You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm trying to run the instr_count tool with the vectoradd example, and it appears to hang on CUDA 12.0. We're on Linux x86_64 with a GV100 and CUDA 12.0, Driver 525. I see that in the README it says, CUDA version: >= 8.0 && <= 11.x and Driver Version < 510. Are there plans to update NVBit to support these newer CUDA and driver versions?
In case it helps, here's a stacktrace I can get from GDB when I interrupt the program:
#0 __futex_abstimed_wait_common (cancel=false, private=0, abstime=0x0, clockid=0, expected=2, futex_word=0x555555638eb8)
at ./nptl/futex-internal.c:103
#1 __GI___futex_abstimed_wait64 (futex_word=futex_word@entry=0x555555638eb8, expected=expected@entry=2,
clockid=clockid@entry=0, abstime=abstime@entry=0x0, private=private@entry=0) at ./nptl/futex-internal.c:128
#2 0x00007ffff7bbb395 in __pthread_rwlock_wrlock_full64 (abstime=0x0, clockid=0, rwlock=0x555555638eb0)
at ./nptl/pthread_rwlock_common.c:829
#3 ___pthread_rwlock_wrlock (rwlock=0x555555638eb0) at ./nptl/pthread_rwlock_wrlock.c:26
#4 0x00007ffff6097687 in ?? () from /lib/x86_64-linux-gnu/libcuda.so.1
#5 0x00007ffff60f16d1 in ?? () from /lib/x86_64-linux-gnu/libcuda.so.1
#6 0x00007ffff7ed6a4c in __cudart667 () from ../../tools/instr_count/instr_count.so
#7 0x00007ffff7f2bbcd in __cudart679 () from ../../tools/instr_count/instr_count.so
#8 0x00007ffff7ed9f0e in __cudart515 () from ../../tools/instr_count/instr_count.so
#9 0x00007ffff7ee6104 in __cudart1329 () from ../../tools/instr_count/instr_count.so
#10 0x00007ffff7bb9f68 in __pthread_once_slow (once_control=0x7ffff7fb9408 <__cudart2818>,
init_routine=0x7ffff7ee5e10 <__cudart1329>) at ./nptl/pthread_once.c:116
#11 0x00007ffff7f31979 in __cudart1608 () from ../../tools/instr_count/instr_count.so
#12 0x00007ffff7eda267 in __cudart513 () from ../../tools/instr_count/instr_count.so
#13 0x00007ffff7f2516f in cudaLaunchKernel () from ../../tools/instr_count/instr_count.so
#14 0x00007ffff7e78110 in __device_stub__Z24load_module_nvbit_kerneli(int) () from ../../tools/instr_count/instr_count.so
#15 0x00007ffff7e78197 in nvbit_at_context_init_hook () from ../../tools/instr_count/instr_count.so
#16 0x00007ffff7e846fa in Nvbit::create_ctx(CUctx_st*) () from ../../tools/instr_count/instr_count.so
#17 0x00007ffff7e8d9d8 in nvbitToolsCallbackFunc(void*, CUtools_cb_domain_enum, unsigned int, void const*) ()
from ../../tools/instr_count/instr_count.so
The text was updated successfully, but these errors were encountered:
I'm trying to run the instr_count tool with the vectoradd example, and it appears to hang on CUDA 12.0. We're on Linux x86_64 with a GV100 and CUDA 12.0, Driver 525. I see that in the README it says,
CUDA version: >= 8.0 && <= 11.x
andDriver Version < 510
. Are there plans to update NVBit to support these newer CUDA and driver versions?In case it helps, here's a stacktrace I can get from GDB when I interrupt the program:
The text was updated successfully, but these errors were encountered: