- Added client autoupdate when running as linux systemd service
- Bugfixes
- Optimized legacy config reader
- Added hybrid/dual mode
- Removed legacy constructs (Qiner, Custom runner and more)
- Bugfixing
- Added support for skylake in modern environment
- Optimized binary file
- Added support for idling in classic environment
- Added PPS+ Support
- Added Live Connection Support
- Added Idle Task Support
- Fixed AVX2 auto detection
- Fixed a bug where trainer couldn't get a task (object reference not set...)
- Re-Added CPU autodetection for new trainer config scheme (Auto AVX2/AVX512)
- Added GPU Cards Feature (select which GPU Cards to use and how many threads or let the client auto tune them)
- Minor Bugfixes
- Native AMD GPU support
- Added new options to select your trainer
- Prepared for combined CPU+GPU training
skipped version
- fixed windows trainer change problem
- optimized trainer handling
- added possibility to configure cpu afinity
- internal optimizations
- fixed cases when the client crashed unexpected
- changed internal solution processing
- added auto re-register of client if registration the first time fails (lead to 'waiting for task' loop)
- fixed some minor bugs
- added ability to resend unsubmitted solution after a client restart
- Log time set to UTC
- added graceful handling of runner
- update to .net 8
- optimized logging of found solutions during epoch
- Fixed wrong "skylake" overwrite for linux client
- Optimized Epoch Change
- Added new Output line for GPU Trainers
GPU #0: 870 it/s | GPU #1: 884 it/s | GPU #2: 870 it/s | GPU #3: 872 it/s
to have this information available in hiveos
- Fixed a bug in calculating local found solutions. (e.g. wrong "SOL: 12/300" display)
- Fiexd a reporting bug
- Added experimental display of epoch to console output
- Fixed a bug in writing
stats.EPOCH.lock
file
- Added optimized console logging. (new format: "SOL: X/Y" where Y is found solutions and X is submitted solutions)
- GPU Release
- Fixed Bug where solution finding was delayed
- Changed main communication Scheme from qiner to qli internal
- Windows Release
- Bugfixes
- UPgrade from .Net 6 to 7
- Added Intel Skylake detection for Linux
- Downgrade from .Net 7 to 6
- Removed AVX2 Fallback for Linux
- Optimized Network Traffic Handling
- Significant performance boost
- Fixed Bug in Solution Delivery in certain environments
- Added compatibility for new Qubic protocol
- Fixed potential Bug in Solution Delivery
- Added http/2 support
- Improved API Handling
- General Bugfixing
- Added/Updated Qiner Protocol
- fixed bug on Windows with stopping to report status
- automatic fallback to AVX2 on linux
- DEPRECATED: custom QLI Protocol
- fixed bug with Qiner communication protocol
- fixed bug which decreases training performance over time
- Optimized Communication between Client and API
- Fixed blocking Bug
- Deprecated:
"AVX2":true
setting - Optimized Detection of AVX512 and added Fallback to AVX2
- Removed Deprecated Settings
- Added Setting
AllowOptimization
to allow the trainer to try to optimize environment. (Default: false) - Reduced Runner Footprint
- Bugfixing
- Automatic optimization of the environment
- Bugfixing
- Possibility to Overwrite Params like AVX512
- new argument
--version
to get current version of Client
- Bugfixing
- New Setting:
"serviceLock": true
to use qiner protocol for custom runner - New Log Format with display of average it/s
- Signaling if the trainer isn't starting properly
- Fixed AVX512 detection on Windows
- Fixed Thread Settings from Portal
- Automatic runner selection from Portal Settings
- Automatic thread selection from Portal Settings
- New Setting:
"useAvx": true
which forces the client to use AVX2 runner - New Setting:
"allowHwInfoCollect": true
which will allow the client to collect CPU Model, RAM Size and CPU Cache Size; with this information the client can better decide which runner should be used - Optimized Detection of AVX2/512
- Optimized Error Handling
- Added automated selection of AVX2/512 runner
- New Setting:
"customRunner": true
to enforce a custom runner