corrected the gpu-id in the profiler record for MI300 individual xcc mode
Change-Id: Iec0163a67dd28b556f73ef7ff376d6d7040ba4c5
Este commit está contenido en:
cometido por
Ammar ELWazir
padre
7ed8e6f37b
commit
b72fa8b125
@@ -212,6 +212,7 @@ The resulting `a.out` will depend on
|
||||
### Added
|
||||
- 'end_time' need to be disabled in roctx_trace.txt
|
||||
- support for hsa_amd_memory_async_copy_on_engine API function trace
|
||||
|
||||
### Fixed
|
||||
- rocprof in ROcm/5.4.0 gpu selector broken.
|
||||
- rocprof in ROCm/5.4.1 fails to generate kernel info.
|
||||
@@ -225,6 +226,7 @@ The resulting `a.out` will depend on
|
||||
- MI300 counters support for rocprof v1 and v2.
|
||||
- Limiting file name sizes for ATT plugin.
|
||||
- Support for MI300 XCC modes for rocprof v2.
|
||||
- MI300 individual XCC counters dumped per-xcc as separate records but with same record-id and kernel dispatch info
|
||||
### Fixed
|
||||
- Samples are fixed to show the new usage of phases.
|
||||
- Plugin option validates the plugin names.
|
||||
|
||||
@@ -466,8 +466,7 @@ bool AsyncSignalHandler(hsa_signal_value_t signal_value, void* data) {
|
||||
// TODO: (sauverma) gpu-id will need to support xcc like so- 1.1, 1.2, 1.3 ... 1.5 for
|
||||
// different xcc
|
||||
record.gpu_id = rocprofiler_agent_id_t{
|
||||
(uint64_t)hsa_support::GetAgentInfo(queue_info_session->agent.handle).getIndex() +
|
||||
xcc_id};
|
||||
(uint64_t)hsa_support::GetAgentInfo(queue_info_session->agent.handle).getIndex()};
|
||||
record.kernel_properties = pending.kernel_properties;
|
||||
record.thread_id = rocprofiler_thread_id_t{pending.thread_id};
|
||||
record.queue_idx = rocprofiler_queue_index_t{pending.queue_index};
|
||||
|
||||
Referencia en una nueva incidencia
Block a user