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

SWDEV-193956 - [hipclang-vdi-rocm][perf]

Browse Source 
~45% to 50% of Performance drop on rocBLAS_int8 test

Use the last command in the queue for a wait.
Add extra print information about processed commands.
Add an option to disable file location printing.

Change-Id: I4187883e1a90e571fde3128af98368108fda8785
This commit is contained in:
German Andryeyev
2020-02-20 16:24:10 -05:00
parent fea14a7201
commit a66d09f5a3
3 changed files with 36 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/utils/debug.hpp
+6 -1
View File
@@ -47,6 +47,7 @@ enum LogMask {
LOG_AQL2 = 0x00002000, //!< Show raw bytes of AQL packet
LOG_CODE = 0x00004000, //!< Show code creation debug
LOG_CMD2 = 0x00008000, //!< More detailed command info, including barrier commands
LOG_LOCATION = 0x00010000, //!< Log message location
LOG_ALWAYS = 0xFFFFFFFF, //!< Log always even mask flag is zero
};
@@ -178,7 +179,11 @@ inline void warning(const char* msg) { amd::report_warning(msg); }
do { \
if (LOG_LEVEL >= level) { \
if (GPU_LOG_MASK & mask || mask == amd::LOG_ALWAYS) { \
amd::log_printf(level, __FILE__, __LINE__, format, ##__VA_ARGS__); \
if (GPU_LOG_MASK & amd::LOG_LOCATION) { \
amd::log_printf(level, __FILE__, __LINE__, format, ##__VA_ARGS__); \
} else { \
amd::log_printf(level, "", 0, format, ##__VA_ARGS__); \
} \
} \
} \
} while (false)