + Parameters: + list - list head + info - idx or data to find + + Returns: + Found item, or NULL if not found. +*/ +list_head *core_list_find(list_head *list,list_data *info) { + if (info->idx>=0) { + while (list && (list->info->idx != info->idx)) + list=list->next; + return list; + } else { + while (list && ((list->info->data16 & 0xff) != info->data16)) + list=list->next; + return list; + } +} +/* Function: core_list_reverse + Reverse a list + + Operation: + Rearrange the pointers so the list is reversed. + + Parameters: + list - list head + info - idx or data to find + + Returns: + Found item, or NULL if not found. +*/ + +list_head *core_list_reverse(list_head *list) { + list_head *next=NULL, *tmp; + while (list) { + tmp=list->next; + list->next=next; + next=list; + list=tmp; + } + return next; +} +/* Function: core_list_mergesort + Sort the list in place without recursion. + + Description: + Use mergesort, as for linked list this is a realistic solution. + Also, since this is aimed at embedded, care was taken to use iterative rather then recursive algorithm. + The sort can either return the list to original order (by idx) , + or use the data item to invoke other other algorithms and change the order of the list. + + Parameters: + list - list to be sorted. + cmp - cmp function to use + + Returns: + New head of the list. + + Note: + We have a special header for the list that will always be first, + but the algorithm could theoretically modify where the list starts. + + */ +list_head *core_list_mergesort(list_head *list, list_cmp cmp, core_results *res) { + list_head *p, *q, *e, *tail; + ee_s32 insize, nmerges, psize, qsize, i; + + insize = 1; + + while (1) { + p = list; + list = NULL; + tail = NULL; + + nmerges = 0; /* count number of merges we do in this pass */ + + while (p) { + nmerges++; /* there exists a merge to be done */ + /* step `insize' places along from p */ + q = p; + psize = 0; + for (i = 0; i < insize; i++) { + psize++; + q = q->next; + if (!q) break; + } + + /* if q hasn't fallen off end, we have two lists to merge */ + qsize = insize; + + /* now we have two lists; merge them */ + while (psize > 0 || (qsize > 0 && q)) { + + /* decide whether next element of merge comes from p or q */ + if (psize == 0) { + /* p is empty; e must come from q. */ + e = q; q = q->next; qsize--; + } else if (qsize == 0 || !q) { + /* q is empty; e must come from p. */ + e = p; p = p->next; psize--; + } else if (cmp(p->info,q->info,res) <= 0) { + /* First element of p is lower (or same); e must come from p. */ + e = p; p = p->next; psize--; + } else { + /* First element of q is lower; e must come from q. */ + e = q; q = q->next; qsize--; + } + + /* add the next element to the merged list */ + if (tail) { + tail->next = e; + } else { + list = e; + } + tail = e; + } + + /* now p has stepped `insize' places along, and q has too */ + p = q; + } + + tail->next = NULL; + + /* If we have done only one merge, we're finished. */ + if (nmerges <= 1) /* allow for nmerges==0, the empty list case */ + return list; + + /* Otherwise repeat, merging lists twice the size */ + insize *= 2; + } +#if COMPILER_REQUIRES_SORT_RETURN + return list; +#endif +} Index: soft/giet_vm/applications/coremark/core_main.c =================================================================== --- soft/giet_vm/applications/coremark/core_main.c (revision 753) +++ soft/giet_vm/applications/coremark/core_main.c (revision 753) @@ -0,0 +1,356 @@ +/* +Author : Shay Gal-On, EEMBC + +This file is part of EEMBC(R) and CoreMark(TM), which are Copyright (C) 2009 +All rights reserved. + +EEMBC CoreMark Software is a product of EEMBC and is provided under the terms of the +CoreMark License that is distributed with the official EEMBC COREMARK Software release. +If you received this EEMBC CoreMark Software without the accompanying CoreMark License, +you must discontinue use and download the official release from www.coremark.org. + +Also, if you are publicly displaying scores generated from the EEMBC CoreMark software, +make sure that you are in compliance with Run and Reporting rules specified in the accompanying readme.txt file. + +EEMBC +4354 Town Center Blvd. Suite 114-200 +El Dorado Hills, CA, 95762 +*/ +/* File: core_main.c + This file contains the framework to acquire a block of memory, seed initial parameters, tun t he benchmark and report the results. +*/ +#include "coremark.h" + +/* Function: iterate + Run the benchmark for a specified number of iterations. + + Operation: + For each type of benchmarked algorithm: + a - Initialize the data block for the algorithm. + b - Execute the algorithm N times. + + Returns: + NULL. +*/ +static ee_u16 list_known_crc[] = {(ee_u16)0xd4b0,(ee_u16)0x3340,(ee_u16)0x6a79,(ee_u16)0xe714,(ee_u16)0xe3c1}; +static ee_u16 matrix_known_crc[] = {(ee_u16)0xbe52,(ee_u16)0x1199,(ee_u16)0x5608,(ee_u16)0x1fd7,(ee_u16)0x0747}; +static ee_u16 state_known_crc[] = {(ee_u16)0x5e47,(ee_u16)0x39bf,(ee_u16)0xe5a4,(ee_u16)0x8e3a,(ee_u16)0x8d84}; +void *iterate(void *pres) { + ee_u32 i; + ee_u16 crc; + core_results *res=(core_results *)pres; + ee_u32 iterations=res->iterations; + res->crc=0; + res->crclist=0; + res->crcmatrix=0; + res->crcstate=0; + + for (i=0; icrc=crcu16(crc,res->crc); + crc=core_bench_list(res,-1); + res->crc=crcu16(crc,res->crc); + if (i==0) res->crclist=res->crc; + } + return NULL; +} + +#if (SEED_METHOD==SEED_ARG) +ee_s32 get_seed_args(int i, int argc, char *argv[]); +#define get_seed(x) (ee_s16)get_seed_args(x,argc,argv) +#define get_seed_32(x) get_seed_args(x,argc,argv) +#else /* via function or volatile */ +ee_s32 get_seed_32(int i); +#define get_seed(x) (ee_s16)get_seed_32(x) +#endif + +#if (MEM_METHOD==MEM_STATIC) +ee_u8 static_memblk[TOTAL_DATA_SIZE]; +#endif +char *mem_name[3] = {"Static","Heap","Stack"}; +/* Function: main + Main entry routine for the benchmark. + This function is responsible for the following steps: + + 1 - Initialize input seeds from a source that cannot be determined at compile time. + 2 - Initialize memory block for use. + 3 - Run and time the benchmark. + 4 - Report results, testing the validity of the output if the seeds are known. + + Arguments: + 1 - first seed : Any value + 2 - second seed : Must be identical to first for iterations to be identical + 3 - third seed : Any value, should be at least an order of magnitude less then the input size, but bigger then 32. + 4 - Iterations : Special, if set to 0, iterations will be automatically determined such that the benchmark will run between 10 to 100 secs + +*/ + +#if MAIN_HAS_NOARGC +MAIN_RETURN_TYPE main(void) { + int argc=0; + char *argv[1]; +#else +MAIN_RETURN_TYPE main(int argc, char *argv[]) { +#endif + ee_u16 i,j=0,num_algorithms=0; + ee_s16 known_id=-1,total_errors=0; + ee_u16 seedcrc=0; + CORE_TICKS total_time; + core_results results[MULTITHREAD]; +#if (MEM_METHOD==MEM_STACK) + ee_u8 stack_memblock[TOTAL_DATA_SIZE*MULTITHREAD]; +#endif + /* first call any initializations needed */ + portable_init(&(results[0].port), &argc, argv); + /* First some checks to make sure benchmark will run ok */ + if (sizeof(struct list_head_s)>128) { + ee_printf("list_head structure too big for comparable data!\n"); + return MAIN_RETURN_VAL; + } + results[0].seed1=get_seed(1); + results[0].seed2=get_seed(2); + results[0].seed3=get_seed(3); + results[0].iterations=get_seed_32(4); +#if CORE_DEBUG + results[0].iterations=1; +#endif + results[0].execs=get_seed_32(5); + if (results[0].execs==0) { /* if not supplied, execute all algorithms */ + results[0].execs=ALL_ALGORITHMS_MASK; + } + /* put in some default values based on one seed only for easy testing */ + if ((results[0].seed1==0) && (results[0].seed2==0) && (results[0].seed3==0)) { /* validation run */ + results[0].seed1=0; + results[0].seed2=0; + results[0].seed3=0x66; + } + if ((results[0].seed1==1) && (results[0].seed2==0) && (results[0].seed3==0)) { /* perfromance run */ + results[0].seed1=0x3415; + results[0].seed2=0x3415; + results[0].seed3=0x66; + } +#if (MEM_METHOD==MEM_STATIC) + results[0].memblock[0]=(void *)static_memblk; + results[0].size=TOTAL_DATA_SIZE; + results[0].err=0; + #if (MULTITHREAD>1) + #error "Cannot use a static data area with multiple contexts!" + #endif +#elif (MEM_METHOD==MEM_MALLOC) + for (i=0 ; i1) + if (default_num_contexts>MULTITHREAD) { + default_num_contexts=MULTITHREAD; + } + for (i=0 ; i=0) { + for (i=0 ; i 0) + ee_printf("Iterations/Sec : %f\n",default_num_contexts*results[0].iterations/time_in_secs(total_time)); +#else + ee_printf("Total time (secs): %d\n",time_in_secs(total_time)); + if (time_in_secs(total_time) > 0) + ee_printf("Iterations/Sec : %d\n",default_num_contexts*results[0].iterations/time_in_secs(total_time)); +#endif + if (time_in_secs(total_time) < 10) { + ee_printf("ERROR! Must execute for at least 10 secs for a valid result!\n"); + total_errors++; + } + + ee_printf("Iterations : %lu\n",(ee_u32)default_num_contexts*results[0].iterations); + ee_printf("Compiler version : %s\n",COMPILER_VERSION); + ee_printf("Compiler flags : %s\n",COMPILER_FLAGS); +#if (MULTITHREAD>1) + ee_printf("Parallel %s : %d\n",PARALLEL_METHOD,default_num_contexts); +#endif + ee_printf("Memory location : %s\n",MEM_LOCATION); + /* output for verification */ + ee_printf("seedcrc : 0x%04x\n",seedcrc); + if (results[0].execs & ID_LIST) + for (i=0 ; i1) + ee_printf(" / %d:%s",default_num_contexts,PARALLEL_METHOD); +#endif + ee_printf("\n"); + } +#endif + } + if (total_errors>0) + ee_printf("Errors detected\n"); + if (total_errors<0) + ee_printf("Cannot validate operation for these seed values, please compare with results on a known platform.\n"); + +#if (MEM_METHOD==MEM_MALLOC) + for (i=0 ; i>(from)) & (~(0xffffffff << (to)))) + +#if CORE_DEBUG +void printmat(MATDAT *A, ee_u32 N, char *name) { + ee_u32 i,j; + ee_printf("Matrix %s [%dx%d]:\n",name,N,N); + for (i=0; i N times, + changing the matrix values slightly by a constant amount each time. +*/ +ee_u16 core_bench_matrix(mat_params *p, ee_s16 seed, ee_u16 crc) { + ee_u32 N=p->N; + MATRES *C=p->C; + MATDAT *A=p->A; + MATDAT *B=p->B; + MATDAT val=(MATDAT)seed; + + crc=crc16(matrix_test(N,C,A,B,val),crc); + + return crc; +} + +/* Function: matrix_test + Perform matrix manipulation. + + Parameters: + N - Dimensions of the matrix. + C - memory for result matrix. + A - input matrix + B - operator matrix (not changed during operations) + + Returns: + A CRC value that captures all results calculated in the function. + In particular, crc of the value calculated on the result matrix + after each step by . + + Operation: + + 1 - Add a constant value to all elements of a matrix. + 2 - Multiply a matrix by a constant. + 3 - Multiply a matrix by a vector. + 4 - Multiply a matrix by a matrix. + 5 - Add a constant value to all elements of a matrix. + + After the last step, matrix A is back to original contents. +*/ +ee_s16 matrix_test(ee_u32 N, MATRES *C, MATDAT *A, MATDAT *B, MATDAT val) { + ee_u16 crc=0; + MATDAT clipval=matrix_big(val); + + matrix_add_const(N,A,val); /* make sure data changes */ +#if CORE_DEBUG + printmat(A,N,"matrix_add_const"); +#endif + matrix_mul_const(N,C,A,val); + crc=crc16(matrix_sum(N,C,clipval),crc); +#if CORE_DEBUG + printmatC(C,N,"matrix_mul_const"); +#endif + matrix_mul_vect(N,C,A,B); + crc=crc16(matrix_sum(N,C,clipval),crc); +#if CORE_DEBUG + printmatC(C,N,"matrix_mul_vect"); +#endif + matrix_mul_matrix(N,C,A,B); + crc=crc16(matrix_sum(N,C,clipval),crc); +#if CORE_DEBUG + printmatC(C,N,"matrix_mul_matrix"); +#endif + matrix_mul_matrix_bitextract(N,C,A,B); + crc=crc16(matrix_sum(N,C,clipval),crc); +#if CORE_DEBUG + printmatC(C,N,"matrix_mul_matrix_bitextract"); +#endif + + matrix_add_const(N,A,-val); /* return matrix to initial value */ + return crc; +} + +/* Function : matrix_init + Initialize the memory block for matrix benchmarking. + + Parameters: + blksize - Size of memory to be initialized. + memblk - Pointer to memory block. + seed - Actual values chosen depend on the seed parameter. + p - pointers to containing initialized matrixes. + + Returns: + Matrix dimensions. + + Note: + The seed parameter MUST be supplied from a source that cannot be determined at compile time +*/ +ee_u32 core_init_matrix(ee_u32 blksize, void *memblk, ee_s32 seed, mat_params *p) { + ee_u32 N=0; + MATDAT *A; + MATDAT *B; + ee_s32 order=1; + MATDAT val; + ee_u32 i=0,j=0; + if (seed==0) + seed=1; + while (jA=A; + p->B=B; + p->C=(MATRES *)align_mem(B+N*N); + p->N=N; +#if CORE_DEBUG + printmat(A,N,"A"); + printmat(B,N,"B"); +#endif + return N; +} + +/* Function: matrix_sum + Calculate a function that depends on the values of elements in the matrix. + + For each element, accumulate into a temporary variable. + + As long as this value is under the parameter clipval, + add 1 to the result if the element is bigger then the previous. + + Otherwise, reset the accumulator and add 10 to the result. +*/ +ee_s16 matrix_sum(ee_u32 N, MATRES *C, MATDAT clipval) { + MATRES tmp=0,prev=0,cur=0; + ee_s16 ret=0; + ee_u32 i,j; + for (i=0; iclipval) { + ret+=10; + tmp=0; + } else { + ret += (cur>prev) ? 1 : 0; + } + prev=cur; + } + } + return ret; +} + +/* Function: matrix_mul_const + Multiply a matrix by a constant. + This could be used as a scaler for instance. +*/ +void matrix_mul_const(ee_u32 N, MATRES *C, MATDAT *A, MATDAT val) { + ee_u32 i,j; + for (i=0; i0) { + for(i=0;i>3) & 0x3]; + next=4; + break; + case 3: /* float */ + case 4: /* float */ + buf=floatpat[(seed>>3) & 0x3]; + next=8; + break; + case 5: /* scientific */ + case 6: /* scientific */ + buf=scipat[(seed>>3) & 0x3]; + next=8; + break; + case 7: /* invalid */ + buf=errpat[(seed>>3) & 0x3]; + next=8; + break; + default: /* Never happen, just to make some compilers happy */ + break; + } + } + size++; + while (total='0') & (c<='9')) ? 1 : 0; + return retval; +} + +/* Function: core_state_transition + Actual state machine. + + The state machine will continue scanning until either: + 1 - an invalid input is detcted. + 2 - a valid number has been detected. + + The input pointer is updated to point to the end of the token, and the end state is returned (either specific format determined or invalid). +*/ + +enum CORE_STATE core_state_transition( ee_u8 **instr , ee_u32 *transition_count) { + ee_u8 *str=*instr; + ee_u8 NEXT_SYMBOL; + enum CORE_STATE state=CORE_START; + for( ; *str && state != CORE_INVALID; str++ ) { + NEXT_SYMBOL = *str; + if (NEXT_SYMBOL==',') /* end of this input */ { + str++; + break; + } + switch(state) { + case CORE_START: + if(ee_isdigit(NEXT_SYMBOL)) { + state = CORE_INT; + } + else if( NEXT_SYMBOL == '+' || NEXT_SYMBOL == '-' ) { + state = CORE_S1; + } + else if( NEXT_SYMBOL == '.' ) { + state = CORE_FLOAT; + } + else { + state = CORE_INVALID; + transition_count[CORE_INVALID]++; + } + transition_count[CORE_START]++; + break; + case CORE_S1: + if(ee_isdigit(NEXT_SYMBOL)) { + state = CORE_INT; + transition_count[CORE_S1]++; + } + else if( NEXT_SYMBOL == '.' ) { + state = CORE_FLOAT; + transition_count[CORE_S1]++; + } + else { + state = CORE_INVALID; + transition_count[CORE_S1]++; + } + break; + case CORE_INT: + if( NEXT_SYMBOL == '.' ) { + state = CORE_FLOAT; + transition_count[CORE_INT]++; + } + else if(!ee_isdigit(NEXT_SYMBOL)) { + state = CORE_INVALID; + transition_count[CORE_INT]++; + } + break; + case CORE_FLOAT: + if( NEXT_SYMBOL == 'E' || NEXT_SYMBOL == 'e' ) { + state = CORE_S2; + transition_count[CORE_FLOAT]++; + } + else if(!ee_isdigit(NEXT_SYMBOL)) { + state = CORE_INVALID; + transition_count[CORE_FLOAT]++; + } + break; + case CORE_S2: + if( NEXT_SYMBOL == '+' || NEXT_SYMBOL == '-' ) { + state = CORE_EXPONENT; + transition_count[CORE_S2]++; + } + else { + state = CORE_INVALID; + transition_count[CORE_S2]++; + } + break; + case CORE_EXPONENT: + if(ee_isdigit(NEXT_SYMBOL)) { + state = CORE_SCIENTIFIC; + transition_count[CORE_EXPONENT]++; + } + else { + state = CORE_INVALID; + transition_count[CORE_EXPONENT]++; + } + break; + case CORE_SCIENTIFIC: + if(!ee_isdigit(NEXT_SYMBOL)) { + state = CORE_INVALID; + transition_count[CORE_INVALID]++; + } + break; + default: + break; + } + } + *instr=str; + return state; +} Index: soft/giet_vm/applications/coremark/core_util.c =================================================================== --- soft/giet_vm/applications/coremark/core_util.c (revision 753) +++ soft/giet_vm/applications/coremark/core_util.c (revision 753) @@ -0,0 +1,210 @@ +/* +Author : Shay Gal-On, EEMBC + +This file is part of EEMBC(R) and CoreMark(TM), which are Copyright (C) 2009 +All rights reserved. + +EEMBC CoreMark Software is a product of EEMBC and is provided under the terms of the +CoreMark License that is distributed with the official EEMBC COREMARK Software release. +If you received this EEMBC CoreMark Software without the accompanying CoreMark License, +you must discontinue use and download the official release from www.coremark.org. + +Also, if you are publicly displaying scores generated from the EEMBC CoreMark software, +make sure that you are in compliance with Run and Reporting rules specified in the accompanying readme.txt file. + +EEMBC +4354 Town Center Blvd. Suite 114-200 +El Dorado Hills, CA, 95762 +*/ +#include "coremark.h" +/* Function: get_seed + Get a values that cannot be determined at compile time. + + Since different embedded systems and compilers are used, 3 different methods are provided: + 1 - Using a volatile variable. This method is only valid if the compiler is forced to generate code that + reads the value of a volatile variable from memory at run time. + Please note, if using this method, you would need to modify core_portme.c to generate training profile. + 2 - Command line arguments. This is the preferred method if command line arguments are supported. + 3 - System function. If none of the first 2 methods is available on the platform, + a system function which is not a stub can be used. + + e.g. read the value on GPIO pins connected to switches, or invoke special simulator functions. +*/ +#if (SEED_METHOD==SEED_VOLATILE) + extern volatile ee_s32 seed1_volatile; + extern volatile ee_s32 seed2_volatile; + extern volatile ee_s32 seed3_volatile; + extern volatile ee_s32 seed4_volatile; + extern volatile ee_s32 seed5_volatile; + ee_s32 get_seed_32(int i) { + ee_s32 retval; + switch (i) { + case 1: + retval=seed1_volatile; + break; + case 2: + retval=seed2_volatile; + break; + case 3: + retval=seed3_volatile; + break; + case 4: + retval=seed4_volatile; + break; + case 5: + retval=seed5_volatile; + break; + default: + retval=0; + break; + } + return retval; + } +#elif (SEED_METHOD==SEED_ARG) +ee_s32 parseval(char *valstring) { + ee_s32 retval=0; + ee_s32 neg=1; + int hexmode=0; + if (*valstring == '-') { + neg=-1; + valstring++; + } + if ((valstring[0] == '0') && (valstring[1] == 'x')) { + hexmode=1; + valstring+=2; + } + /* first look for digits */ + if (hexmode) { + while (((*valstring >= '0') && (*valstring <= '9')) || ((*valstring >= 'a') && (*valstring <= 'f'))) { + ee_s32 digit=*valstring-'0'; + if (digit>9) + digit=10+*valstring-'a'; + retval*=16; + retval+=digit; + valstring++; + } + } else { + while ((*valstring >= '0') && (*valstring <= '9')) { + ee_s32 digit=*valstring-'0'; + retval*=10; + retval+=digit; + valstring++; + } + } + /* now add qualifiers */ + if (*valstring=='K') + retval*=1024; + if (*valstring=='M') + retval*=1024*1024; + + retval*=neg; + return retval; +} + +ee_s32 get_seed_args(int i, int argc, char *argv[]) { + if (argc>i) + return parseval(argv[i]); + return 0; +} + +#elif (SEED_METHOD==SEED_FUNC) +/* If using OS based function, you must define and implement the functions below in core_portme.h and core_portme.c ! */ +ee_s32 get_seed_32(int i) { + ee_s32 retval; + switch (i) { + case 1: + retval=portme_sys1(); + break; + case 2: + retval=portme_sys2(); + break; + case 3: + retval=portme_sys3(); + break; + case 4: + retval=portme_sys4(); + break; + case 5: + retval=portme_sys5(); + break; + default: + retval=0; + break; + } + return retval; +} +#endif + +/* Function: crc* + Service functions to calculate 16b CRC code. + +*/ +ee_u16 crcu8(ee_u8 data, ee_u16 crc ) +{ + ee_u8 i=0,x16=0,carry=0; + + for (i = 0; i < 8; i++) + { + x16 = (ee_u8)((data & 1) ^ ((ee_u8)crc & 1)); + data >>= 1; + + if (x16 == 1) + { + crc ^= 0x4002; + carry = 1; + } + else + carry = 0; + crc >>= 1; + if (carry) + crc |= 0x8000; + else + crc &= 0x7fff; + } + return crc; +} +ee_u16 crcu16(ee_u16 newval, ee_u16 crc) { + crc=crcu8( (ee_u8) (newval) ,crc); + crc=crcu8( (ee_u8) ((newval)>>8) ,crc); + return crc; +} +ee_u16 crcu32(ee_u32 newval, ee_u16 crc) { + crc=crc16((ee_s16) newval ,crc); + crc=crc16((ee_s16) (newval>>16) ,crc); + return crc; +} +ee_u16 crc16(ee_s16 newval, ee_u16 crc) { + return crcu16((ee_u16)newval, crc); +} + +ee_u8 check_data_types() { + ee_u8 retval=0; + if (sizeof(ee_u8) != 1) { + ee_printf("ERROR: ee_u8 is not an 8b datatype!\n"); + retval++; + } + if (sizeof(ee_u16) != 2) { + ee_printf("ERROR: ee_u16 is not a 16b datatype!\n"); + retval++; + } + if (sizeof(ee_s16) != 2) { + ee_printf("ERROR: ee_s16 is not a 16b datatype!\n"); + retval++; + } + if (sizeof(ee_s32) != 4) { + ee_printf("ERROR: ee_s32 is not a 32b datatype!\n"); + retval++; + } + if (sizeof(ee_u32) != 4) { + ee_printf("ERROR: ee_u32 is not a 32b datatype!\n"); + retval++; + } + if (sizeof(ee_ptr_int) != sizeof(int *)) { + ee_printf("ERROR: ee_ptr_int is not a datatype that holds an int pointer!\n"); + retval++; + } + if (retval>0) { + ee_printf("ERROR: Please modify the datatypes in core_portme.h!\n"); + } + return retval; +} Index: soft/giet_vm/applications/coremark/coremark.h =================================================================== --- soft/giet_vm/applications/coremark/coremark.h (revision 753) +++ soft/giet_vm/applications/coremark/coremark.h (revision 753) @@ -0,0 +1,174 @@ +/* +Author : Shay Gal-On, EEMBC + +This file is part of EEMBC(R) and CoreMark(TM), which are Copyright (C) 2009 +All rights reserved. + +EEMBC CoreMark Software is a product of EEMBC and is provided under the terms of the +CoreMark License that is distributed with the official EEMBC COREMARK Software release. +If you received this EEMBC CoreMark Software without the accompanying CoreMark License, +you must discontinue use and download the official release from www.coremark.org. + +Also, if you are publicly displaying scores generated from the EEMBC CoreMark software, +make sure that you are in compliance with Run and Reporting rules specified in the accompanying readme.txt file. + +EEMBC +4354 Town Center Blvd. Suite 114-200 +El Dorado Hills, CA, 95762 +*/ +/* Topic: Description + This file contains declarations of the various benchmark functions. +*/ + +/* Configuration: TOTAL_DATA_SIZE + Define total size for data algorithms will operate on +*/ +#ifndef TOTAL_DATA_SIZE +#define TOTAL_DATA_SIZE 2*1000 +#endif + +#define SEED_ARG 0 +#define SEED_FUNC 1 +#define SEED_VOLATILE 2 + +#define MEM_STATIC 0 +#define MEM_MALLOC 1 +#define MEM_STACK 2 + +#include "core_portme.h" + +#if HAS_STDIO +#include +#endif +#if HAS_PRINTF +#define ee_printf printf +#endif + +/* Actual benchmark execution in iterate */ +void *iterate(void *pres); + +/* Typedef: secs_ret + For machines that have floating point support, get number of seconds as a double. + Otherwise an unsigned int. +*/ +#if HAS_FLOAT +typedef double secs_ret; +#else +typedef ee_u32 secs_ret; +#endif + +#if MAIN_HAS_NORETURN +#define MAIN_RETURN_VAL +#define MAIN_RETURN_TYPE void +#else +#define MAIN_RETURN_VAL 0 +#define MAIN_RETURN_TYPE int +#endif + +void start_time(void); +void stop_time(void); +CORE_TICKS get_time(void); +secs_ret time_in_secs(CORE_TICKS ticks); + +/* Misc useful functions */ +ee_u16 crcu8(ee_u8 data, ee_u16 crc); +ee_u16 crc16(ee_s16 newval, ee_u16 crc); +ee_u16 crcu16(ee_u16 newval, ee_u16 crc); +ee_u16 crcu32(ee_u32 newval, ee_u16 crc); +ee_u8 check_data_types(); +void *portable_malloc(ee_size_t size); +void portable_free(void *p); +ee_s32 parseval(char *valstring); + +/* Algorithm IDS */ +#define ID_LIST (1<<0) +#define ID_MATRIX (1<<1) +#define ID_STATE (1<<2) +#define ALL_ALGORITHMS_MASK (ID_LIST|ID_MATRIX|ID_STATE) +#define NUM_ALGORITHMS 3 + +/* list data structures */ +typedef struct list_data_s { + ee_s16 data16; + ee_s16 idx; +} list_data; + +typedef struct list_head_s { + struct list_head_s *next; + struct list_data_s *info; +} list_head; + + +/*matrix benchmark related stuff */ +#define MATDAT_INT 1 +#if MATDAT_INT +typedef ee_s16 MATDAT; +typedef ee_s32 MATRES; +#else +typedef ee_f16 MATDAT; +typedef ee_f32 MATRES; +#endif + +typedef struct MAT_PARAMS_S { + int N; + MATDAT *A; + MATDAT *B; + MATRES *C; +} mat_params; + +/* state machine related stuff */ +/* List of all the possible states for the FSM */ +typedef enum CORE_STATE { + CORE_START=0, + CORE_INVALID, + CORE_S1, + CORE_S2, + CORE_INT, + CORE_FLOAT, + CORE_EXPONENT, + CORE_SCIENTIFIC, + NUM_CORE_STATES +} core_state_e ; + + +/* Helper structure to hold results */ +typedef struct RESULTS_S { + /* inputs */ + ee_s16 seed1; /* Initializing seed */ + ee_s16 seed2; /* Initializing seed */ + ee_s16 seed3; /* Initializing seed */ + void *memblock[4]; /* Pointer to safe memory location */ + ee_u32 size; /* Size of the data */ + ee_u32 iterations; /* Number of iterations to execute */ + ee_u32 execs; /* Bitmask of operations to execute */ + struct list_head_s *list; + mat_params mat; + /* outputs */ + ee_u16 crc; + ee_u16 crclist; + ee_u16 crcmatrix; + ee_u16 crcstate; + ee_s16 err; + /* ultithread specific */ + core_portable port; +} core_results; + +/* Multicore execution handling */ +#if (MULTITHREAD>1) +ee_u8 core_start_parallel(core_results *res); +ee_u8 core_stop_parallel(core_results *res); +#endif + +/* list benchmark functions */ +list_head *core_list_init(ee_u32 blksize, list_head *memblock, ee_s16 seed); +ee_u16 core_bench_list(core_results *res, ee_s16 finder_idx); + +/* state benchmark functions */ +void core_init_state(ee_u32 size, ee_s16 seed, ee_u8 *p); +ee_u16 core_bench_state(ee_u32 blksize, ee_u8 *memblock, + ee_s16 seed1, ee_s16 seed2, ee_s16 step, ee_u16 crc); + +/* matrix benchmark functions */ +ee_u32 core_init_matrix(ee_u32 blksize, void *memblk, ee_s32 seed, mat_params *p); +ee_u16 core_bench_matrix(mat_params *p, ee_s16 seed, ee_u16 crc); + Index: soft/giet_vm/applications/coremark/coremark.ld =================================================================== --- soft/giet_vm/applications/coremark/coremark.ld (revision 753) +++ soft/giet_vm/applications/coremark/coremark.ld (revision 753) @@ -0,0 +1,36 @@ +/**************************************************************************** +* Definition of the base address for code and data vsegs +*****************************************************************************/ + +seg_code_base = 0x10000000; +seg_data_base = 0x20000000; + +/*************************************************************************** +* Grouping sections into segments for code and data +***************************************************************************/ + +SECTIONS +{ + . = seg_code_base; + seg_code : + { + *(.text) + } + + . = seg_data_base; + seg_data : + { + *(.ctors) + *(.rodata) + *(.rodata.*) + *(.data) + *(.lit8) + *(.lit4) + *(.sdata) + *(.bss) + *(COMMON) + *(.sbss) + *(.scommon) + } +} + Index: soft/giet_vm/applications/coremark/coremark.md5 =================================================================== --- soft/giet_vm/applications/coremark/coremark.md5 (revision 753) +++ soft/giet_vm/applications/coremark/coremark.md5 (revision 753) @@ -0,0 +1,6 @@ +ef848e21d3268820c4830c8332708c1e *core_list_join.c +7631bd852cf622ee2c3f58d30d5a6a92 *core_main.c +aba83b72d1179503e17d182cf809a2f1 *core_matrix.c +e494356cd3a4adc07f12d330b25a2d72 *core_state.c +f37e364b4f4123722066218707cc476e *core_util.c +e8eac57239af0eedf934f9e8b62d8ec6 *coremark.h Index: soft/giet_vm/applications/coremark/coremark.py =================================================================== --- soft/giet_vm/applications/coremark/coremark.py (revision 753) +++ soft/giet_vm/applications/coremark/coremark.py (revision 753) @@ -0,0 +1,125 @@ +#!/usr/bin/env python + +from mapping import * + +################################################################################### +# file : coremark.py +# date : january 2016 +# author : Cesar FUGUET TORTOLERO +################################################################################### + +###################### +def extend( mapping ): + + x_size = mapping.x_size + y_size = mapping.y_size + nprocs = mapping.nprocs + x_width = mapping.x_width + y_width = mapping.y_width + + # define vsegs base & size + code_base = 0x10000000 + code_size = 0x00010000 # 64 Kbytes (per cluster) + + data_base = 0x20000000 + data_size = 0x00010000 # 64 Kbytes (non replicated) + + heap_base = 0x30000000 + heap_size = 0x00200000 # 2M bytes (per cluster) + + stack_base = 0x40000000 + stack_size = 0x00010000 # 64 Kbytes (per thread) + + # create vspace + vspace = mapping.addVspace( name = 'coremark', + startname = 'coremark_data', + active = True ) + + # data vseg : shared / cluster[0][0] + mapping.addVseg( vspace, 'coremark_data', data_base , data_size, + 'C_WU', vtype = 'ELF', x = 0, y = 0, pseg = 'RAM', + binpath = 'bin/coremark/appli.elf', + local = False ) + + # heap vsegs : shared (one per cluster) + for x in xrange (x_size): + for y in xrange (y_size): + cluster_id = (x * y_size) + y + if ( mapping.clusters[cluster_id].procs ): + size = heap_size + base = heap_base + (cluster_id * size) + mapping.addVseg( vspace, 'coremark_heap_%d_%d' %(x,y), base , size, + 'C_WU', vtype = 'HEAP', x = x, y = y, pseg = 'RAM', + local = False, big = True ) + + # code vsegs : local (one copy per cluster) + for x in xrange (x_size): + for y in xrange (y_size): + cluster_id = (x * y_size) + y + if ( mapping.clusters[cluster_id].procs ): + mapping.addVseg( vspace, 'coremark_code_%d_%d' %(x,y), + code_base , code_size, + 'CXWU', vtype = 'ELF', x = x, y = y, pseg = 'RAM', + binpath = 'bin/coremark/appli.elf', + local = True ) + + # stacks vsegs: local (one stack per thread => nprocs stacks per cluster) + # Define also the main_stack in cluster[0][0]. The main_stack cannot be + # local as the main thread passes to workers a local parameter. + mapping.addVseg( vspace, 'main_stack', + stack_base, stack_size, 'C_WU', vtype = 'BUFFER', + x = 0 , y = 0 , pseg = 'RAM', + local = False ) + + for x in xrange (x_size): + for y in xrange (y_size): + cluster_id = (x * y_size) + y + if ( mapping.clusters[cluster_id].procs ): + for p in xrange( nprocs ): + proc_id = (((x * y_size) + y) * nprocs) + p + base = stack_base + (proc_id * stack_size) + stack_size + mapping.addVseg( vspace, 'coremark_stack_%d_%d_%d' % (x,y,p), + base, stack_size, 'C_WU', vtype = 'BUFFER', + x = x , y = y , pseg = 'RAM', + local = True ) + + # distributed threads / one thread per processor + # ... plus main on P[0][0][0] + IS_MAIN = True + mapping.addThread( vspace, 'main', IS_MAIN, 0, 0, 0, + 'main_stack', + 'coremark_heap_0_0', + 1 ) # index in start_vector + + for x in xrange (x_size): + for y in xrange (y_size): + cluster_id = (x * y_size) + y + if ( mapping.clusters[cluster_id].procs ): + for p in xrange( nprocs ): + thread_name = 'iterate_%d_%d_%d' % (x,y,p) + mapping.addThread( vspace, thread_name, not IS_MAIN, x, y, p, + 'coremark_stack_%d_%d_%d' % (x,y,p), + 'coremark_heap_%d_%d' % (x,y), + 0 ) + + # extend mapping name + mapping.name += '_coremark' + + return vspace # useful for test + +################################ test ############################################ + +if __name__ == '__main__': + vspace = extend( Mapping( 'test', 2, 2, 4 ) ) + print vspace.xml() + + +# Local Variables: +# tab-width: 4; +# c-basic-offset: 4; +# c-file-offsets:((innamespace . 0)(inline-open . 0)); +# indent-tabs-mode: nil; +# End: +# +# vim: filetype=python:expandtab:shiftwidth=4:tabstop=4:softtabstop=4 + Index: soft/giet_vm/applications/coremark/mach/core_portme.c =================================================================== --- soft/giet_vm/applications/coremark/mach/core_portme.c (revision 753) +++ soft/giet_vm/applications/coremark/mach/core_portme.c (revision 753) @@ -0,0 +1,228 @@ +/* + File : core_portme.c +*/ +/* + Author : Shay Gal-On, EEMBC + Legal : TODO! +*/ +#include "coremark.h" + +#include +#include + +#if VALIDATION_RUN +volatile ee_s32 seed1_volatile=0x3415; +volatile ee_s32 seed2_volatile=0x3415; +volatile ee_s32 seed3_volatile=0x66; +#endif + +#if PERFORMANCE_RUN +volatile ee_s32 seed1_volatile=0x0; +volatile ee_s32 seed2_volatile=0x0; +volatile ee_s32 seed3_volatile=0x66; +#endif + +#if PROFILE_RUN +volatile ee_s32 seed1_volatile=0x8; +volatile ee_s32 seed2_volatile=0x8; +volatile ee_s32 seed3_volatile=0x8; +#endif + +/* Define : seed4_volatile + It defines the number of iterations. + + When set to 0, the number of iterations is automatically computed + during the benchmark's execution +*/ +volatile ee_s32 seed4_volatile=0; + + +/* Define : seed5_volatile + It defines which algorithms are executed + + When set to 0, all algorithms are executed (matrix, list_join, state) +*/ +volatile ee_s32 seed5_volatile=0; + + +/* Porting : Timing functions + How to capture time and convert to seconds must be ported to whatever is supported by the platform. + + Define : TIMER_RES_DIVIDER + Divider to trade off timer resolution and total time that can be measured. + + Use lower values to increase resolution, but make sure that overflow does not occur. + If there are issues with the return value overflowing, increase this value. +*/ +#define TIMER_RES_DIVIDER 1 +#define NSECS_PER_SEC 1000000 /* 1 GHz */ +#define EE_TICKS_PER_SEC (NSECS_PER_SEC / TIMER_RES_DIVIDER) +#define GETMYTIME(_t) (*_t=giet_proctime()) + + +/* global time variables.*/ +static CORE_TICKS start_time_val, stop_time_val; + + +/* Function : start_time + This function will be called right before starting the timed portion of the benchmark. + + Implementation may be capturing a system timer (as implemented in the example code) + or zeroing some system parameters - e.g. setting the cpu clocks cycles to 0. +*/ +void start_time(void) { + GETMYTIME(&start_time_val ); +} + + +/* Function : stop_time + This function will be called right after ending the timed portion of the benchmark. + + Implementation may be capturing a system timer (as implemented in the example code) + or other system parameters - e.g. reading the current value of cpu cycles counter. +*/ +void stop_time(void) { + GETMYTIME(&stop_time_val ); +} + + +/* Function : get_time + Return an abstract "ticks" number that signifies time on the system. + + Actual value returned may be cpu cycles, milliseconds or any other value, + as long as it can be converted to seconds by . + This methodology is taken to accomodate any hardware or simulated platform. + The sample implementation returns millisecs by default, + and the resolution is controlled by +*/ +CORE_TICKS get_time(void) { + CORE_TICKS elapsed=(CORE_TICKS)(stop_time_val - start_time_val); + return elapsed; +} + + +/* Function : time_in_secs + Convert the value returned by get_time to seconds. + + The type is used to accomodate systems with no support for floating point. + Default implementation implemented by the EE_TICKS_PER_SEC macro above. +*/ +secs_ret time_in_secs(CORE_TICKS ticks) { + secs_ret retval=((secs_ret)ticks) / (secs_ret)EE_TICKS_PER_SEC; + return retval; +} + +ee_u32 default_num_contexts=1; + + +/* Function : portable_init + Target specific initialization code + Test for some common mistakes. +*/ +void portable_init(core_portable *p, int *argc, char *argv[]) +{ + if (sizeof(ee_ptr_int) != sizeof(ee_u8 *)) { + ee_printf("ERROR! Please define ee_ptr_int to a type that holds a pointer!\n"); + } + if (sizeof(ee_u32) != 4) { + ee_printf("ERROR! Please define ee_u32 to a 32b unsigned type!\n"); + } + + // allocate a shared TTY + giet_tty_alloc(1); + +#if MULTITHREAD>1 + unsigned int x_size, y_size, nprocs; + giet_procs_number( &x_size, &y_size, &nprocs ); + + // one context per core when using multiple threads + default_num_contexts = x_size * y_size * nprocs; + + // initialize distributed heaps + int x, y; + for (x=0; x < x_size; ++x) { + for (y=0; y < y_size; ++y) { + heap_init(x, y); + } + } + + // FIXME: init a lock to access the TTY, and create a printf wrapper, + // requiring and releasing the lock +#else + // initialize local heap + unsigned int lx, ly, lp; + giet_proc_xyp(&lx, &ly, &lp); + heap_init(lx, ly); +#endif + + p->portable_id=1; +} + + +/* Function : portable_fini + Target specific final code +*/ +void portable_fini(core_portable *p) +{ + p->portable_id=0; +} + + +/* Function : __iterate + Target specific wrapper for the iterate function, which allows to add + the iterate function in the constructor list + + This is the entry function for the secondary threads +*/ +__attribute__ ((constructor)) +void *__iterate(void *pres) { + iterate(pres); + giet_pthread_exit(NULL); + return NULL; +} + + +/* Function : __main + Target specific wrapper for the main function, which allows to add + the main function in the constructor list. + + This is the entry function for the main thread +*/ +extern MAIN_RETURN_TYPE main(void); +__attribute__ ((constructor)) +void __main() { + main(); + giet_pthread_exit("Coremark execution completed"); +} + + +/* Function: portable_malloc + Provide malloc() functionality in a platform specific way. +*/ +void *portable_malloc(ee_size_t size) { + return malloc(size); +} + + +/* Function: portable_free + Provide free() functionality in a platform specific way. +*/ +void portable_free(void *p) { + free(p); +} + + +/* Function: core_start_parallel + Start benchmarking in a parallel context. +*/ +ee_u8 core_start_parallel(core_results *res) { + return (ee_u8)giet_pthread_create(&(res->port.thread),NULL,__iterate,(void*)res); +} + + +/* Function: core_stop_parallel + Stop a parallel context execution of coremark, and gather the results. +*/ +ee_u8 core_stop_parallel(core_results *res) { + return (ee_u8)giet_pthread_join(res->port.thread, NULL); +} Index: soft/giet_vm/applications/coremark/mach/core_portme.h =================================================================== --- soft/giet_vm/applications/coremark/mach/core_portme.h (revision 753) +++ soft/giet_vm/applications/coremark/mach/core_portme.h (revision 753) @@ -0,0 +1,201 @@ +/* File : core_portme.h */ + +/* + Author : Shay Gal-On, EEMBC + Legal : TODO! +*/ +/* Topic : Description + This file contains configuration constants required to execute on different platforms +*/ +#ifndef CORE_PORTME_H +#define CORE_PORTME_H + +/************************/ +/* Data types and settings */ +/************************/ + +/* Configuration : HAS_FLOAT + Define to 1 if the platform supports floating point. +*/ +#ifndef HAS_FLOAT +#define HAS_FLOAT 0 +#endif + + +/* Configuration : HAS_TIME_H + Define to 1 if platform has the time.h header file, + and implementation of functions thereof. +*/ +#ifndef HAS_TIME_H +#define HAS_TIME_H 0 +#endif + + +/* Configuration : USE_CLOCK + Define to 1 if platform has the time.h header file, + and implementation of functions thereof. +*/ +#ifndef USE_CLOCK +#define USE_CLOCK 0 +#endif + + +/* Configuration : HAS_STDIO + Define to 1 if the platform has stdio.h. +*/ +#ifndef HAS_STDIO +#define HAS_STDIO 1 +#include +#endif + + +/* Configuration : HAS_PRINTF + Define to 1 if the platform has stdio.h and implements the printf function. +*/ +#ifndef HAS_PRINTF +#define HAS_PRINTF 1 + +/* printf wrapper macro */ +#define printf(...) giet_tty_printf(__VA_ARGS__) +#endif + + +/* Configuration : CORE_TICKS + Define type of return from the timing functions. + */ +typedef unsigned int CORE_TICKS; + + +/* Definitions : COMPILER_VERSION, COMPILER_FLAGS, MEM_LOCATION + Initialize these strings per platform +*/ +#ifndef COMPILER_VERSION + #ifdef __GNUC__ + #define COMPILER_VERSION "GCC"__VERSION__ + #else + #define COMPILER_VERSION "Please put compiler version here (e.g. gcc 4.1)" + #endif +#endif +#ifndef COMPILER_FLAGS + #define COMPILER_FLAGS FLAGS_STR /* "Please put compiler flags here (e.g. -o3)" */ +#endif +#ifndef MEM_LOCATION + #define MEM_LOCATION "Data on heap" +#endif + + +/* Data Types : + To avoid compiler issues, define the data types that need ot be used for 8b, 16b and 32b in . + + *Important* : + ee_ptr_int needs to be the data type used to hold pointers, otherwise coremark may fail!!! +*/ +typedef signed short ee_s16; +typedef unsigned short ee_u16; +typedef signed int ee_s32; +typedef unsigned char ee_u8; +typedef unsigned int ee_u32; +typedef ee_u32 ee_ptr_int; +typedef ee_u32 ee_size_t; + +#if HAS_FLOAT +typedef double ee_f32; +#endif + + +/* align_mem : + This macro is used to align an offset to point to a 32b value. It is used in the Matrix algorithm to initialize the input memory blocks. +*/ +#define align_mem(x) (void *)(4 + (((ee_ptr_int)(x) - 1) & ~3)) + + +/* Configuration : SEED_METHOD + Defines method to get seed values that cannot be computed at compile time. + + Valid values : + SEED_ARG - from command line. + SEED_FUNC - from a system function. + SEED_VOLATILE - from volatile variables. +*/ +#ifndef SEED_METHOD +#define SEED_METHOD SEED_VOLATILE +#endif + + +/* Configuration : MEM_METHOD + Defines method to get a block of memry. + + Valid values : + MEM_MALLOC - for platforms that implement malloc and have malloc.h. + MEM_STATIC - to use a static memory array. + MEM_STACK - to allocate the data block on the stack (NYI). +*/ +#ifndef MEM_METHOD +#define MEM_METHOD MEM_MALLOC +#endif + + +/* Configuration : MULTITHREAD + Define for parallel execution + + Valid values : + 1 - only one context (default). + N>1 - will execute N copies in parallel. + + Note : + If this flag is defined to more then 1, an implementation for launching parallel contexts must be defined. + + It is valid to have a different implementation of and in , + to fit a particular architecture. +*/ +#ifndef MULTITHREAD +#define MULTITHREAD 1 +#endif + +#define PARALLEL_METHOD "pthread" + +/* Configuration : MAIN_HAS_NOARGC + Needed if platform does not support getting arguments to main. + + Valid values : + 0 - argc/argv to main is supported + 1 - argc/argv to main is not supported + + Note : + This flag only matters if MULTITHREAD has been defined to a value greater then 1. +*/ +#ifndef MAIN_HAS_NOARGC +#define MAIN_HAS_NOARGC 1 +#endif + + +/* Configuration : MAIN_HAS_NORETURN + Needed if platform does not support returning a value from main. + + Valid values : + 0 - main returns an int, and return value will be 0. + 1 - platform does not support returning a value from main +*/ +#ifndef MAIN_HAS_NORETURN +#define MAIN_HAS_NORETURN 0 +#endif + + +/* Variable : default_num_contexts + Not used for this simple port, must cintain the value 1. +*/ +extern ee_u32 default_num_contexts; + +typedef struct CORE_PORTABLE_S { + ee_u8 portable_id; + pthread_t thread; +} core_portable; + + +/* target specific init/fini */ +void portable_init(core_portable *p, int *argc, char *argv[]); +void portable_fini(core_portable *p); +void *portable_malloc(ee_size_t size); +void portable_free(void *p); + +#endif /* CORE_PORTME_H */ Index: soft/giet_vm/applications/coremark/readme.txt =================================================================== --- soft/giet_vm/applications/coremark/readme.txt (revision 753) +++ soft/giet_vm/applications/coremark/readme.txt (revision 753) @@ -0,0 +1,227 @@ +File: CoreMark + +Topic: Welcome +Copyright © 2009 EEMBC All rights reserved. +CoreMark is a trademark of EEMBC and EEMBC is a registered trademark of the Embedded Microprocessor Benchmark Consortium. + +CoreMark’s primary goals are simplicity and providing a method for testing only a processor’s core features. + +For more information about EEMBC's comprehensive embedded benchmark suites, please see www.eembc.org. + +Topic: Building and running + Download the release files from the www.coremark.org. + You can verify the download using the coremark_.md5 file + > md5sum -c coremark_.md5 + + Unpack the distribution (tar -vzxf coremark_.tgz && tar -vzxf coremark__docs.tgz) + then change to the coremark_ folder. + + To build and run the benchmark, type + > make + Full results are available in the files run1.log and run2.log. + CoreMark result can be found in run1.log. + + For self hosted Linux or Cygwin platforms, a simple make should work. + + Cross Compile: + For cross compile platforms please adjust , (and possibly ) + according to the specific platform used. + When porting to a new platform, it is recommended to copy one of the default port folders + (e.g. mkdir && cp linux/* ), adjust the porting files, and run + > make PORT_DIR= + + Systems without make: + The following files need to be compiled: + - + - + - + - + - + - / + + For example + > gcc -O2 -o coremark.exe core_list_join.c core_main.c core_matrix.c core_state.c core_util.c simple/core_portme.c -DPERFORMANCE_RUN=1 -DITERATIONS=1000 + > ./coremark.exe > run1.log + The above will compile the benchmark for a performance run and 1000 iterations. Output is redirected to run1.log. + + Make targets: + run - Default target, creates run1.log and run2.log. + run1.log - Run the benchmark with performance parameters, and output to run1.log + run2.log - Run the benchmark with validation parameters, and output to run2.log + run3.log - Run the benchmark with profile generation parameters, and output to run3.log + compile - compile the benchmark executable + link - link the benchmark executable + check - test MD5 of sources that may not be modified + clean - clean temporary files + + ITERATIONS: + By default, the benchmark will run between 10-100 seconds. + To override, use ITERATIONS=N + > make ITERATIONS=10 + Will run the benchmark for 10 iterations. + It is recommended to set a specific number of iterations in certain situations e.g.: + - Running with a simulator + - Measuring power/energy + - Timing cannot be restarted + + Minimum required run time: + Results are only valid for reporting if the benchmark ran for at least 10 secs! + + XCFLAGS: + To add compiler flags from the command line, use XCFLAGS e.g. + > make XCFLAGS="-g -DMULTITHREAD=4 -DUSE_FORK=1" + + o CORE_DEBUG + + Define to compile for a debug run if you get incorrect CRC. + > make XCFLAGS="-DCORE_DEBUG=1" + + o Parallel Execution + + Use XCFLAGS=-DMULTITHREAD=N where N is number of threads to run in parallel. + Several implementations are available to execute in multiple contexts, + or you can implement your own in . + > make XCFLAGS="-DMULTITHREAD=4 -DUSE_PTHREAD" + Above will compile the benchmark for execution on 4 cores, using POSIX Threads API. + + REBUILD: + To force rebuild, add the flag REBUILD to the command line + > make REBUILD=1 + + Check core_portme.mak for more important options. + + Run parameters for the benchmark executable: + Coremark executable takes several parameters as follows (if main accepts arguments). + 1st - A seed value used for initialization of data. + 2nd - A seed value used for initialization of data. + 3rd - A seed value used for initialization of data. + 4th - Number of iterations (0 for auto : default value) + 5th - Reserved for internal use. + 6th - Reserved for internal use. + 7th - For malloc users only, ovreride the size of the input data buffer. + + The run target from make will run coremark with 2 different data initialization seeds. + + Alternative parameters: + If not using malloc or command line arguments are not supported, the buffer size + for the algorithms must be defined via the compiler define TOTAL_DATA_SIZE. + TOTAL_DATA_SIZE must be set to 2000 bytes (default) for standard runs. + The default for such a target when testing different configurations could be ... + > make XCFLAGS="-DTOTAL_DATA_SIZE=6000 -DMAIN_HAS_NOARGC=1" + +Topic: Documentation + When you unpack the documentation (tar -vzxf coremark__docs.tgz) a docs folder will be created. + Check the file docs/html/index.html and the website http://www.coremark.org for more info. + +Topic: Submitting results + CoreMark results can be submitted on the web. + + Open a web browser and go to http://www.coremark.org/benchmark/index.php?pg=benchmark + Select the link to add a new score and follow the instructions. + +Topic: Run rules + What is and is not allowed. + + Required: + 1 - The benchmark needs to run for at least 10 seconds. + 2 - All validation must succeed for seeds 0,0,0x66 and 0x3415,0x3415,0x66, + buffer size of 2000 bytes total. + o If not using command line arguments to main: + > make XCFLAGS="-DPERFORMANCE_RUN=1" REBUILD=1 run1.log + > make XCFLAGS="-DVALIDATION_RUN=1" REBUILD=1 run2.log + 3 - If using profile guided optimization, profile must be generated using seeds of 8,8,8, + and buffer size of 1200 bytes total. + > make XCFLAGS="-DTOTAL_DATA_SIZE=1200 -DPROFILE_RUN=1" REBUILD=1 run3.log + 4 - All source files must be compiled with the same flags. + 5 - All data type sizes must match size in bits such that: + o ee_u8 is an 8 bits datatype. + o ee_s16 is an 16 bits datatype. + o ee_u16 is an 16 bits datatype. + o ee_s32 is an 32 bits datatype. + o ee_u32 is an 32 bits datatype. + + Allowed: + - Changing number of iterations + - Changing toolchain and build/load/run options + - Changing method of acquiring a data memory block + - Changing the method of acquiring seed values + - Changing implementation in core_portme.c + - Changing configuration values in core_portme.h + - Changing core_portme.mak + + Not allowed: + - Changing of source file other then core_portme* (use make check to validate) + +Topic: Reporting rules + How to report results on a data sheet? + + CoreMark 1.0 : N / C [/ P] [/ M] + + N - Number of iterations per second with seeds 0,0,0x66,size=2000) + C - Compiler version and flags + P - Parameters such as data and code allocation specifics + - This parameter *may* be omitted if all data was allocated on the heap in RAM. + - This parameter *may not* be omitted when reporting CoreMark/MHz + M - Type of parallel execution (if used) and number of contexts + This parameter may be omitted if parallel execution was not used. + + e.g. + > CoreMark 1.0 : 128 / GCC 4.1.2 -O2 -fprofile-use / Heap in TCRAM / FORK:2 + or + > CoreMark 1.0 : 1400 / GCC 3.4 -O4 + + If reporting scaling results, the results must be reported as follows: + + CoreMark/MHz 1.0 : N / C / P [/ M] + + P - When reporting scaling results, memory parameter must also indicate memory frequency:core frequency ratio. + - If the core has cache and cache frequency to core frequency ratio is configurable, that must also be included. + + e.g. + > CoreMark/MHz 1.0 : 1.47 / GCC 4.1.2 -O2 / DDR3(Heap) 30:1 Memory 1:1 Cache + + +Topic: Log File Format + The log files have the following format +(start example) +2K performance run parameters for coremark. (Run type) +CoreMark Size : 666 (Buffer size) +Total ticks : 25875 (platform dependent value) +Total time (secs) : 25.875000 (actual time in seconds) +Iterations/Sec : 3864.734300 (Performance value to report) +Iterations : 100000 (number of iterations used) +Compiler version : GCC3.4.4 (Compiler and version) +Compiler flags : -O2 (Compiler and linker flags) +Memory location : Code in flash, data in on chip RAM +seedcrc : 0xe9f5 (identifier for the input seeds) +[0]crclist : 0xe714 (validation for list part) +[0]crcmatrix : 0x1fd7 (validation for matrix part) +[0]crcstate : 0x8e3a (validation for state part) +[0]crcfinal : 0x33ff (iteration dependent output) +Correct operation validated. See readme.txt for run and reporting rules. (*Only when run is successful*) +CoreMark 1.0 : 6508.490622 / GCC3.4.4 -O2 / Heap (*Only on a successful performance run*) +(end example) + +Topic: Legal +See LICENSE.txt or the word document file under docs/LICENSE.doc. +For more information on your legal rights to use this benchmark, please see +http://www.coremark.org/download/register.php?pg=register + +Topic: Credits +Many thanks to all of the individuals who helped with the development or testing of CoreMark including (Sorted by company name) +o Alan Anderson, ADI +o Adhikary Rajiv, ADI +o Elena Stohr, ARM +o Ian Rickards, ARM +o Andrew Pickard, ARM +o Trent Parker, CAVIUM +o Shay Gal-On, EEMBC +o Markus Levy, EEMBC +o Ron Olson, IBM +o Eyal Barzilay, MIPS +o Jens Eltze, NEC +o Hirohiko Ono, NEC +o Ulrich Drees, NEC +o Frank Roscheda, NEC +o Rob Cosaro, NXP +o Shumpei Kawasaki, RENESAS Index: soft/giet_vm/giet_kernel/sys_handler.c =================================================================== --- soft/giet_vm/giet_kernel/sys_handler.c (revision 752) +++ soft/giet_vm/giet_kernel/sys_handler.c (revision 753) @@ -959,10 +959,5 @@ // print exit message - if ( string == NULL ) - { - _printf("\n[GIET WARNING] Exit thread %x in application %s\n", - trdid , vspace[vsid].name ); - } - else + if ( string != NULL ) { _printf("\n[GIET WARNING] Exit thread %x in vspace %s\n" Index: soft/giet_vm/giet_python/genmap =================================================================== --- soft/giet_vm/giet_python/genmap (revision 752) +++ soft/giet_vm/giet_python/genmap (revision 753) @@ -43,4 +43,5 @@ # - shell # - transpose +# - coremark ################################################################################### @@ -162,4 +163,8 @@ default = False, help = 'map the "transpose" application for the GietVM' ) + +parser.add_option( '--coremark', action = 'store_true', dest = 'coremark', + default = False, + help = 'map the "coremark" application for the GietVM' ) ################################################################################### @@ -202,4 +207,5 @@ map_sort = options.sort # map "sort" application if True map_transpose = options.transpose # map "transpose" application if True +map_coremark = options.coremark # map "coremark" application if True ################################################################################### @@ -290,4 +296,9 @@ print '[genmap] application "transpose" will be loaded' +if ( map_coremark ): + appli = __import__( 'coremark' ) + appli.extend( mapping ) + print '[genmap] application "coremark" will be loaded' + ################################################################################### # Generate xml file if required.