TheRock/debug-tools/rocgdb/source/CHANGELOG_AMD.md

Changelog for ROCgdb

Full documentation for ROCgdb is available at rocm.docs.amd.com/rocgdb

ROCgdb-X for ROCm-next

Added

• Dumping core of AMD GPU programs with the "gcore" command is now significantly faster, particularly for kernels that use small amounts of VRAM.

ROCgdb-16-3 for ROCm-7.13

Added

• Support for the gfx1153 architecture.

ROCgdb-16-3 for ROCm-7.11

Added

• Support for the HIP language and its built-in variables:
  • threadIdx
  • blockIdx
  • blockDim
  • gridDim
  • warpSize

ROCgdb-16-3 for ROCm-7.2

Added

• GDB now determines the name of AMD GPU threads based on the name of their kernel function.

ROCgdb-16-3 for ROCm-7.0

Added

• Support for the following architectures:
  • gfx950
  • gfx1150
  • gfx1151
• Support for FP4, FP6 and FP8 micro-scaling (MX) data types with the gfx950 architecture.

Removed

• Support for the gfx940 and gfx941 architectures.

ROCgdb-15.2 for ROCm-6.4

Added

• Support for debugging shaders compiled for the following generic targets:
  • gfx9-generic
  • gfx9-4-generic
  • gfx10-1-generic
  • gfx10-3-generic
  • gfx11-generic
  • gfx12-generic

ROCgdb-15.2 (for ROCm-6.3)

Added

• Support for gfx1200 and gfx1201 architectures.
• Support for precise ALU exception reporting for supported architectures. Precise ALU exceptions reporting is controlled with the following commands:
  • set amdgpu precise-alu-exceptions
  • show amdgpu precise-alu-exceptions

Changed

• The sysroot or solib-search-path settings can now be used to locate files containing GPU code objects when opening a core dump. This allows opening GPU code objects on systems different from the one where the core dump was generated.

Resolved issues

• Fixed possible hangs when opening some AMDGPU core dumps in ROCgdb.
• Addressed cases where the roccoremerge utility improperly handled LOAD segment copy from the host core dump to the combined core dump.

ROCgdb-14.2 (for ROCm-6.2)

Added

• 'info agents' now prints the agent location as "DDD:BB:DD.F", where "DDDD" is the agent's PCI domain.
• Introduce the coremerge utility to merge a host core dump and a GPU-only AMDGPU core dump into a unified AMDGPU corefile.
• Support for generating and opening core files for heterogeneous processes.

ROCgdb-13.2 (for ROCm-6.0)

Added

• Add support for gfx1100, gfx1101, and gfx1102.
• Added support for gfx940, gfx941 and gfx942.

Known Issues

• ROCgdb cannot debug a program on an AMDGPU device past a s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) instruction. If an exception is reported after this instruction has been executed (including asynchronous exceptions), the wave is killed and the exceptions are only reported by the ROCm runtime.

ROCgdb-13.2 (for ROCm-5.7)

• Support for watchpoints on scratch memory addresses.

ROCgdb-13.1 (for ROCm-5.6)

Optimized

• Improved performances when handling the end of a process with a large number of threads.

Known Issues

• On certain configurations, ROCgdb can show the following warning message:

  warning: Probes-based dynamic linker interface failed. Reverting to original interface.

  This does not affect ROCgdb's functionalities.