Collecting feedback from ACM/IEEE MICRO'23 authors to improve CM automation about ck HOT 10 OPEN

Turning on cache explicitly works.

The desired behavior was to allow the user to specify the path to their local version of the resource before continuing with the artifact reproduction process

This command seems to have the desired behavior:

cmr "download file _url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso" --local_path=/home/n869p538/cpu2017-1_0_5.iso --force_cache=true

Thanks for your help

from ck.

1. need to provide specs of _cm.yaml of CM scripts (https://github.com/mlcommons/ck/blob/master/docs/specs/cm-automation-script.md) and how to pass keys and ENV to sub dependencies

from ck.

Add Docker installation (many artifacts use it): #941

from ck.

Need FAQ about CM scripts ...

from ck.

--local_path option still results in downloading of remote file in spite of local resouce's being available:

- tags: download,file,_url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso
  env:
    CM_DOWNLOAD_FINAL_ENV_NAME: 'SPEC_ISO'
  extra_cache_tags: micro23,artifact,ae,xfm,spec
  force_cache: true

Still results in downloading the file in spite of the local path to the resource being provided

(cm) n869p538@bristol:reproduce-paper-m2023-xfm$ cmr "reproduce project micro-2023 xfm _install_spec_deps" 
* cm run script "reproduce project micro-2023 xfm _install_spec_deps"
  * cm run script "detect os"
  * cm run script "get python" 
  * cm run script "get generic-python-lib _pandas"
  * cm run script "get generic-python-lib _matplotlib"
  * cm run script "get git repo _repo.https://github.com/neel-patel-1/XFM_MICRO2023"
  * cm run script "download file _url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso"

Downloaded:   4%

from ck.

--local_path option still results in downloading of remote file in spite of local resouce's being available:

- tags: download,file,_url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso
  env:
    CM_DOWNLOAD_FINAL_ENV_NAME: 'SPEC_ISO'
  extra_cache_tags: micro23,artifact,ae,xfm,spec
  force_cache: true

Still results in downloading the file in spite of the local path to the resource being provided

(cm) n869p538@bristol:reproduce-paper-m2023-xfm$ cmr "reproduce project micro-2023 xfm _install_spec_deps" 
* cm run script "reproduce project micro-2023 xfm _install_spec_deps"
  * cm run script "detect os"
  * cm run script "get python" 
  * cm run script "get generic-python-lib _pandas"
  * cm run script "get generic-python-lib _matplotlib"
  * cm run script "get git repo _repo.https://github.com/neel-patel-1/XFM_MICRO2023"
  * cm run script "download file _url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso"

Downloaded:   4%

Thanks for your feedback! CC @arjunsuresh

from ck.

@neel-patel-1 can you please share how exactly you added --local_path option?

from ck.

(cm) n869p538@bristol:reproduce-paper-m2023-xfm$ cmr "download file _url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso" --local_path=/home/n869p538/cpu2017-1_0_5.iso

from ck.

@arjunsuresh The command sequence looked as follows:

cmr "download file _url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso" --local_path=/home/n869p538/cpu2017-1_0_5.iso

(cm) n869p538@bristol:reproduce-paper-m2023-xfm$ cmr "reproduce project micro-2023 xfm _install_spec_deps" 
* cm run script "reproduce project micro-2023 xfm _install_spec_deps"
  * cm run script "detect os"
  * cm run script "get python" 
  * cm run script "get generic-python-lib _pandas"
  * cm run script "get generic-python-lib _matplotlib"
  * cm run script "get git repo _repo.https://github.com/neel-patel-1/XFM_MICRO2023"
  * cm run script "download file _url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso"

Downloaded:   4%

from ck.

Thank you @neel-patel-1 for sharing the details. Actually, for the download file CM script we are not turning on cache by default. When we want to use cache, we have to turn it on explicitly as follows:

cmr "download file _url.https://spec2017iso.s3.us-east-2.amazonaws.com/cpu2017-1_0_5.iso" --json --force_cache=true

And in the _cm.json also force_cache must be set for the URL download dependency like done here

Please let us know if this works.

from ck.