You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2022/08/15 22:18:54 UTC

[GitHub] [tvm] vinx13 opened a new pull request, #12445: [MetaSchedule] Add logging of usage of tensor intrinsics

vinx13 opened a new pull request, #12445:
URL: https://github.com/apache/tvm/pull/12445

   cc @zxybazh @AndrewZhaoLuo 


-- 
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: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org

[GitHub] [tvm] vinx13 merged pull request #12445: [MetaSchedule] Add logging of usage of tensor intrinsics

Posted by GitBox <gi...@apache.org>.
vinx13 merged PR #12445:
URL: https://github.com/apache/tvm/pull/12445


-- 
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: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org

[GitHub] [tvm] vinx13 commented on a diff in pull request #12445: [MetaSchedule] Add logging of usage of tensor intrinsics

Posted by GitBox <gi...@apache.org>.
vinx13 commented on code in PR #12445:
URL: https://github.com/apache/tvm/pull/12445#discussion_r946178085


##########
src/meta_schedule/schedule_rule/multi_level_tiling_with_intrin.cc:
##########
@@ -55,8 +55,10 @@ class MultiLevelTilingWithIntrinNode : public MultiLevelTilingNode {
     auto res = MultiLevelTilingNode::Apply(sch->Copy(), block_rv);
 
     if (res.empty()) {
+      TVM_PY_LOG(INFO, logging_func) << "The workload cannot be tensorized.";

Review Comment:
   also added logging in the above case, `CheckAutoTensorizeApplicable` is non-exhaustive, tensorization can still fail even if it passes the check



-- 
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: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org

[GitHub] [tvm] AndrewZhaoLuo commented on a diff in pull request #12445: [MetaSchedule] Add logging of usage of tensor intrinsics

Posted by GitBox <gi...@apache.org>.
AndrewZhaoLuo commented on code in PR #12445:
URL: https://github.com/apache/tvm/pull/12445#discussion_r946175474


##########
src/meta_schedule/schedule_rule/multi_level_tiling.h:
##########
@@ -186,6 +186,8 @@ class MultiLevelTilingNode : public ScheduleRuleNode {
   int thread_warp_size_;
   /*! \brief The maximum number of threads to be used size of a thread warp */
   int max_threads_per_block_;
+  /*! \brief The logging function */
+  PackedFunc logging_func;

Review Comment:
   Where is this set? (If at all yet)



##########
src/meta_schedule/schedule_rule/multi_level_tiling_with_intrin.cc:
##########
@@ -55,8 +55,10 @@ class MultiLevelTilingWithIntrinNode : public MultiLevelTilingNode {
     auto res = MultiLevelTilingNode::Apply(sch->Copy(), block_rv);
 
     if (res.empty()) {
+      TVM_PY_LOG(INFO, logging_func) << "The workload cannot be tensorized.";

Review Comment:
   What is the difference between hitting this and `!CheckAutoTensorizeApplicable`? 



##########
src/meta_schedule/schedule_rule/multi_level_tiling_with_intrin.cc:
##########
@@ -55,8 +55,10 @@ class MultiLevelTilingWithIntrinNode : public MultiLevelTilingNode {
     auto res = MultiLevelTilingNode::Apply(sch->Copy(), block_rv);
 
     if (res.empty()) {
+      TVM_PY_LOG(INFO, logging_func) << "The workload cannot be tensorized.";

Review Comment:
   What is the difference between hitting this and `!CheckAutoTensorizeApplicable` branch? 



-- 
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: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org