[ViewOpGraph] Improve GraphViz output

[ehein6]

This patch improves the GraphViz output of ViewOpGraph (--view-op-graph).

• Switch to rectangular record-based nodes, inspired by a similar visualization in Glow. Rectangles make more efficient use of space when printing text.
• Add input and output ports for each operand and result, and remove edge labels.
• Switch to a muted color palette to reduce eye strain.

Before:

After:

More comparisons here: compare.tar.gz

This visualization displays the short name of each operand along the top edge of each node.
Result types are displayed along the bottom edge of each node. This differs somewhat from
reading the mlir source code, where the name of the result comes before the operation name.

But in my opinion this style works better in the visualization for the following reasons:

1. Most ops have many operands but few results. So we need a short
   representation for the operands, but we can use a longer representation for the results.
2. Since a value has only one producer, we only need to print the full type in one place for each value.
3. The dataflow dependencies flow from top to bottom, so it makes sense to put the
   operands on top and the results on the bottom.
4. The viewer's eye can naturally follow the edge upwards to match the value number to its type.

The graphviz documentation notes some limitations with record-based nodes, and
recommends switching to the HTML-like syntax. I'm sticking with record-based
nodes for now since it's less verbose and I don't need any of the newer features,
but it might be worth switching to the HTML-like syntax in the future.

[github-actions]

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

[weiweichen]

this looks awesome, could you please add some documentations for this change (either in this PR or a follow-up one)?

[ehein6]

this looks awesome, could you please add some documentations for this change (either in this PR or a follow-up one)?

This is an existing pass, I just changed the output.

Example: mlir-opt --view-op-graph="print-attrs=1 print-result-types=1 max-label-len=80" mlir/test/Transforms/print-op-graph-back-edges.mlir 2> output.dot

--view-op-graph                                        -   Print Graphviz visualization of an operation
        --max-label-len=<uint>                               - Limit attribute/type length to number of chars
        --print-attrs                                        - Print attributes of operations
        --print-control-flow-edges                           - Print control flow edges
        --print-data-flow-edges                              - Print data flow edges
        --print-result-types                                 - Print result types of operations

[MaskRay]

Agreed that a rectangle shape utilizes the image size better :) The generated image files are also usually smaller.
Have you considered the non-fancy "box"?

[ehein6]

Agreed that a rectangle shape utilizes the image size better :) The generated image files are also usually smaller. Have you considered the non-fancy "box"?

Using the "Mrecord" shape turns on support for input and output ports, which this patch uses extensively. Without them, edges attach to any part of the node instead of matching up with an input operand. I could switch to "record" to bring back the square corners.

To make this work with "box", I would need to switch to HTML-like labels. But the render would look the same.

[github-actions]

@ehein6 Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!