nvtop-3.0.2+2-bp156.1.3

- Update to version 3.0.2+2:
  * Fix a potential issue when using multiple AMDGPUs.
  * Fix amdgpu wrong tx/rx computation
  * Remove workaround for driver bug
  * Uniqueness is pdev+id on amdgpu
  * Uniqueness is pdev+id on intel
  * README MSM information.
  * MSM and Adreno support

- Update to version 3.0.1+29:
  + Fix typo in pci id.
  + Add pre-parsed hardware info.
  + Update copyrights.
  + Fix monitored gpu selection logic.
  + Fix multiple gpus selection.
  + Fix a unused case.
  + Add license.
  + Add copyright info.
  + New gpus support using a local gpu type database.
  + hjkl motion keys.
- Add %%doc README.markdown.

- Update to version 3.0.1:
  * Don't overwrite usage for graphics+compute processes.
  * Expose processes that are both graphical and compute.
  * Fix nvtop process hiding.
  * Move misplaced assertion.
- Changes from version 3.0.0:
  * Update documentation.
  * Option to hide nvtop from the process list.
  * Clang-format all.
- Changes from 2.0.3+4:
  * Option to toggle startup messages.
  * Don't print the infos to stdout.
  * Fix spacing.
  * Save message option and fix multidevice save.
  * Support info window.
  * Support info messages.
  * Save monitored gpus to config.
  * Setup window to select gpus.
  * Dynamic gpu monitored list.
  * Rework interface for gpu select option.
  * Remove GPU mask option.
  * Extract pdev for all devices.
  * Fix for libudev wrapper.
  * Readme update for Intel support.
  * Don't need device discovery for NVIDIA only.
  * Fix FindUDev and FindSystemd for older cmake.
  * Show intel ENC/DEC together.
  * Fix integrated AMDGPU drm and fdinfo usage mismatch.
  * Fix macro extra arguments.
  * Mark AMDGPU integrated gpu as such.
  * Fix amdgpu null pointer deref.
  * Fix intel and integrated GPU support.
  * Use device discovery for Intel.
  * Use device discovery for AMDGPU.
  * Udev/sd-device wrapper.
  * Fix intel support.
  * Preliminary Intel support.
  * nvtop metainfo file.
  * Auto date for manpage.
  * Nvtop logo.
  * add process-control plug to allow killing processes.
  * Guard against huge jump in usage.
  * Uint64 literals.
  * Guard against suspect values reported by fdinfo.
  * Better validity macros.
  * Mark valid AMDGPU cache entries.
  * AMDGPU compute process support.
  * Add desktop file.
  * Code reorganization to sweep /proc once per update.

- Update to version 2.0.3+3:
  * Enable format warnings.
  * Fix format-security warnings.
  * Fix drm node file descriptor opening logic.
  * Fix fdinfo name typo.
  * Remove unnecessary list and free amdgpu resources.
  * Support new amdgpu fdinfo interface kernel >=5.19.
  * Display time on x axis.
  * Resolve some warning messages.

- Update to version 2.0.2:
  * Fix encode/decode window cleaning on AMDGPU.
  * Fix fast resize event not registering.
  * Put expensive tests behind an option.
  * Fix plot for a maximum of 64 GPUs.
  * Expensive test at the end.
  * Add test for infinite loop fix.
  * Fix wrong size UCHAR_MAX -> UINT_MAX.
  * Test for expected empty process window.
  * Don't redistribute when no rows are left.
  * Header stack of minimum 1 to avoid divide by zero.
  * More interface tests.
  * Resolve compilation warnings.
  * C++ friendly header for testing.
  * Don't build testing by default unless in Debug.
  * Simplified and fixed plot merge algorithm.
  * Expose max lines in a plot as a macro.
  * Enable CMake testing using GTest.
  * Add assert in info_in_plot.

- Update to version 2.0.1+12:
  * Project wide allocarray availability check.
  * Metrics reset cleanup.
  * Fix AMDGPU encoder value stored in decoder usage.
  * Update encode and decode rates for AMDGPU.
  * Alternative path to get GPU usage from process info.
  * Linear realloc strategy.
  * realloc to reallocarray to catch potential overflows.
  * Check for reallocarray availability.

- Update to version 2.0.1+2:
  * docs(README): fix typo (querry -> query).
  * Use pp_dpm_pcie to get the current PCIe link configuration.
  * Guard GPU family to support older libdrm.
  * Bump initial major release with AMDGPU support.
  * Disable over-verbose warnings and enable useful ones.
  * Fix goto label not preceding a statement.
  * Dev tooling improvement.
  * Update README and manpage with AMDGPU support info.
  * Use libdrm for AMDGPU support.
  * AMDGPU power cap info.
  * AMDGPU max temps and PCIe gen/width and usage.
  * Fan speed extraction from sysfs.
  * Don't close negative file descriptors and fix error check.
  * Convert AMDGPU watts to the datastructure milliwatts.
  * Reset return status when handle or function are not loaded.
  * Get the right gpu name in the setup window.
  * extract_gpuinfo_amdgpu: Support per-process GPU usage.
  * Add basic global (not per-process) GPU usage for AMDGPU.
  * Prepare mask for inclusion of multiple vendors.
  * Make GPU vendors use a hooking mechanism.
  * Make devices into a linked list and make use container_of.
  * typedef struct foo -> struct foo.
  * extract_gpuinfo_nvidia: Get rid of dlsym cast.
  * optimization_flags.cmake: Disable quite a few warnings.

- Update to version 1.2.2+5:
  * Remove printf statement.
  * Use the maximum of graphical and sm as GPU speed.
  * Fix: filter out nvidia utilization samples.

- Update to version 1.2.1:
  * Remove TODO file.
  * Line plot algorithm rewrite.
  * Call most specific ncurses function.
  * Rename weirdly named function.
  * Clear only the updated part of the process window.
  * Merge process related options to process logic.
  * Separate the shortcut window logic from process and setup.
  * end/dec: clear the windows only when needed.
  * Make changes to please clang-tidy in some cases.
  * Fix: improve backward compatibility for old drivers.
  * Fix: potential buffer overflow on process memory buffer.
  * Fix: search in the two pid info caches for a match.
  * Fix: not assuming at least one process for pid selection.
  * Up-to-date screenshots and README.

- Update to version 1.1.0+34:
  * All the charts have the same scale.
  * Fix: chart layout now correctly handles multiple GPUs.
  * Bump nvtop version to 1.2.
  * Integration of per process utilization to the interface.
  * NVIDIA per process utilization info retrieval.
  * Add the process list metric options to the setup window.
  * Add a notice about the maximum number of metrics per GPU.
  * Add the GPU/MEM clock rates to the list of metrics that can be
    plotted.
  * Add the fan speed to the list of metrics that can be plotted.
  * Add the power draw rate to the list of metrics that can be
    plotted.
  * Add the a no-modification notice to the generated config file.
  * Reduce legend size.
  * Update README and manpage.
  * Fix: the memory usage was initialized with a wrong metric.
  * Update interface codebase.
  * Use a ring buffer to store the chart data.
  * Fix: encode/decode now correctly cleared and displayed.
  * Consistency size reduction of menu shortcuts.
  * Add the update interval as an option.
  * Setup window implementation.
  * Configuration file support.
  * Redisign of program option handling.
  * Update README and doc following the gpuinfo redesign.
  * Build system update following the gpuinfo redesign.
  * Update interface and main compilation units to use the new
    gpuinfo functions.
  * Processus info field rename and add missing header.
  * Per vendor GPU extraction.
  * Bump uthash version to 2.3.0.
  * Fix: /proc/<pid>/cmdline may be empty so initialize the string
    as empty.
  * More paths for NVML library.
  * Fix: initialize_device_info failed cause double free.

- Update to version 1.1.0+3:
  * Interface: plot refresh rate respects the -d option.
  * Fix: sorting function relied on integer conversion.
  * Add additional path hints for nvml.

- Update to version 1.1.0:
  * Bump version 1.1.0
  * README.markdown: add gentoo install commands
- Update _service: Change to @PARENT_TAG@+@TAG_OFFSET@ version
  scheme, with +0 being stripped when detected.

- Update to version 1.0.0+git20201111.aa48ce2:
  * ITF: print power draw or max draw even when both are not
    provided by the driver

- Update to version 1.0.0+git20201012.087a025:
  * Plot: reverse to march other monitoring programs behavior
  * Build: clear cached variable for build check
  * Fix: imprecise information from nvml library
  * Fix: wrong variable was used for column computation
  * Fix: initialize the correct ammount of plot location
  * Build: do not use nvml header automatic api upgrade
  * CMake: add nvml path hints
  * Doc: fix typo and add encoder/decoder info to manpage
  * Fix: increase PID field size supporting linux range
  * Update Arch Linux installation instructions

- Initial build.