liuneng1994 commented on code in PR #5893:
URL: https://github.com/apache/incubator-gluten/pull/5893#discussion_r1618567907
##########
cpp-ch/local-engine/Common/CHUtil.cpp:
##########
@@ -672,6 +678,21 @@ void
BackendInitializerUtil::initSettings(std::map<std::string, std::string> & b
settings.set("function_json_value_return_type_allow_complex", true);
settings.set("function_json_value_return_type_allow_nullable", true);
settings.set("precise_float_parsing", true);
+ if (backend_conf_map.contains(GLUTEN_TASK_OFFHEAP))
+ {
+ auto task_memory =
std::stoull(backend_conf_map.at(GLUTEN_TASK_OFFHEAP));
+ if (!backend_conf_map.contains(CH_RUNTIME_SETTINGS_PREFIX +
"max_bytes_before_external_sort"))
+ {
+ settings.max_bytes_before_external_sort = static_cast<size_t>(0.8
* task_memory);
+ }
+ if (!backend_conf_map.contains(CH_RUNTIME_SETTINGS_PREFIX +
"prefer_external_sort_block_bytes"))
+ {
+ auto mem_gb = task_memory / static_cast<double>(1_GiB);
+ // 2.8x+5, Heuristics calculate the block size of external sort,
[8,16]
Review Comment:
大概定了一下 1G 8M 2G 10M 3G 14M 4G 16M四个数据点然后做了一个线性回归，数据点是跟据测试效果大致选择的
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
