Full documentation for ROCgdb is available at rocm.docs.amd.com/rocgdb
- 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
- The
sysroot
orsolib-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.
- 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.
- '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.
- Add support for gfx1100, gfx1101, and gfx1102.
- Added support for gfx940, gfx941 and gfx942.
- 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.
- Support for watchpoints on scratch memory addresses.
- Improved performances when handling the end of a process with a large number of threads.
-
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.