#ifdef MULTICORE_GC
#ifdef SMEMF
+#define NUM_CORES2TEST 5
#ifdef GC_1
-int core2test[1][5] = {
+int core2test[1][NUM_CORES2TEST] = {
{0, -1, -1, -1, -1}
};
#elif defined GC_56
-int core2test[56][5] = {
+int core2test[56][NUM_CORES2TEST] = {
{ 0, -1, 7, -1, 1}, { 1, -1, 8, 0, 2}, { 2, -1, 9, 1, 3},
{ 3, -1, 10, 2, 4}, { 4, -1, 11, 3, 5}, { 5, -1, 12, 4, 6},
{ 6, -1, 13, 5, -1}, { 7, 0, 14, -1, 8}, { 8, 1, 15, 7, 9},
{54, 47, -1, 53, 55}, {55, 48, -1, 54, -1}
};
#elif defined GC_62
-int core2test[62][5] = {
+int core2test[62][NUM_CORES2TEST] = {
{ 0, -1, 6, -1, 1}, { 1, -1, 7, 0, 2}, { 2, -1, 8, 1, 3},
{ 3, -1, 9, 2, 4}, { 4, -1, 10, 3, 5}, { 5, -1, 11, 4, -1},
{ 6, 0, 14, -1, 7}, { 7, 1, 15, 6, 8}, { 8, 2, 16, 7, 9},
};
#endif // GC_1
#elif defined SMEMM
+unsigned int gcmem_mixed_threshold = 0;
+unsigned int gcmem_mixed_usedmem = 0;
+#define NUM_CORES2TEST 9
+#ifdef GC_1
+int core2test[1][NUM_CORES2TEST] = {
+ {0, -1, -1, -1, -1, -1, -1, -1, -1}
+};
+#elif defined GC_56
+int core2test[56][5] = {
+ { 0, -1, 7, -1, 1, -1, 14, -1, 2}, { 1, -1, 8, 0, 2, -1, 15, -1, 3},
+ { 2, -1, 9, 1, 3, -1, 16, 0, 4}, { 3, -1, 10, 2, 4, -1, 17, 1, 5},
+ { 4, -1, 11, 3, 5, -1, 18, 2, 6}, { 5, -1, 12, 4, 6, -1, 19, 3, -1},
+ { 6, -1, 13, 5, -1, -1, 20, 4, -1}, { 7, 0, 14, -1, 8, -1, 21, -1, 9},
+ { 8, 1, 15, 7, 9, -1, 22, -1, 10}, { 9, 2, 16, 8, 10, -1, 23, 7, 11},
+ {10, 3, 17, 9, 11, -1, 24, 8, 12}, {11, 4, 18, 10, 12, -1, 25, 9, 13},
+ {12, 5, 19, 11, 13, -1, 26, 10, -1}, {13, 6, 20, 12, -1, -1, 27, 11, -1},
+ {14, 7, 21, -1, 15, 0, 28, -1, 16}, {15, 8, 22, 14, 16, 1, 29, -1, 17},
+ {16, 9, 23, 15, 17, 2, 30, 14, 18}, {17, 10, 24, 16, 18, 3, 31, 15, 19},
+ {18, 11, 25, 17, 19, 4, 32, 16, 20}, {19, 12, 26, 18, 20, 5, 33, 17, -1},
+ {20, 13, 27, 19, -1, 6, 34, 18, -1}, {21, 14, 28, -1, 22, 7, 35, -1, 23},
+ {22, 15, 29, 21, 23, 8, 36, -1, 24}, {23, 16, 30, 22, 24, 9, 37, 21, 25},
+ {24, 17, 31, 23, 25, 10, 38, 22, 26}, {25, 18, 32, 24, 26, 11, 39, 23, 27},
+ {26, 19, 33, 25, 27, 12, 40, 24, -1}, {27, 20, 34, 26, -1, 13, 41, 25, -1},
+ {28, 21, 35, -1, 29, 14, 42, -1, 30}, {29, 22, 36, 28, 30, 15, 43, -1, 31},
+ {30, 23, 37, 29, 31, 16, 44, 28, 32}, {31, 24, 38, 30, 32, 17, 45, 29, 33},
+ {32, 25, 39, 31, 33, 18, 46, 30, 34}, {33, 26, 40, 32, 34, 19, 47, 31, -1},
+ {34, 27, 41, 33, -1, 20, 48, 32, -1}, {35, 28, 42, -1, 36, 21, 49, -1, 37},
+ {36, 29, 43, 35, 37, 22, 50, -1, 38}, {37, 30, 44, 36, 38, 23, 51, 35, 39},
+ {38, 31, 45, 37, 39, 24, 52, 36, 40}, {39, 32, 46, 38, 40, 25, 53, 37, 41},
+ {40, 33, 47, 39, 41, 26, 54, 38, -1}, {41, 34, 48, 40, -1, 27, 55, 39, -1},
+ {42, 35, 49, -1, 43, 28, -1, -1, 44}, {43, 36, 50, 42, 44, 29, -1, -1, 45},
+ {44, 37, 51, 43, 45, 30, -1, 42, 46}, {45, 38, 52, 44, 46, 31, -1, 43, 47},
+ {46, 39, 53, 45, 47, 32, -1, 44, 48}, {47, 40, 54, 46, 48, 33, -1, 45, -1},
+ {48, 41, 55, 47, -1, 34, -1, 46, -1}, {49, 42, -1, -1, 50, 35, -1, -1, 51},
+ {50, 43, -1, 49, 51, 36, -1, -1, 52}, {51, 44, -1, 50, 52, 37, -1, 49, 53},
+ {52, 45, -1, 51, 53, 38, -1, 50, 54}, {53, 46, -1, 52, 54, 39, -1, 51, 55},
+ {54, 47, -1, 53, 55, 40, -1, 52, -1}, {55, 48, -1, 54, -1, 41, -1, 53, -1}
+};
+#elif defined GC_62
+int core2test[62][5] = {
+ { 0, -1, 6, -1, 1, -1, 14, -1, 2}, { 1, -1, 7, 0, 2, -1, 15, -1, 3},
+ { 2, -1, 8, 1, 3, -1, 16, 0, 4}, { 3, -1, 9, 2, 4, -1, 17, 1, 5},
+ { 4, -1, 10, 3, 5, -1, 18, 2, -1}, { 5, -1, 11, 4, -1, -1, 19, 3, -1},
+ { 6, 0, 14, -1, 7, -1, 22, -1, 8}, { 7, 1, 15, 6, 8, -1, 23, -1, 9},
+ { 8, 2, 16, 7, 9, -1, 24, 6, 10}, { 9, 3, 17, 8, 10, -1, 25, 7, 11},
+ {10, 4, 18, 9, 11, -1, 26, 8, 12}, {11, 5, 19, 10, 12, -1, 27, 9, 13},
+ {12, -1, 20, 11, 13, -1, 28, 10, -1}, {13, -1, 21, 12, -1, -1, 29, 11, -1},
+ {14, 6, 22, -1, 15, 0, 30, -1, 16}, {15, 7, 23, 14, 16, 1, 31, -1, 17},
+ {16, 8, 24, 15, 17, 2, 32, 14, 18}, {17, 9, 25, 16, 18, 3, 33, 15, 19},
+ {18, 10, 26, 17, 19, 4, 34, 16, 20}, {19, 11, 27, 18, 20, 5, 35, 17, 21},
+ {20, 12, 28, 19, 21, -1, 36, 18, -1}, {21, 13, 29, 28, -1, -1, 37, 19, -1},
+ {22, 14, 30, -1, 23, 6, 38, -1, 24}, {23, 15, 31, 22, 24, 7, 39, -1, 25},
+ {24, 16, 32, 23, 25, 8, 40, 22, 26}, {25, 17, 33, 24, 26, 9, 41, 23, 27},
+ {26, 18, 34, 25, 27, 10, 42, 24, 28}, {27, 19, 35, 26, 28, 11, 43, 25, 29},
+ {28, 20, 36, 27, 29, 12, 44, 26, -1}, {29, 21, 37, 28, -1, 13, 45, 27, -1},
+ {30, 22, 38, -1, 31, 22, 46, -1, 32}, {31, 23, 39, 30, 32, 15, 47, -1, 33},
+ {32, 24, 40, 31, 33, 16, 48, 30, 34}, {33, 25, 41, 32, 34, 17, 49, 31, 35},
+ {34, 26, 42, 33, 35, 18, 50, 32, 36}, {35, 27, 43, 34, 36, 19, 51, 33, 37},
+ {36, 28, 44, 35, 37, 20, 52, 34, -1}, {37, 29, 45, 36, -1, 21, 53, 35, -1},
+ {38, 30, 46, -1, 39, 22, 54, -1, 40}, {39, 31, 47, 38, 40, 23, 55, -1, 41},
+ {40, 32, 48, 39, 41, 24, 56, 38, 42}, {41, 33, 49, 40, 42, 25, 57, 39, 43},
+ {42, 34, 50, 41, 43, 26, 58, 40, 44}, {43, 35, 51, 42, 44, 27, 59, 41, 45},
+ {44, 36, 52, 43, 45, 28, 60, 42, -1}, {45, 37, 53, 44, -1, 29, 61, 43, -1},
+ {46, 38, 54, -1, 47, 30, -1, -1, 48}, {47, 39, 55, 46, 48, 31, -1, -1, 49},
+ {48, 40, 56, 47, 49, 32, -1, 46, 50}, {49, 41, 57, 48, 50, 33, -1, 47, 51},
+ {50, 42, 58, 49, 51, 34, -1, 48, 52}, {51, 43, 59, 50, 52, 35, -1, 49, 53},
+ {52, 44, 60, 51, 53, 36, -1, 50, -1}, {53, 45, 61, 52, -1, 37, -1, 51, -1},
+ {54, 46, -1, -1, 55, 38, -1, -1, 56}, {55, 47, -1, 54, 56, 39, -1, -1, 57},
+ {56, 48, -1, 55, 57, 40, -1, 54, 58}, {57, 49, -1, 56, 59, 41, -1, 55, 59},
+ {58, 50, -1, 57, 59, 42, -1, 56, 60}, {59, 51, -1, 58, 60, 43, -1, 57, 61},
+ {60, 52, -1, 59, 61, 44, -1, 58, -1}, {61, 53, -1, 60, -1, 45, -1, 59, -1}
+};
+#endif // GC_1
#endif
inline __attribute__((always_inline))
#elif defined SMEMM
// Allocate the local shared memory to each core with the highest priority,
// if a core has used up its local shared memory, try to allocate the
- // shared memory that belong to its neighbours first, if failed, allocate
- // the shared memory globally. If all the shared memory has been used up,
- // start gc.
+ // shared memory that belong to its neighbours first, if failed, check
+ // current memory allocation rate, if it has already reached the threshold,
+ // start gc, otherwise, allocate the shared memory globally. If all the
+ // shared memory has been used up, start gc.
bamboo_smem_mode = SMEMMIXED;
#elif defined SMEMG
// Allocate all the memory chunks globally, do not consider the host cores
}
BAMBOO_MEMSET_WH(gcrpointertbls,0,sizeof(mgcsharedhashtbl_t *)*NUMCORES4GC);
//sizeof(struct RuntimeHash *)*NUMCORES4GC);
+#ifdef SMEMM
+ gcmem_mixed_threshold =
+ (unsigned int)((BAMBOO_SHARED_MEM_SIZE-(gcbaseva-BAMBOO_BASE_VA))*0.8);
+ gcmem_mixed_usedmem = 0;
+#endif
#else
// create the lock table, lockresult table and obj queue
locktable.size = 20;
return mem;
} // void * localmalloc_I(int, int, int *)
+#ifdef SMEMF
// Allocate the local shared memory to each core with the highest priority,
// if a core has used up its local shared memory, try to allocate the
// shared memory that belong to its neighbours, if also failed, start gc.
int i = 0;
int j = 0;
int k = 0;
- //int core2test[5]={coren,-1,-1,-1,-1};//(x,y),(x-1,y),(x+1,y),(x,y-1),(x,y+1)
int coords_x = bamboo_cpu2coords[coren*2];
int coords_y = bamboo_cpu2coords[coren*2+1];
int ii = 1;
- /*if(coords_x != 0) {
- core2test[ii++] = bamboo_coords2cpu[coords_x-1][coords_y];
- }
- if(coords_x != 7) {
- core2test[ii++] = bamboo_coords2cpu[coords_x+1][coords_y];
- }
- if(coords_y != 0) {
- core2test[ii++] = bamboo_coords2cpu[coords_x][coords_y-1];
- }
- if(coords_y != 7) {
- core2test[ii++] = bamboo_coords2cpu[coords_x][coords_y+1];
- }*/
int tofindb = gc_core2block[2*core2test[coren][k]+i]+(NUMCORES4GC*2)*j;
int totest = tofindb;
int bound = BAMBOO_SMEM_SIZE_L;
// try to malloc shared memory assigned to the neighbour cores
do{
k++;
- if(k > 4) {
+ if(k >= NUM_CORES2TEST) {
// no more memory available on either coren or its neighbour cores
foundsmem = 2;
goto memsearchresult;
return mem;
} // void * fixedmalloc_I(int, int, int *)
+#endif // #ifdef SMEMF
+
+#ifdef SMEMM
+// Allocate the local shared memory to each core with the highest priority,
+// if a core has used up its local shared memory, try to allocate the
+// shared memory that belong to its neighbours first, if failed, check
+// current memory allocation rate, if it has already reached the threshold,
+// start gc, otherwise, allocate the shared memory globally. If all the
+// shared memory has been used up, start gc.
+void * mixedmalloc_I(int coren,
+ int isize,
+ int * allocsize) {
+ void * mem = NULL;
+ int i = 0;
+ int j = 0;
+ int k = 0;
+ int coords_x = bamboo_cpu2coords[coren*2];
+ int coords_y = bamboo_cpu2coords[coren*2+1];
+ int ii = 1;
+ int tofindb = gc_core2block[2*core2test[coren][k]+i]+(NUMCORES4GC*2)*j;
+ int totest = tofindb;
+ int bound = BAMBOO_SMEM_SIZE_L;
+ int foundsmem = 0;
+ int size = 0;
+ do {
+ bound = (totest < NUMCORES4GC) ? BAMBOO_SMEM_SIZE_L : BAMBOO_SMEM_SIZE;
+ int nsize = bamboo_smemtbl[totest];
+ bool islocal = true;
+ if(nsize < bound) {
+ bool tocheck = true;
+ // have some space in the block
+ if(totest == tofindb) {
+ // the first partition
+ size = bound - nsize;
+ } else if(nsize == 0) {
+ // an empty partition, can be appended
+ size += bound;
+ } else {
+ // not an empty partition, can not be appended
+ // the last continuous block is not big enough, go to check the next
+ // local block
+ islocal = true;
+ tocheck = false;
+ } // if(totest == tofindb) else if(nsize == 0) else ...
+ if(tocheck) {
+ if(size >= isize) {
+ // have enough space in the block, malloc
+ foundsmem = 1;
+ break;
+ } else {
+ // no enough space yet, try to append next continuous block
+ // TODO may consider to go to next local block?
+ islocal = false;
+ } // if(size > isize) else ...
+ } // if(tocheck)
+ } // if(nsize < bound)
+ if(islocal) {
+ // no space in the block, go to check the next block
+ i++;
+ if(2==i) {
+ i = 0;
+ j++;
+ }
+ tofindb=totest=gc_core2block[2*core2test[coren][k]+i]+(NUMCORES4GC*2)*j;
+ } else {
+ totest += 1;
+ } // if(islocal) else ...
+ if(totest > gcnumblock-1-bamboo_reserved_smem) {
+ // no more local mem, do not find suitable block on local mem
+ // try to malloc shared memory assigned to the neighbour cores
+ do{
+ k++;
+ if(k >= NUM_CORES2TEST) {
+ if(gcmem_mixed_usedmem >= gcmem_mixed_threshold) {
+ // no more memory available on either coren or its neighbour cores
+ foundsmem = 2;
+ goto memmixedsearchresult;
+ } else {
+ // try allocate globally
+ mem = globalmalloc_I(coren, isize, allocsize);
+ return mem;
+ }
+ }
+ } while(core2test[coren][k] == -1);
+ i = 0;
+ j = 0;
+ tofindb=totest=gc_core2block[2*core2test[coren][k]+i]+(NUMCORES4GC*2)*j;
+ } // if(totest > gcnumblock-1-bamboo_reserved_smem) ...
+ } while(true);
+
+memmixedsearchresult:
+ if(foundsmem == 1) {
+ // find suitable block
+ mem = gcbaseva+bamboo_smemtbl[tofindb]+((tofindb<NUMCORES4GC) ?
+ (BAMBOO_SMEM_SIZE_L*tofindb) : (BAMBOO_LARGE_SMEM_BOUND+
+ (tofindb-NUMCORES4GC)*BAMBOO_SMEM_SIZE));
+ *allocsize = size;
+ // set bamboo_smemtbl
+ for(i = tofindb; i <= totest; i++) {
+ bamboo_smemtbl[i]=(i<NUMCORES4GC)?BAMBOO_SMEM_SIZE_L:BAMBOO_SMEM_SIZE;
+ }
+ gcmem_mixed_usedmem += size;
+ if(tofindb == bamboo_free_block) {
+ bamboo_free_block = totest+1;
+ }
+ } else if(foundsmem == 2) {
+ // no suitable block
+ *allocsize = 0;
+ }
+
+ return mem;
+} // void * mixedmalloc_I(int, int, int *)
+#endif // #ifdef SMEMM
// Allocate all the memory chunks globally, do not consider the host cores
// When all the shared memory are used up, start gc.
}
case SMEMFIXED: {
- // TODO not supported yet
- //BAMBOO_EXIT(0xe001);
+#ifdef SMEMF
mem = fixedmalloc_I(coren, isize, allocsize);
+#else
+ // not supported yet
+ BAMBOO_EXIT(0xe001);
+#endif
break;
}
case SMEMMIXED: {
- // TODO not supported yet
+#ifdef SMEMM
+ mem = mixedmalloc_I(coren, isize, allocsize);
+#else
+ // not supported yet
BAMBOO_EXIT(0xe002);
+#endif
break;
}
projects / IRC.git / commitdiff