llama.cpp/report-long.txt at master · Seunghhon/llama.cpp · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
# To display the perf.data header info, please use --header/--header-only options.
#
#
# Total Lost Samples: 0
#
# Samples: 2M of event 'cpu-clock:pppH'
# Event count (approx.): 22644272500830
#
# Overhead  Command      Shared Object           Symbol
# ........  ...........  ......................  .........................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................
#
    48.22%  llama-bench  libggml-cpu.so.0.9.11   [.] ggml_vec_dot_f16
            |
            ---ggml_vec_dot_f16
               |
                --1.52%--0

    42.20%  llama-bench  libggml-cpu.so.0.9.11   [.] ggml_vec_dot_q4_K_q8_K_generic
            |
            ---ggml_vec_dot_q4_K_q8_K_generic
               |
                --8.81%--0

     7.99%  llama-bench  libggml-cpu.so.0.9.11   [.] ggml_vec_dot_q6_K_q8_K_generic
            |
            ---ggml_vec_dot_q6_K_q8_K_generic
               |
                --1.08%--0

     0.42%  llama-bench  libggml-cpu.so.0.9.11   [.] ggml_compute_forward_mul_mat
     0.27%  llama-bench  libggml-cpu.so.0.9.11   [.] ggml_cpu_fp32_to_fp16
     0.22%  llama-bench  libggml-cpu.so.0.9.11   [.] ggml_vec_soft_max_f32
     0.16%  llama-bench  libggml-cpu.so.0.9.11   [.] ggml_compute_forward_soft_max
     0.12%  llama-bench  libgomp.so.1.0.0        [.] 0x00000000000207ba
     0.06%  llama-bench  libggml-base.so.0.9.11  [.] quantize_row_q8_K_ref
     0.03%  llama-bench  [kernel.kallsyms]       [k] handle_softirqs
     0.03%  llama-bench  libgomp.so.1.0.0        [.] 0x00000000000207be
     0.03%  llama-bench  libggml-cpu.so.0.9.11   [.] rotate_pairs<float>
     0.03%  llama-bench  libc.so.6               [.] __memmove_sse2_unaligned_erms
     0.03%  llama-bench  libgomp.so.1.0.0        [.] 0x00000000000207c6