Windows
Launch application AMDTClassicMatMul.exe and collect the samples for
CYCLES_NOT_IN_HALT and RETIRED_INST events and generate report
C:\> AMDuProfCLI.exe profile -e cycles-not-in-halt -e retired-inst --interval 1000000
-o c:\Temp\cpuprof-custom AMDTClassicMatMul.exe
$ ./AMDuProfCLI.exe profile -e event=cycles-not-in-halt,interval=250000
-e event=retired-inst,interval=500000 -o c:\Temp\cpuprof-custom AMDTClassicMatMul.exe
Launch 'AMDTClassicMatMul-bin' and collect 'IBS' samples and generate thread-wise imix
C:\> AMDuProfCLI.exe profile --config ibs --imix --imix-group-by thread -o c:\Temp\cpuprof-tbp AMDTClassicMatMul.exe
Launch AMDTClassicMatMul.exe and perform Assess Performance profile
for 10 seconds and generate report
C:\> AMDuProfCLI.exe profile --config assess -o c:\Temp\cpuprof-assess -d 10 AMDTClassicMatMul.exe
Launch AMDTClassicMatMul.exe and collect the IBS samples in the SWP
mode and generate report sorted on ibs-op event
C:\> AMDuProfCLI.exe profile --config ibs -a -s event=ibs-op -o c:\Temp\cpuprof-ibs-swp AMDTClassicMatMul.exe
Collect the TBP samples in SWP mode for 10 seconds and generate report
C:\> AMDuProfCLI.exe profile -a -o c:\Temp\cpuprof-tbp-swp -d 10
Launch AMDTClassicMatMul.exe, collect TBP with callstack sampling
and generate report
C:\> AMDuProfCLI.exe profile --config tbp -g -o c:\Temp\cpuprof-tbp AMDTClassicMatMul.exe
Launch AMDTClassicMatMul.exe, collect TBP with callstack sampling
(unwind FPO optimized stack) and generate report
C:\> AMDuProfCLI.exe profile --config tbp --call-graph-mode fpo --call-graph-type user -o c:\Temp\cpuprof-tbp AMDTClassicMatMul.exe
Launch AMDTClassicMatMul.exe and collect the samples for PMCx076 and
PMCx0C0 and generate report sorted on pmcxc0 event
C:\> AMDuProfCLI.exe profile -e event=pmcx76,interval=250000 -e event=pmcxc0,user=1,os=0,interval=250000 -s event=pmcxc0 -o c:\Temp\cpuprof-tbp AMDTClassicMatMul.exe
Launch AMDTClassicMatMul.exe and collect the samples for IBS OP with
an interval of 50000 and generate report sorted on ibs-op event
C:\> AMDuProfCLI.exe profile -e event=ibs-op,interval=50000 -s event=ibs-op -o c:\Temp\cpuprof-tbp AMDTClassicMatMul.exe
Launch AMDTClassicMatMul.exe and do TBP samples profile for thread
concurrency, name, and generate report
C:\> AMDuProfCLI.exe profile --config tbp --thread thread=concurrency,name -o c:\Temp\cpuproftbp AMDTClassicMatMul.exe
Collect samples for PMCx076 and PMCx0C0, but collect the call graph info only for PMCx0C0 and generate report
C:\> AMDuProfCLI.exe profile -e event=pmcx76,interval=250000 -e event=pmcxc0,interval=250000,call-graph -o c:\Temp\cpuprof-pmc AMDTClassicMatMul-bin
Launch AMDTClassicMatMul.exe and collect the samples for predefined
event RETIRED_INST and L1_DC_REFILLS.ALL events and generate report
C:\> AMDuProfCLI.exe profile -e event=RETIRED_INST,interval=250000 -e event=L1_DC_REFILLS.ALL,user=1,os=0,interval=250000 -o c:\Temp\cpuprof-pmc AMDTClassicMatMul.exe
Launch AMDTClassicMatMul.exe. Collect the TBP, Assess Performance
samples, and generate report
C:\> AMDuProfCLI.exe profile --config tbp --config assess -o c:\Temp\cpuprof-tbp-assess AMDTClassicMatMul.exe
Linux
Launch the application AMDTClassicMatMul.bin. Collect the samples
for CYCLES_NOT_IN_HALT and RETIRED_INST events and generate report
$ ./AMDuProfCLI profile -e cycles-not-in-halt -e retired-inst
--interval 1000000 -o /tmp/cpuprof-custom AMDTClassicMatMul-bin
$ ./AMDuProfCLI profile -e event=cycles-not-in-halt,interval=250000
-e event=retired-inst,interval=500000 -o /tmp/cpuprof-custom AMDTClassicMatMul-bin
Launch the application AMDTClassicMatMul.bin.
Collect the IBS samples and generate thread-wise IMIX report from the raw data
file
$ ./AMDuProfCLI profile --config ibs --imix --imix-group-by thread -o /tmp/cpuprof-tbp AMDTClassicMatMul-bin
Launch AMDTClassicMatMul.bin. Perform Assess Performance profile for
10 seconds and generate report
$ ./AMDuProfCLI profile --config assess -o /tmp/cpuprof-assess -d 10 AMDTClassicMatMul-bin
Launch AMDTClassicMatMul.bin. Collect the IBS samples in the SWP
mode and generate report sorted based on ibs_op event
$ ./AMDuProfCLI profile --config ibs -a -s event=ibs_op -o /tmp/cpuprof-ibs-swp AMDTClassicMatMul-bin
Collect the TBP samples in SWP mode for 10 seconds and generate report
$ ./AMDuProfCLI profile -a -o /tmp/cpuprof-tbp-swp -d 10
Launch AMDTClassicMatMul.bin. Collect TBP with callstack sampling
and generate report
$ ./AMDuProfCLI profile --config tbp -g -o /tmp/cpuprof-tbp AMDTClassicMatMul-bin
Launch AMDTClassicMatMul.bin and collect TBP with callstack sampling
(unwind FPO optimized stack) and generate report
$ ./AMDuProfCLI profile --config tbp --call-graph-mode fpo --call-graph-size 512 -o /tmp/cpuprof-tbp AMDTClassicMatMul-bin
Launch AMDTClassicMatMul.bin. Collect the samples for PMCx076 and
PMCx0C0 and generate report
$ ./AMDuProfCLI profile -e event=pmcx76,interval=250000 -e event=pmcxc0,user=1,os=0,interval=250000 -o /tmp/cpuprof-tbp AMDTClassicMatMul-bin
Launch AMDTClassicMatMul.bin. Collect the samples for IBS OP with
interval 50000 and generate report sorted on ibs-op event
$ ./AMDuProfCLI profile -e event=ibs-op,interval=50000 -s event=ibs-op -o /tmp/cpuprof-tbp AMDTClassicMatMulbin
Attach to a thread, collect TBP samples for 10 seconds, and generate report
$ AMDuProfCLI profile --config tbp -o /tmp/cpuprof-tbp-attach -d 10 --tid <TID>
Collect basic OpenMP trace info of an OpenMP application compiled with GCC OpenMP library and generate the report
$ AMDuProfCLI profile --trace openmp --openmp-impl omplib -o /tmp/cpuprof-omp <path-to-openmp-exe>
Collect the samples for PMCx076 and PMCx0C0, but collect the call graph info only for PMCx0C0 and generate report
$ AMDuProfCLI profile -e event=pmcx76,interval=250000 -e event=pmcxc0,interval=250000,callgraph -o /tmp/cpuprof-pmc
AMDTClassicMatMul-bin
Launch 'AMDTClassicMatMul-bin' and collect pthread runtime trace with default threshold
$ AMDuProfCLI collect --trace osrt --osrt-event pthread -o /tmp/cpuprof-os AMDTClassicMatMul-bin
Launch 'AMDTClassicMatMul-bin' and collect syscall which are taking more than or equal to 1ms and generate report
$ AMDuProfCLI profile --trace osrt --osrt-event syscall --osrt-threshold syscall:1000000 -o /tmp/cpuprof-os AMDTClassicMatMul-bin
Launch 'AMDTClassicMatMul-bin' and collect the GPU Traces and generate gpu trace report
$ AMDuProfCLI profile --trace gpu -o /tmp/cpuprof-gpu AMDTClassicMatMul-bin --category gputrace
Launch AMDTClassicMatMul.bin. Collect the TBP samples, GPU Traces and generate report
$ AMDuProfCLI profile --config tbp --trace gpu -o /tmp/cpuprof-gpu AMDTClassicMatMul-bin
Launch 'AMDTClassicMatMul-bin' and collect 'GPU' samples and generate report
$ AMDuProfCLI profile --config gpu -o /tmp/gpuprof-gpu AMDTClassicMatMul-bin
$ AMDuProfCLI profile --config gpu --ip-block SQ -o /tmp/gpuprof-gpu AMDTClassicMatMul-bin
$ AMDuProfCLI profile --trace osrt --osrt-event function --osrt-funcs AMDTClassicMatMul-bin:* -o /tmp/cpuprof-os AMDTClassicMatMul-bin
$ AMDuProfCLI profile --trace osrt --osrt-event function --osrt-func-size 64 --osrt-threshold function:10000 --osrt-funcs AMDTClassicMatMul-bin:* -o /tmp/cpuprof-os AMDTClassicMatMul-bin