[SDK][rocprofv3] Buffer tracing records with args (HIP) (#285)

* [SDK][rocprofv3] HIP API buffer records with args (ext)

- New buffer tracing domain(s) for HIP APIs which include the arguments and the return value in the buffer records
- Update HIP stream support for extended HIP buffer tracing
- Update rocprofv3 tool library and output library to use extended HIP buffer tracing recods

* Update stream.cpp

- handle hipStream_t address being reused for a new stream

* Update doxygen docs for rocprofiler_iterate_buffer_tracing_record_args

* Update rocprofv3 tool.cpp

- configure buffer tracing services with HIP_*_API_EXT variants
- tweak logging level for hip_stream_display_callback

* Fix validation tests

- add HIP_RUNTIME_API_EXT and HIP_COMPILER_API_EXT to valid domain names

* Serialization support for buffer tracing args

* Disable stream service for __hipPopCallConfiguration

- this is interpreted as a stream create but it doesn't create a stream

* Fix execute_buffer_record_emplace for HIP extended contexts

* Add uint64_t_retval to rocprofiler_hip_api_retval_t union

- reading in hipError_t_retval during serialization of pointer return value causes undefined behavior

* Fix compilation warning about unused but set parameter

- in hip/stream.cpp

* Add synchronization for async_copy_data

* Fix compilation error

* Fix compilation error

---------

Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
Этот коммит содержится в:
Madsen, Jonathan
2025-03-22 19:57:32 -05:00
коммит произвёл GitHub
родитель 2d072f9217
Коммит e33dff7ad0
30 изменённых файлов: 548 добавлений и 118 удалений
+1 -1
Просмотреть файл
@@ -138,7 +138,7 @@ buffered_output<Tp, DomainT>::destroy()
}
using hip_buffered_output_t =
buffered_output<rocprofiler_buffer_tracing_hip_api_record_t, domain_type::HIP>;
buffered_output<rocprofiler_buffer_tracing_hip_api_ext_record_t, domain_type::HIP>;
using hsa_buffered_output_t =
buffered_output<rocprofiler_buffer_tracing_hsa_api_record_t, domain_type::HSA>;
using marker_buffered_output_t =
+4 -4
Просмотреть файл
@@ -316,10 +316,10 @@ generate_csv(const output_config& cfg,
}
void
generate_csv(const output_config& cfg,
const metadata& tool_metadata,
const generator<rocprofiler_buffer_tracing_hip_api_record_t>& data,
const stats_entry_t& stats)
generate_csv(const output_config& cfg,
const metadata& tool_metadata,
const generator<rocprofiler_buffer_tracing_hip_api_ext_record_t>& data,
const stats_entry_t& stats)
{
if(data.empty()) return;
+4 -4
Просмотреть файл
@@ -46,10 +46,10 @@ generate_csv(const output_config& cfg,
const stats_entry_t& stats);
void
generate_csv(const output_config& cfg,
const metadata& tool_metadata,
const generator<rocprofiler_buffer_tracing_hip_api_record_t>& data,
const stats_entry_t& stats);
generate_csv(const output_config& cfg,
const metadata& tool_metadata,
const generator<rocprofiler_buffer_tracing_hip_api_ext_record_t>& data,
const stats_entry_t& stats);
void
generate_csv(const output_config& cfg,
+1 -1
Просмотреть файл
@@ -187,7 +187,7 @@ write_json(json_output& json_ar,
const output_config& /*cfg*/,
const metadata& /*tool_metadata*/,
const domain_stats_vec_t& domain_stats,
generator<rocprofiler_buffer_tracing_hip_api_record_t>&& hip_api_gen,
generator<rocprofiler_buffer_tracing_hip_api_ext_record_t>&& hip_api_gen,
generator<rocprofiler_buffer_tracing_hsa_api_record_t> hsa_api_gen,
generator<tool_buffer_tracing_kernel_dispatch_with_stream_record_t> kernel_dispatch_gen,
generator<tool_buffer_tracing_memory_copy_with_stream_record_t> memory_copy_gen,
+1 -1
Просмотреть файл
@@ -85,7 +85,7 @@ write_json(json_output& j
const output_config& cfg,
const metadata& tool_metadata,
const domain_stats_vec_t& domain_stats,
generator<rocprofiler_buffer_tracing_hip_api_record_t>&& hip_api_gen,
generator<rocprofiler_buffer_tracing_hip_api_ext_record_t>&& hip_api_gen,
generator<rocprofiler_buffer_tracing_hsa_api_record_t> hsa_api_gen,
generator<tool_buffer_tracing_kernel_dispatch_with_stream_record_t> kernel_dispatch_gen,
generator<tool_buffer_tracing_memory_copy_with_stream_record_t> memory_copy_gen,
+1 -1
Просмотреть файл
@@ -360,7 +360,7 @@ write_otf2(
const metadata& tool_metadata,
uint64_t pid,
const std::vector<agent_info>& agent_data,
std::deque<rocprofiler_buffer_tracing_hip_api_record_t>* hip_api_data,
std::deque<rocprofiler_buffer_tracing_hip_api_ext_record_t>* hip_api_data,
std::deque<rocprofiler_buffer_tracing_hsa_api_record_t>* hsa_api_data,
std::deque<tool_buffer_tracing_kernel_dispatch_with_stream_record_t>* kernel_dispatch_data,
std::deque<tool_buffer_tracing_memory_copy_with_stream_record_t>* memory_copy_data,
+1 -1
Просмотреть файл
@@ -40,7 +40,7 @@ write_otf2(
const metadata& tool_metadata,
uint64_t pid,
const std::vector<agent_info>& agent_data,
std::deque<rocprofiler_buffer_tracing_hip_api_record_t>* hip_api_data,
std::deque<rocprofiler_buffer_tracing_hip_api_ext_record_t>* hip_api_data,
std::deque<rocprofiler_buffer_tracing_hsa_api_record_t>* hsa_api_data,
std::deque<tool_buffer_tracing_kernel_dispatch_with_stream_record_t>* kernel_dispatch_data,
std::deque<tool_buffer_tracing_memory_copy_with_stream_record_t>* memory_copy_data,
+1 -1
Просмотреть файл
@@ -68,7 +68,7 @@ write_perfetto(
const output_config& ocfg,
const metadata& tool_metadata,
std::vector<agent_info> agent_data,
const generator<rocprofiler_buffer_tracing_hip_api_record_t>& hip_api_gen,
const generator<rocprofiler_buffer_tracing_hip_api_ext_record_t>& hip_api_gen,
const generator<rocprofiler_buffer_tracing_hsa_api_record_t>& hsa_api_gen,
const generator<tool_buffer_tracing_kernel_dispatch_with_stream_record_t>& kernel_dispatch_gen,
const generator<tool_buffer_tracing_memory_copy_with_stream_record_t>& memory_copy_gen,
+1 -1
Просмотреть файл
@@ -40,7 +40,7 @@ write_perfetto(
const output_config& cfg,
const metadata& tool_metadata,
std::vector<agent_info> agent_data,
const generator<rocprofiler_buffer_tracing_hip_api_record_t>& hip_api_gen,
const generator<rocprofiler_buffer_tracing_hip_api_ext_record_t>& hip_api_gen,
const generator<rocprofiler_buffer_tracing_hsa_api_record_t>& hsa_api_gen,
const generator<tool_buffer_tracing_kernel_dispatch_with_stream_record_t>& kernel_dispatch_gen,
const generator<tool_buffer_tracing_memory_copy_with_stream_record_t>& memory_copy_gen,
+2 -2
Просмотреть файл
@@ -83,8 +83,8 @@ generate_stats(const output_config& /*cfg*/,
stats_entry_t
generate_stats(const output_config& /*cfg*/,
const metadata& tool_metadata,
const generator<rocprofiler_buffer_tracing_hip_api_record_t>& data)
const metadata& tool_metadata,
const generator<rocprofiler_buffer_tracing_hip_api_ext_record_t>& data)
{
auto hip_stats = stats_map_t{};
for(auto ditr : data)
+3 -3
Просмотреть файл
@@ -37,9 +37,9 @@ generate_stats(const output_config& cfg,
const generator<tool_buffer_tracing_kernel_dispatch_with_stream_record_t>& data);
stats_entry_t
generate_stats(const output_config& cfg,
const metadata& tool_metadata,
const generator<rocprofiler_buffer_tracing_hip_api_record_t>& data);
generate_stats(const output_config& cfg,
const metadata& tool_metadata,
const generator<rocprofiler_buffer_tracing_hip_api_ext_record_t>& data);
stats_entry_t
generate_stats(const output_config& cfg,