griefith/test
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Depsgraph: Separate loop for ID node flags clear

Browse Source 
This ends up in less number of memory writes, which should have positive effect
on performance.
This commit is contained in:
Sergey Sharybin
2017-12-01 17:30:36 +01:00
parent 59e8b72881
commit 6365ddcab5
1 changed files with 19 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
source/blender/depsgraph/intern/eval/deg_eval_flush.cc
View File

@@ -65,32 +65,35 @@ typedef std::deque<OperationDepsNode *> FlushQueue;
namespace {
void flush_init_func(void *data_v, int i)
void flush_init_operation_node_func(void *data_v, int i)
{
/* ID node's done flag is used to avoid multiple editors update
* for the same ID.
*/
Depsgraph *graph = (Depsgraph *)data_v;
OperationDepsNode *node = graph->operations[i];
ComponentDepsNode *comp_node = node->owner;
IDDepsNode *id_node = comp_node->owner;
id_node->done = 0;
comp_node->done = COMPONENT_STATE_NONE;
node->scheduled = false;
}
void flush_init_id_node_func(void *data_v, int i)
{
Depsgraph *graph = (Depsgraph *)data_v;
IDDepsNode *id_node = graph->id_nodes[i];
id_node->done = 0;
GHASH_FOREACH_BEGIN(ComponentDepsNode *, comp_node, id_node->components)
comp_node->done = COMPONENT_STATE_NONE;
GHASH_FOREACH_END();
}
BLI_INLINE void flush_prepare(Depsgraph *graph)
{
/* TODO(sergey): With a bit of flag magic we can get rid of this
* extra loop.
*/
const int num_operations = graph->operations.size();
const bool do_threads = num_operations > 256;
BLI_task_parallel_range(0,
num_operations,
BLI_task_parallel_range(0, num_operations,
graph,
flush_init_func,
do_threads);
flush_init_operation_node_func,
(num_operations > 256));
const int num_id_nodes = graph->id_nodes.size();
BLI_task_parallel_range(0, num_id_nodes,
graph,
flush_init_id_node_func,
(num_id_nodes > 256));
}
BLI_INLINE void flush_schedule_entrypoints(Depsgraph *graph, FlushQueue *queue)