Repository navigation
Add bounded portal presentation performance smoke mode - #709
Merged
Merged
Conversation
This file contains hidden or 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add an explicit
--presentation-performancemode to the real portal example. It disposes the five-tick setup before five warmup cycles and twenty samples of 1,200 actual interpolation/projection views. Every frame is retained and checked outside timing. Raw stopwatch, process CPU/GC and current-thread allocation observations include per-view normalization and workload/assembly identity.One admitted fresh offline seven-package receiver at source
13a8cdebpassed unchanged locked restore, assets/nuspec checks, Release build, CLI 0/0/2 and the explicit smoke mode. All twenty raw samples and 24,000 views are retained; elapsed median was 886.82 ns/view with its outliers preserved. These remain observations of that exact source/assembly, with no historical comparison, timing threshold, steady-state or release verdict.Seven native jobs (Ubuntu/Windows property, full-test and deterministic builds, plus surface-baseline) failed the same nullable environment-variable guard during compilation, before property tests or baseline refresh. Refine through
Option.ofObjbefore string access. Local compilation now explicitly matches null checking, latest language and warnings-as-errors, and six compiled refusal controls pass. The measured computational body is byte-identical to13a8cdeb; no measurement was repeated and no baseline files changed. First local and native failures are retained.Historical baseline is unknown; P3 performance release and fresh/retained product adoption remain open. Original resume04/resume05 custody and cleanup measurement limits are unchanged. Parent owns native rerun and remote integration.