AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Optimize rotcx markers

Browse Source 
Improve the roctx markers performance when the tracer is not engaged
(the application is not running with rocprof).

The performance of roctx push/pop, measured with:

-----------------------------------------------------------------------
  auto start = std::chrono::steady_clock::now();
  for (int i = 0; i < 10000000; ++i) {
    roctxRangePush ("A");
    roctxRangePop ();
  }
  auto end = std::chrono::steady_clock::now();
  std::cout << "ns = " << std::chrono::nanoseconds(end - start).count()
      / 10000000 << std::endl;
-----------------------------------------------------------------------

w/o rocprof | with rocprof | commit
       92ns |       770ns  | 0d6e132: Cleanup CallbackTable::Get
       28ns |       712ns  | 6421bd5: Cleanup ROCTX's implementation
       20ns |       664ns  | 7f0e5e5: Remove the roctx range message...
        6ns |       665ns  | this commit

Change-Id: Id679dcbd0fb190a3179be98a9b2c1db151efee3d
This commit is contained in:
Laurent Morichetti
2022-05-06 19:46:09 -07:00
parent 3d0198c395
commit a794247c55
3 changed files with 33 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files
script/hsaap.py
+1 -1
View File
@@ -395,7 +395,7 @@ class API_DescrParser:
# generate API callbacks
def gen_callbacks(self, n, name, call, struct):
if n == -1:
self.content += 'typedef CallbackTable<HSA_API_ID_NUMBER> cb_table_t;\n'
self.content += 'typedef CallbackTable<ACTIVITY_DOMAIN_HSA_API, HSA_API_ID_NUMBER> cb_table_t;\n'
self.content += 'extern cb_table_t cb_table;\n'
self.content += '\n'
if call != '-':