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 2020/10/28 18:31:13 UTC
[GitHub] [incubator-tvm] areusch opened a new pull request #6789: [µTVM] Add serial transport, parameterize µTVM Zephyr test, run on physical HW
areusch opened a new pull request #6789:
URL: https://github.com/apache/incubator-tvm/pull/6789
This PR adds a pyserial transport to µTVM, allowing it to be used with physical hardware attached to the machine. This splits #6703 in half and adds just the zephyr-specific runtime components. To test this change, you can either run it locally if you have the necessary dependencies or use the VM in #6703
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [incubator-tvm] areusch commented on a change in pull request #6789: [µTVM] Add serial transport, parameterize µTVM Zephyr test, run on physical HW
Posted by GitBox <gi...@apache.org>.
areusch commented on a change in pull request #6789:
URL: https://github.com/apache/incubator-tvm/pull/6789#discussion_r513774283
##########
File path: python/tvm/target/target.py
##########
@@ -220,19 +220,24 @@ def intel_graphics(model="unknown", options=None):
return Target(" ".join(["opencl"] + opts))
-def micro(hardware="unknown", options=None):
+def micro(model="unknown", options=None):
"""Returns a microTVM target.
Parameters
----------
- hardware : str
- Canonically identifies the target device; typicaly one of cortex-mX, or a specific SoC model
- when that model has been tested to work with microTVM.
+ model : str
+ Canonically identifies the target device. This is typically a CPU or board level name (other
+ flags such as -mcpu identify the ISA).
options : str or list of str
Additional options
"""
- trans_table = {"host": ["-mcpu=native"]}
- opts = _merge_opts(trans_table[hardware] + ["-runtime=c", "--system-lib"], options)
+ trans_table = {
Review comment:
you can always pass in a specific CPU by manually creating a target: `tvm.target.create("c -mcpu=cortex-m4")`. I do agree that this might become unworkable in the long run--i'm mostly doing this here since it was past convention. for now i don't want to mess with that convention, but i'm open to it in the future. feel free to suggest too if you have a better idea!
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [incubator-tvm] tom-gall commented on a change in pull request #6789: [µTVM] Add serial transport, parameterize µTVM Zephyr test, run on physical HW
Posted by GitBox <gi...@apache.org>.
tom-gall commented on a change in pull request #6789:
URL: https://github.com/apache/incubator-tvm/pull/6789#discussion_r513742956
##########
File path: python/tvm/target/target.py
##########
@@ -220,19 +220,24 @@ def intel_graphics(model="unknown", options=None):
return Target(" ".join(["opencl"] + opts))
-def micro(hardware="unknown", options=None):
+def micro(model="unknown", options=None):
"""Returns a microTVM target.
Parameters
----------
- hardware : str
- Canonically identifies the target device; typicaly one of cortex-mX, or a specific SoC model
- when that model has been tested to work with microTVM.
+ model : str
+ Canonically identifies the target device. This is typically a CPU or board level name (other
+ flags such as -mcpu identify the ISA).
options : str or list of str
Additional options
"""
- trans_table = {"host": ["-mcpu=native"]}
- opts = _merge_opts(trans_table[hardware] + ["-runtime=c", "--system-lib"], options)
+ trans_table = {
Review comment:
Someday we're going to have to / want to work with targets other then cpu. Having a per board hard code here makes me a little uneasy. Should we be open to a board / -mcpu setting passed in to override or add to the trans_table?
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [incubator-tvm] areusch commented on pull request #6789: [µTVM] Add serial transport, parameterize µTVM Zephyr test, run on physical HW
Posted by GitBox <gi...@apache.org>.
areusch commented on pull request #6789:
URL: https://github.com/apache/incubator-tvm/pull/6789#issuecomment-718173898
cc @tmoreau89 @tqchen @u99127 @manupa-arm @leandron @tom-gall @liangfu @mshawcroft
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org