Repository navigation
Keep the manual handshake working where no activity kind is required - #172
Merged
Merged
Conversation
PC sampling and SASS safe mode require no CUPTI activity kind, so the required-activity tally recorded nothing and reported activity_not_enabled. Readiness then failed and GpuFlightInitializeInjectionAfterCuda returned 0 in those passes, which the Rust and Python hosts treat as fatal. The tally now becomes ready once the required set was enabled, even when the set is empty; a recorded failure still latches and is reported first. The default initialization deadline goes from 15 s to 60 s: before this release nothing bounded gpufl::init, and a slow but healthy init (PerfWorks setup on a large GPU) must not fail a manual handshake or be killed by the launcher watchdog.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
#171 made
GpuFlightInitializeInjectionAfterCudareturn 1 only when every required CUPTI activity kind was enabled. PC sampling and SASS safe mode require no activity kind, becauseKernelLaunchHandlerandMemTransferHandlerreturn{}when kernel events are not collected. The required-activity tally therefore recorded nothing and reportedactivity_not_enabled, readiness failed, and the handshake returned 0 in those passes.The Rust bootstrap and the backend's
WorkbenchPythonBootstraptreat anything but 1 as fatal. Every PcSampling and SassMetrics pass of a Rust or Python lab would have aborted. This was found while preparing v1.3.0rc5, before any release.Change
RequiredActivityTallygetsEvaluated(), called once after the start-time enable loop, even when the required set is empty.Ready()reports a recorded failure first (failures still latch), and otherwise requiresEvaluated().GPUFL_INJECT_LAUNCH_WAIT_MS) goes from 15 s to 60 s. Before Gate the first CUDA call on checked capture readiness (Linux) #171 nothing boundedgpufl::init. A slow but healthy init, such as PerfWorks setup on a large GPU, must not fail a manual handshake or be killed by the launcher watchdog. Init times will be measured in the L4 requalification.Tests
Evaluated().🤖 Generated with Claude Code