Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

enhance pragma to support single point copy #863

Merged
merged 28 commits into from
Feb 4, 2018
Merged

enhance pragma to support single point copy #863

merged 28 commits into from
Feb 4, 2018

Conversation

libing4752
Copy link
Contributor

to enhance pragma to support single point copy by adding a special path to generate buffers with stride = 1,and offset = 0

Copy link
Member

@tqchen tqchen left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am thinking a better way to do this. Maybe we can simply generalize DetectLinearEquation to accept empty list of loop_var, which returns everything as base. This is well-defined and consistent with original defition

@tqchen tqchen merged commit f14e7ac into apache:master Feb 4, 2018
tqchen pushed a commit to tqchen/tvm that referenced this pull request Jul 6, 2018
* modified schedule_dataflow_rewrite.cc to fix losing tensor problem

* modified schedule_dataflow_rewrite.cc for lint scan

* modified schedule_dataflow_rewrite.cc for lint scan

* using tensor's value_index to index output of stage op

* repare address offset for different kinds of dtype

* bc

* aaa

* aaaaa

* repare address for different dtypes

* remove nonsense files

* add whitespace of line 581

* use base alloc elem_type

* enhance the testcast of basic buffer is 64bits,32bits,16bits,8bits

* use extends[0]->type() as dtype of offset

* clear program writes

* enhance inject_copy_intin to support of pragma stmt with no loops

* fix cpplint errors

* fix cpplint error of !

* enhance detectLinearEquation to support with no loop vars

* fix cpplint errors
sergei-mironov pushed a commit to sergei-mironov/tvm that referenced this pull request Aug 8, 2018
* modified schedule_dataflow_rewrite.cc to fix losing tensor problem

* modified schedule_dataflow_rewrite.cc for lint scan

* modified schedule_dataflow_rewrite.cc for lint scan

* using tensor's value_index to index output of stage op

* repare address offset for different kinds of dtype

* bc

* aaa

* aaaaa

* repare address for different dtypes

* remove nonsense files

* add whitespace of line 581

* use base alloc elem_type

* enhance the testcast of basic buffer is 64bits,32bits,16bits,8bits

* use extends[0]->type() as dtype of offset

* clear program writes

* enhance inject_copy_intin to support of pragma stmt with no loops

* fix cpplint errors

* fix cpplint error of !

* enhance detectLinearEquation to support with no loop vars

* fix cpplint errors
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants