The pre-warming feature was removed to simplify the audio input supervisor implementation. This feature added complexity and was not providing significant latency improvements in practice.
Add LSB depth parameter for improved bit allocation and disable MMAP access for compatibility.
Adjust buffer sizing logic to better handle constrained environments while maintaining stability.
- Replace direct atomic updates with sampling to reduce contention
- Simplify metrics tracking by removing buffering and using direct updates
- Optimize logging by adding level checks and sampling
- Improve validation performance using cached config values
- Replace CGO function variable aliases with direct function calls to eliminate indirection
- Simplify audio frame validation by using cached max size and removing error formatting
- Optimize buffer pool operations by removing metrics collection and streamlining cache access
- Improve batch audio processor by pre-calculating values and reducing config lookups
- Streamline IPC message processing with inline validation and reduced error logging
Add mutex locking around config cache expiration checks to prevent race conditions. The cache now properly checks initialization status before attempting updates.
- Skip logging in frame validation to reduce overhead
- Only update cache when expired to avoid unnecessary operations
- Remove duplicate config caching system and simplify buffer handling
- Optimize batch processing with pre-allocated buffers and conditional time tracking
- Add PCM buffer pool and config for optimized decode-write operations
- Implement separate buffer handling in CGO audio processing
- Update batch processor to support both legacy and optimized paths
- Add batch write processing functionality to match existing read processing
- Improve thread pinning logic with separate controls for read/write
- Add new batch processing configuration parameters
- Update build tags to exclude arm architecture
Add atomic fields to AudioConfigCache for validation parameters to enable lock-free access
Optimize validation functions to use cached values for common cases
Move AudioFrameBatch to separate file and update validation logic
Add SizedBufferPool for efficient memory management with size tracking and statistics
Introduce AudioConfigCache to minimize GetConfig() calls in hot paths
Add batch processing support for audio frames to reduce CGO overhead
Extend AudioBufferPoolDetailedStats with total bytes and average size metrics
Add goroutine pool implementation to manage reusable workers for audio processing tasks
Add configuration constants for pool sizing and behavior
Modify audio server components to use pool for goroutine management
Add fallback to direct goroutine creation when pools are full
* Clean up Virtual Keyboard styling
Make the header a bit taller
Add keyboard layout name to header
Make the detached keyboard render smaller key text so you can read them.
Updated the settings text for keyboard layout.
* Add the key graphics and missing keys
* style(ui): add cursor-pointer class to Button component for better UX
* refactor(ui): Improve header styling and detach bug
- Remove unused AttachIcon and related SVG asset.
- Replace icon usage with a styled LinkButton to improve consistency.
- Simplify and reformat VirtualKeyboard component for better readability.
* refactor(ui): Hide keyboard layout settings on mobile and fix minor styling
---------
Co-authored-by: Marc Brooks <IDisposable@gmail.com>
Add more robust process state checking and error handling during audio input server shutdown. Use signal 0 to verify process existence before killing and handle various edge cases. Also improve logging to better track shutdown outcomes.
Add detailed logging when sending Opus configuration to audio input subprocess
Include supervisor running status in log when subprocess is not connected
Add support for dynamic Opus encoder configuration updates without requiring subprocess restart. This allows quality changes to be applied immediately while maintaining audio stream continuity.
Optimize audio quality parameters to reduce CPU load and prevent mouse lag on RV1106 devices. Lower bitrates and complexity while adjusting signal types and bandwidths for better performance.
Add build tags for CGO requirements and improve audio input supervisor behavior to check for existing processes before starting new ones.
Implement periodic cleanup of stale goroutine buffer caches to prevent memory leaks
Add ability to detect and reuse existing audio input server processes
Adjust bitrates, frame sizes, and OPUS parameters to balance quality and performance on RV1106. Reduce channel count for low quality to minimize CPU load. Update CGO constants for better memory efficiency.
add audio input supervisor with opus configuration support
update audio quality presets and configuration handling
restructure audio subprocess management with environment variables
* chore/Deprecate browser mount
No longer supported.
* Remove device-side go code
* Removed diskChannel and localFile
* Removed RemoteVirtualMediaState.WebRTC
Also removed dead go code (to make that lint happy!)
Introduce MetricsRegistry to serve as single source of truth for audio metrics
Remove duplicate metrics collection logic from web endpoints
Add callback mechanism for metrics updates
This commit removes the granular latency metrics collection and histogram visualization functionality across the codebase. The changes include:
- Removing latency histogram tracking from audio input/output processing
- Removing latency histogram UI components and related types
- Removing granular metrics collector's latency tracking capabilities
- Updating Prometheus metrics to use milliseconds instead of seconds
- Removing related tests and benchmarks
Add granular metrics collection for audio processing latency in both input and output paths. This enables better performance monitoring through histograms.
Also update build tags to include ARM platform and rename UI label for clarity.
- Add LatencyHistogramData interface and implement histogram collection in granular metrics
- Create LatencyHistogram component for visualizing latency distribution
- Update audio metrics events to include histogram data
- Add comprehensive tests for histogram functionality
- Improve error handling for OPUS encoder parameter updates
- Optimize validation cache initialization
Split generic frame data validation error into specific cases for empty and oversized frames to provide better error context. The optimized validation maintains performance while improving debuggability.
Consolidate audio frame validation functions into a single optimized implementation and add dynamic OPUS encoder parameter updates based on quality settings. Initialize validation cache at startup for consistent performance.
Add latency profiler for end-to-end audio pipeline monitoring. Update test cases to use unified validation function and initialize cache.
The changes improve performance by reducing function call overhead and enabling runtime optimization of audio encoding parameters based on quality presets.
Alex P
Aveline
Marc Brooks
dependabot[bot]
Adam Shiervani