core(lantern): divide throughput only on network node count #14564
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
paulirish
approved these changes
Nov 28, 2022
brendankenny
commented
Nov 28, 2022
| for (const connection of this._connectionPool.connectionsInUse()) { | ||
| connection.setThroughput(this._throughput / this._nodes[NodeState.InProgress].size); |
There was a problem hiding this comment.
for the historical record: the reason it did this was because the original prototype graph only had network nodes, not cpu ones. @paulirish actually noted at the time that we should make sure not to keep including all active nodes when cpu nodes were added to the graph, but we did not remember.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes a bug in how the simulator divides throughput across network nodes. Previously it was counting any in-progress CPU tasks in the number of requests currently being made.
Since the simulator only simulates a single thread this was only ever off by (at most) 1, but that can be significant, especially in early page load where the request count is usually lower (e.g. if two active requests and a busy CPU, each request is allocated 1/3 of the available throughput instead of 1/2).
Looking at the updated tests gives a good overview: every simulated load metric gets a little faster, to varying degrees. The tests may show more pronounced improvements than many real pages because they tend to have fewer requests.
Impact on lantern test accuracy is mixed, but, the current thing is incorrect, so... :)