Whether CUDA supports GPU devices with 8.6 Compute Capability？
I’m using NTX3090,
CUDA Driver Version / Runtime Version 12.0 / 11.8
CUDA Capability Major/Minor version number: 8.6
Excuse me, I mean I use 3090 graphics card, is CUPTI still available？ I am a novice, want to use the terminal to measure GPU performance indicators, but I can not successfully run CUPTI examples, i want to know what the problem is.（sorry my english is poor）
Can you paste the output of cupti samples ?
Also please clarify your CUDA and driver version. Thanks !
Sorry, I’m not sure if I can do this. I used the make command in cupti’s example directory to generate the executable file of cupti_query, and then ran the executable file and got the above output.
CUDA Version output by nvidia-smi: 12.0，
nvcc --version
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0
I made the content of cputi query in the root directory, and then ran the executable file.
The output is as follows
//
Assuming default device id 0
CUDA Device Id : 0
CUDA Device Name: NVIDIA GeForce RTX 3090
cupti_query.cpp:489:Error CUPTI_ERROR_LEGACY_PROFILER_NOT_SUPPORTED for CUPTI API function ‘cuptiDeviceEnumEventDomains’.
Hi, @1802725948
The error info is expected actually. As you are using Ampere, it doesn’t support legacy profile. Legacy profile is only supported on GPU sm < =7.0
You can try cupti sample with next-gen profile such as autorange_profiling, userrange_profiling, callback_profiling etc. Those should work.
thank you very much，I want to know where I can read the manual for nest-gen. I don’t seem to see it in the CUPTI-documentation.
nvcc -o ojbk auto_range_profiling.o -I/usr/local/cuda-11.8/extras/CUPTI/include -L/usr/local/cuda-11.8/extras/CUPTI/lib64 -lcuda -lcupti -lnvpw -lnvpwapi -v
#$ _NVVM_BRANCH_=nvvm
#$ _SPACE_=
#$ _CUDART_=cudart
#$ _HERE_=/usr/local/cuda/bin
#$ _THERE_=/usr/local/cuda/bin
#$ _TARGET_SIZE_=
#$ _TARGET_DIR_=
#$ _TARGET_DIR_=targets/x86_64-linux
#$ TOP=/usr/local/cuda/bin/..
#$ NVVMIR_LIBRARY_DIR=/usr/local/cuda/bin/../nvvm/libdevice
#$ LD_LIBRARY_PATH=/usr/local/cuda/bin/../lib:/usr/local/cuda-11.8/extras/CUPTI/lib64:/usr/local/cuda/lib64:/home/minghuan/instalexport/lib:/usr/local/cuda-11.8/extras/CUPTI/lib64:/usr/local/cuda/lib64:/home/minghuan/instalexport/lib:
#$ PATH=/usr/local/cuda/bin/../nvvm/bin:/usr/local/cuda/bin:/home/minghuan/anaconda3/bin:/home/minghuan/anaconda3/condabin:/home/minghuan/NSComput:/usr/local/cuda/bin:/home/minghuan/instalexport/bin:/home/minghuan/.nix-profile/bin:/nix/var/nix/profiles/default/bin:/home/minghuan/.vscode-server/bin/1a5daa3a0231a0fbba4f14db7ec463cf99d7768e/bin/remote-cli:/home/minghuan/.local/bin:/home/minghuan/NSComput:/usr/local/cuda/bin:/home/minghuan/instalexport/bin:/home/minghuan/.nix-profile/bin:/nix/var/nix/profiles/default/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin
#$ INCLUDES="-I/usr/local/cuda/bin/../targets/x86_64-linux/include"
#$ LIBRARIES= "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib"
#$ CUDAFE_FLAGS=
#$ PTXAS_FLAGS=
#$ nvlink -m64 --arch=sm_52 --register-link-binaries="/tmp/tmpxft_0034771c_00000000-3_ojbk_dlink.reg.c" -L"/usr/local/cuda-11.8/extras/CUPTI/lib64" -lcuda -lcupti -lnvpw -lnvpwapi "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib" -cpu-arch=X86_64 "auto_range_profiling.o" -lcudadevrt -o "/tmp/tmpxft_0034771c_00000000-5_ojbk_dlink.sm_52.cubin" --host-ccbin "gcc"
#$ fatbinary -64 -link "--image3=kind=elf,sm=52,file=/tmp/tmpxft_0034771c_00000000-5_ojbk_dlink.sm_52.cubin" --embedded-fatbin="/tmp/tmpxft_0034771c_00000000-4_ojbk_dlink.fatbin.c"
#$ rm /tmp/tmpxft_0034771c_00000000-4_ojbk_dlink.fatbin
#$ gcc -D__CUDA_ARCH_LIST__=520 -c -x c++ -DFATBINFILE="\"/tmp/tmpxft_0034771c_00000000-4_ojbk_dlink.fatbin.c\"" -DREGISTERLINKBINARYFILE="\"/tmp/tmpxft_0034771c_00000000-3_ojbk_dlink.reg.c\"" -I. -D__NV_EXTRA_INITIALIZATION= -D__NV_EXTRA_FINALIZATION= -D__CUDA_INCLUDE_COMPILER_INTERNAL_HEADERS__ -I"/usr/local/cuda-11.8/extras/CUPTI/include" "-I/usr/local/cuda/bin/../targets/x86_64-linux/include" -D__CUDACC_VER_MAJOR__=11 -D__CUDACC_VER_MINOR__=8 -D__CUDACC_VER_BUILD__=89 -D__CUDA_API_VER_MAJOR__=11 -D__CUDA_API_VER_MINOR__=8 -D__NVCC_DIAG_PRAGMA_SUPPORT__=1 -m64 "/usr/local/cuda/bin/crt/link.stub" -o "/tmp/tmpxft_0034771c_00000000-6_ojbk_dlink.o"
#$ g++ -D__CUDA_ARCH_LIST__=520 -m64 -Wl,--start-group "/tmp/tmpxft_0034771c_00000000-6_ojbk_dlink.o" "auto_range_profiling.o" -L"/usr/local/cuda-11.8/extras/CUPTI/lib64" -lcuda -lcupti -lnvpw -lnvpwapi "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib/stubs" "-L/usr/local/cuda/bin/../targets/x86_64-linux/lib" -lcudadevrt -lcudart_static -lrt -lpthread -ldl -Wl,--end-group -o "ojbk"
/usr/local/bin/ld: cannot find -lnvpw
/usr/local/bin/ld: cannot find -lnvpwapi
collect2: error: ld returned 1 exit status
# --error 0x1 --
After I make auto_range_profiling, I use auto_range_profiling.o to compile. What should I do with these errors?
Sorry @1802725948 . I am not sure what you intend to do. Can you clarify ?
nvcc -o ojbk auto_range_profiling.o -I/usr/local/cuda-11.8/extras/CUPTI/include -L/usr/local/cuda-11.8/extras/CUPTI/lib64 -lcuda -lcupti -lnvpw -lnvpwapi -v
Sorry, I may have done some rather foolish things, but this is what I believe is right. I obtained this program with a .o extension through make, and I want to compile this program using nvcc. The information following is added path information. I learned about these methods through GPT, and I’m not sure what the correct method is.
Which program do you want to compile ? After you do “make”, you already have auto_range_profiling generated. You don’t have to use nvcc to compile again.
Please execute “make clean” and then “make” in the sample directory directly. Thanks ！
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.