Reintroduce autoranging

[psyhtest]

Fixes #253. Adding a new hash to the list of acceptable checksums allows submitters to use any commit on this r1.1 branch for v1.1 submission - both with max ranging (before this PR) and auto ranging (after this PR).

[github-actions]

MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅

[s-idgunji]

Hi Anton (@psyhtest ) - Thanks. How does one select max vs auto if both are supported ? Or have we deprecated max with this ?

[nvpohanh]

LGTM.

[psyhtest]

Sachin (@s-idgunji) Every submitter needs to clone the r1.1 branch:

$ git clone https://github.com/mlcommons/power-dev/commits/r1.1 power-dev-r1.1

If they do it now:

• If they want to use auto ranging (the r1.0 behaviour and the new default behaviour), they should do nothing.
• If they want to use max ranging (the r1.1 behaviour at the time the random seeds got distributed), they should do "downgrade":

$ cd power-dev-r1.1
$ git checkout 020bd3cb33e

(020bd3cb33e is the hash of the last commit on the r1.1 branch with the max ranging behaviour.)

If they did it before today:

• If they want to use auto ranging, they should do "upgrade":

$ cd power-dev-r1.1
$ git pull

• If they want to use max ranging, they should do nothing.

Results obtained with either approach will be accepted by the checker. Does it make sense?

In the future, we should discuss having explicit options:

• Auto ranging (default)?
• Max ranging
• User-selected ranging (without the need for a ranging run).

[s-idgunji]

@psyhtest Thanks for the write up It's likely that some submitters have used versions prior to your changes last week, so we want to make sure that those submitters do not face issues due to the version used. We can discuss this in our Power WG this week. I'm not sure how many will "do it now" given we're 3 days away from submission , but we can discuss all options.

Sachin (@s-idgunji) Every submitter needs to clone the r1.1 branch:

$ git clone https://github.com/mlcommons/power-dev/commits/r1.1 power-dev-r1.1

If they do it now:

• If they want to use auto ranging (the r1.0 behaviour and the new default behaviour), they should do nothing.
• If they want to use max ranging (the r1.1 behaviour at the time the random seeds got distributed), they should do "downgrade":

$ cd power-dev-r1.1
$ git checkout 020bd3cb33e

(020bd3cb33e is the hash of the last commit on the r1.1 branch with the max ranging behaviour.)

If they did it before today:

• If they want to use auto ranging, they should do "upgrade":

$ cd power-dev-r1.1
$ git pull

• If they want to use max ranging, they should do nothing.

Results obtained with either approach will be accepted by the checker. Does it make sense?

In the future, we should discuss having explicit options:

• Auto ranging (default)?
• Max ranging
• User-selected ranging (without the need for a ranging run).