I am trying to do “perf top” and I met an address of libcuda.
Is there any idea of getting function name?
Following output is comming from “perf top” for tf32TensorCoreGemm (cuda-samples). it shows 0x00000000001d5e90 for libcuda.
‘‘‘
Samples: 542K of event ‘cycles’, 4000 Hz, Event count (approx.): 37556326977 lost: 0/0 drop: 0/0
Children Self Shared Object Symbol
- 66.54% 66.32% libc-2.31.so [.] random `
- 15.94% 12.08% tf32TensorCoreGemm [.] init_host_matrices a
- 11.71% 11.68% libc-2.31.so [.] random_r a
- 3.80% 0.07% [kernel] [k] page_fault a
- 2.86% 0.01% [kernel] [k] do_page_fault a
- 2.85% 0.07% [kernel] [k] __do_page_fault a
- 2.66% 0.11% [kernel] [k] handle_mm_fault a
- 2.47% 0.27% [kernel] [k] __handle_mm_fault a
- 1.89% 1.88% libc-2.31.so [.] rand a
- 1.23% 1.23% libcuda.so.470.57.02 [.] 0x00000000001d5e90 a
- 0.85% 0.04% [kernel] [k] alloc_pages_vma a
- 0.79% 0.08% [kernel] [k] __alloc_pages_nodemask a
0.75% 0.01% [kernel] [k] mem_cgroup_try_charge_delay a
0.68% 0.24% [kernel] [k] get_page_from_freelist a
0.62% 0.10% [kernel] [k] mem_cgroup_try_charge a
0.56% 0.00% [kernel] [k] do_syscall_64 a - 0.50% 0.50% [kernel] [k] native_irq_return_iret a
0.48% 0.00% [kernel] [k] unmap_region a - 0.44% 0.07% [kernel] [k] unmap_page_range a
0.42% 0.05% [kernel] [k] error_entry a
0.42% 0.41% tf32TensorCoreGemm [.] rand@plt a
0.40% 0.04% [kernel] [k] lru_cache_add_active_or_unevictable a
0.37% 0.37% [kernel] [k] sync_regs a
0.37% 0.00% [kernel] [k] unmap_vmas a - 0.37% 0.11% [kernel] [k] release_pages a
0.37% 0.36% [kernel] [k] clear_page_erms a
0.36% 0.03% [kernel] [k] __lru_cache_add a - 0.34% 0.01% [kernel] [k] free_pages_and_swap_cache
‘‘‘