Discussion in 'MSI AfterBurner Application Development Forum' started by Unwinder, Feb 20, 2017.
Confirmed and reproduced, will fix it in the next version.
If that NIC data source is coming from PerfCounter plugin than it is not actually a NIC polling time only but total polling time for all performance counters you've enabled in this plugin. Due to specific of Windows performance counters, they cannot be read independently of each other so collection of all counters is being polled in a single pass. So the first source from this plugin in the profiler is a total polling time for all performance counters and the rest data sources from this plugin are virtually "free" because they are already polled. The same apply to some internal MSI AB data sources, for example both core and memory clock are being read from NV driver in a single NVAP call so the first source (e.g. Core clock) is displaying polling time for both of them and the second one (e.g. Memory clock) is virtually "free" as well if core clock is enabled.
All HwInfo data could be added before as well via editing the CFG and there were instructions on doing so in this thread.
No, you don't need to restart AB, but you need to close application properties and apply changes you did there with "OK" button to show new data sources on the graphs.
CPU is polled directly inside MSI AB, GPU data is read via NV/AMD driver, which takes more time.
Negative on that, sorry, I'd like to see new data source browser window similar to HwInfo's native sensors window, where the nodes are expanded.
There will be one more enhancement in PerfCounter plugin. I'll add new option for enabling dynamic mode for data sources exported by this plugin. PerfCounter plugin is validating performance counters during creation, which means that it will invalidate and it won't display the performance counters that supply no valid data during creation (e.g. HDD4 usage is hidden on the system with less than 4 drives). However, some data sources may report no valid data sometimes, but you still may wish to have them displayed. To achieve that you may enable new dynamic mode for them. Dynamic data sources are not validated during creation, such sources are allowed to export no data until some condition is met (e.g. “CPU usage” performance counters specific to some target process export data only when the process is running)
Well then, maybe make the windows bigger by default and/or resizable? Its a little small for the amount of data in there.
I prefer it as is, sorry.
BTW, I noticed that sensor values column is displaying no units in HwInfo plugin on your system. Currently HwInfo itself behaves this way (i.e. provides blank units info via shared memory to external applications) if you enable "Profiling time" in "Show columns" in HwInfo sensor properties.
4.4.3 Beta 3 is online:
o Added experimental support for Intel i7 and i9 7XXXX LGA2066 CPUs
o Added CPU package power monitoring for Intel CPUs
o Fixed one more bug in HwInfo plugin GUI, which could cause duplicated sensor readings in “Add new data source” window
o Improved PerfCounter plugin. Added support for dynamic data sources. Dynamic data sources are not validated during creation, such sources are allowed to export no data until some condition is met (e.g. “CPU usage” performance counters specific to some target process export data only when the process is running)
o Added new open source CPU.DLL plugin. The plugin fully duplicates MSI Afterburner’s built-in CPU temperature monitoring implementation (with the only exception of AMD Ryzen codepath, which cannot be open sourced due to NDA) and shows third party plugin developers how to implement low-level CPU specific sensors in plugins using CPU MSR registers and PCI configuration space access API
o Now uninstaller is also removing user plugin configuration files when you’re performing full application uninstallation
How does this work? I'm looking around for new options for adding a bunch of counters for the different disks on my home server and it still looks the same.
Say I've got 10 different disk counters for 4 different disks, thats adding and customizing multiple fields for 40 counters doing it in the current GUI would be quite time consuming. I could generate them all once and then open up a notepad and do some copy+paste+modify but a GUI way to speed this up should be doable. For instances in the add counter dropdown add a checkbox to allow multiples being selected and allow for an instance variable of %instance to be put in the counters that it will use to instantiate the multiple counters.
I'm experiencing a bug all of a sudden starting with Beta 2 where multiple instances (more than a dozen) of RTSS Hooks Loader are appearing in task manager and it drives up the disk usage. It prevents loading games too. Uninstalled and tried Beta 3, same thing. Uninstalled and reverted to 4.4.2 Final and same issue persists. I guess it's an issue with 18.3.1?
The only way of doing that in current implementation is adding/configuring full set of required counters for the first disk, then selecting them in the list in main plugin setup window and clicking <Add> while holding <Ctrl> key. This will add exact copy of all selected counters, but you'll need to reconfigure instance for each new cloned counter yourself. Probably it worth adding one more hotkey modifier, which will also add a clone of selected counters but with incremented instance index for each of them.
Please do not use development thread for discussing anything non-related to testing new beta's functionality. No, RTSS hooks loader handshaking principles have not changed for years. No, they are not and cannot be anyway related to display driver version. That's a questions of properly configuring antivirus/system security application. And once again, it is offtopic here so PLEASE avoid discussing it further in development thread. If you wish, create a different thread and probably somebody else besides me can help you to configure your system properly there.
Thanks for this. Tried it on my i7 7800X on MSI X299 SLI Plus running Win10 v1709: clock speed and power values are similar to what Core Temp 1.11 on 2nd screen reports.
Thanks for confirming. Whith new CPU plugin introduced in the last beta it should be also much easier for third party developers to implement support for newer CPUs unsupported by MSI AB core.
4.4.3 beta 4 is available:
Changes list includes the following items comparing to the previous beta:
- Added new keyboard modifiers to "Add new source" window in AIDA/HwInfo/PerfCounter plugins. Now you may hold <Ctrl>+<Shift> when clicking <Add> button to clone currently selected data sources and automatically increment instance indices (if specified) in data source properties. This way you can easily clone the same set of sensors for multiple similar devices, e.g. simply select the set of performance counters specific to HDD4 in PerfCounter plugin then easily create the same set of sensors for HDD5 with single click.
- Now new added data sources are automatically selected in the list of available data sources in AIDA/HwInfo/PerfCounter plugins. Combining with previously mentioned keyboard modifiers, this feature simplify the process of cloning the same sets of sensors for multiple similar devices
- Minimum sampling interval for CPU usage calculation has been reduced from 1000ms to 100ms
- Fixed incorrect tracking markers rendering on the graphs dynamically added to hardware monitoring module
- Fixed incorrect tracking markers rendering when monitoring ring data buffer is full
One more feature from hardcore RivaTuner's hardware monitoring module is coming to MSI AB soon. Original RivaTuner’s monitoring history fragment selection and selected fragment processing functionality is back! Now you may hold <Shift> and press left mouse button on initial desired position in hardware monitoring window to begin fragment selection, then drag mouse cursor to desired final position and release left mouse button to select a fragment. The following functionality is available for selected fragments:
- When the fragment is selected, minimum and maximum displayed values are estimated within selected fragment only. This feature can be used if you don’t need the global minimum and maximum values, estimated during whole monitoring session, and wish to see local minimum and maximum values estimated within some shorter period of time (e.g. during benchmark application runtime only)
- When the fragment is selected you may use the context menu of any graph to set tracking marker into the local minimum or maximum position within the selection. This feature can be used to automate the process of identifying and visualizing the moments of reaching peak values (e.g. peak GPU temperature) on desired fragments
That's pretty cool feature, forgot RT did that.
Some more news on development. During testing and debugging plugins configuration GUI last days I needed something to switch between multiple application configurations frequently. So I added restore points system for my needs. I'm not sure if it will be useful for large user base, but I decided to leave it as power user feature in public version too:
- Added power user oriented restore points system. Experienced users may use new /backup and /restore command line switches to create multiple named restore points containing full backups of MSI Afterburner’s settings (including hardware profiles) and restore settings from any previously created restore point if necessary
I kind of do that now, manual backup profile folder for full reinstalls if needed. That to me would be a good feature for HW reviewers were they are switching vid cards to bench them, might be handy for that.