0.00.065.049 I log_info: verbosity = 3 (adjust with the `-lv N` CLI arg) 0.00.065.052 I device_info: 0.00.065.072 I - ROCm0 : AMD Radeon Graphics (32752 MiB, 32728 MiB free) 0.00.065.077 I - ROCm1 : AMD Radeon Graphics (32752 MiB, 32728 MiB free) 0.00.065.084 I - CPU : AMD Ryzen 9 3900X 12-Core Processor (128730 MiB, 128730 MiB free) 0.00.065.146 I system_info: n_threads = 12 (n_threads_batch = 12) / 24 | ROCm : NO_VMM = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 | 0.00.065.149 I srv llama_server: n_parallel is set to auto, using n_parallel = 4 and kv_unified = true 0.00.065.257 I srv init: running without SSL 0.00.065.327 I srv init: using 23 threads for HTTP server 0.00.065.652 I srv start: binding port with default address family 0.00.066.854 I srv llama_server: loading model 0.00.066.862 I srv load_model: loading model '/home/qrf/.cache/llama.cpp/Qwen_Qwen3.6-27B-Q4_K_M.gguf' 0.00.066.915 I common_init_result: fitting params to device memory ... 0.00.066.917 I common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on) 0.06.459.240 W llama_context: n_ctx_seq (32768) < n_ctx_train (262144) -- the full capacity of the model will not be utilized 0.06.556.967 I common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable) 0.06.809.006 I srv load_model: initializing slots, n_slots = 4 0.06.924.991 W srv load_model: speculative decoding will use checkpoints 0.06.925.006 W common_speculative_init: no implementations specified for speculative decoding 0.06.925.008 I slot load_model: id 0 | task -1 | new slot, n_ctx = 32768 0.06.925.012 I slot load_model: id 1 | task -1 | new slot, n_ctx = 32768 0.06.925.012 I slot load_model: id 2 | task -1 | new slot, n_ctx = 32768 0.06.925.013 I slot load_model: id 3 | task -1 | new slot, n_ctx = 32768 0.06.925.065 I srv load_model: prompt cache is enabled, size limit: 8192 MiB 0.06.925.066 I srv load_model: use `--cache-ram 0` to disable the prompt cache 0.06.925.066 I srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391 0.06.925.067 I srv load_model: context checkpoints enabled, max = 32, min spacing = 256 0.06.925.087 I srv init: idle slots will be saved to prompt cache and cleared upon starting a new task 0.06.943.031 I init: chat template, example_format: '<|im_start|>system You are a helpful assistant<|im_end|> <|im_start|>user Hello<|im_end|> <|im_start|>assistant Hi there<|im_end|> <|im_start|>user How are you?<|im_end|> <|im_start|>assistant ' 0.06.956.768 I srv init: init: chat template, thinking = 1 0.06.956.808 I srv llama_server: model loaded 0.06.956.813 I srv llama_server: server is listening on http://127.0.0.1:8090 0.06.956.817 I srv update_slots: all slots are idle 0.10.002.849 I srv operator(): Chat format: peg-native 0.10.003.502 I slot get_availabl: id 3 | task -1 | selected slot by LRU, t_last = -1 0.10.003.505 I srv get_availabl: updating prompt cache 0.10.003.511 I srv load: - looking for better prompt, base f_keep = -1.000, sim = 0.000 0.10.003.515 I srv update: - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 32768 tokens, 8589934592 est) 0.10.003.516 I srv get_availabl: prompt cache update took 0.01 ms 0.10.003.597 I slot launch_slot_: id 3 | task 0 | processing task, is_child = 0 0.10.003.599 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 0.10.003.600 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 0.10.003.627 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 0.10.003.627 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 0.10.003.646 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 0.10.003.647 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 0.12.007.540 I slot create_check: id 3 | task 0 | created context checkpoint 1 of 32 (pos_min = 298, pos_max = 298, n_tokens = 299, size = 149.626 MiB) 0.17.998.885 I slot print_timing: id 3 | task 0 | n_decoded = 100, tg = 18.85 t/s, tg_3s = 18.85 t/s 0.21.014.947 I slot print_timing: id 3 | task 0 | n_decoded = 156, tg = 18.75 t/s, tg_3s = 18.57 t/s 0.23.485.547 I slot print_timing: id 3 | task 0 | prompt eval time = 2691.42 ms / 406 tokens ( 6.63 ms per token, 150.85 tokens per second) 0.23.485.551 I slot print_timing: id 3 | task 0 | eval time = 10790.44 ms / 202 tokens ( 53.42 ms per token, 18.72 tokens per second) 0.23.485.552 I slot print_timing: id 3 | task 0 | total time = 13481.86 ms / 608 tokens 0.23.485.556 I slot print_timing: id 3 | task 0 | graphs reused = 200 0.23.485.629 I slot release: id 3 | task 0 | stop processing: n_tokens = 607, truncated = 0 0.23.485.640 I srv update_slots: all slots are idle 0.23.502.839 I srv operator(): Chat format: peg-native 0.23.503.813 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.692 (> 0.100 thold), f_keep = 0.666 0.23.503.872 I slot launch_slot_: id 3 | task 205 | processing task, is_child = 0 0.23.503.873 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 0.23.503.874 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 0.23.503.896 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 0.23.503.896 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 0.23.503.913 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 0.23.503.913 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 0.23.503.934 I slot update_slots: id 3 | task 205 | Checking checkpoint with [298, 298] against 404... 0.23.539.342 W slot update_slots: id 3 | task 205 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 0.24.937.347 I slot create_check: id 3 | task 205 | created context checkpoint 2 of 32 (pos_min = 557, pos_max = 557, n_tokens = 558, size = 149.626 MiB) 0.27.451.822 I slot print_timing: id 3 | task 205 | prompt eval time = 1784.69 ms / 285 tokens ( 6.26 ms per token, 159.69 tokens per second) 0.27.451.825 I slot print_timing: id 3 | task 205 | eval time = 2163.19 ms / 41 tokens ( 52.76 ms per token, 18.95 tokens per second) 0.27.451.826 I slot print_timing: id 3 | task 205 | total time = 3947.88 ms / 326 tokens 0.27.451.827 I slot print_timing: id 3 | task 205 | graphs reused = 239 0.27.451.881 I slot release: id 3 | task 205 | stop processing: n_tokens = 624, truncated = 0 0.27.451.890 I srv update_slots: all slots are idle 0.27.485.100 I srv operator(): Chat format: peg-native 0.27.486.105 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.860 (> 0.100 thold), f_keep = 0.933 0.27.486.179 I slot launch_slot_: id 3 | task 249 | processing task, is_child = 0 0.27.486.181 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 0.27.486.182 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 0.27.486.206 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 0.27.486.206 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 0.27.486.225 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 0.27.486.226 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 0.27.486.248 I slot update_slots: id 3 | task 249 | Checking checkpoint with [557, 557] against 582... 0.27.521.632 W slot update_slots: id 3 | task 249 | restored context checkpoint (pos_min = 557, pos_max = 557, n_tokens = 558, n_past = 558, size = 149.626 MiB) 0.32.589.913 I slot print_timing: id 3 | task 249 | prompt eval time = 830.76 ms / 119 tokens ( 6.98 ms per token, 143.24 tokens per second) 0.32.589.918 I slot print_timing: id 3 | task 249 | eval time = 4272.90 ms / 80 tokens ( 53.41 ms per token, 18.72 tokens per second) 0.32.589.918 I slot print_timing: id 3 | task 249 | total time = 5103.66 ms / 199 tokens 0.32.589.919 I slot print_timing: id 3 | task 249 | graphs reused = 317 0.32.589.978 I slot release: id 3 | task 249 | stop processing: n_tokens = 756, truncated = 0 0.32.589.987 I srv update_slots: all slots are idle 0.32.623.442 I srv operator(): Chat format: peg-native 0.32.624.069 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.790 (> 0.100 thold), f_keep = 0.402 0.32.624.072 I srv get_availabl: updating prompt cache 0.32.624.141 W srv prompt_save: - saving prompt with length 756, total state size = 196.891 MiB (draft: 0.000 MiB) 0.32.913.646 I srv load: - looking for better prompt, base f_keep = 0.402, sim = 0.790 0.32.913.653 I srv update: - cache state: 1 prompts, 496.143 MiB (limits: 8192.000 MiB, 32768 tokens, 32768 est) 0.32.913.654 I srv update: - prompt 0x63ea542b1aa0: 756 tokens, checkpoints: 2, 496.143 MiB 0.32.913.655 I srv get_availabl: prompt cache update took 289.58 ms 0.32.913.737 I slot launch_slot_: id 3 | task 332 | processing task, is_child = 0 0.32.913.738 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 0.32.913.739 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 0.32.913.760 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 0.32.913.761 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 0.32.913.777 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 0.32.913.778 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 0.32.913.802 I slot update_slots: id 3 | task 332 | Checking checkpoint with [557, 557] against 304... 0.32.913.803 I slot update_slots: id 3 | task 332 | Checking checkpoint with [298, 298] against 304... 0.32.949.348 W slot update_slots: id 3 | task 332 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 0.32.949.351 W slot update_slots: id 3 | task 332 | erased invalidated context checkpoint (pos_min = 557, pos_max = 557, n_tokens = 558, n_swa = 0, pos_next = 299, size = 149.626 MiB) 0.37.728.070 I slot print_timing: id 3 | task 332 | prompt eval time = 654.67 ms / 86 tokens ( 7.61 ms per token, 131.36 tokens per second) 0.37.728.074 I slot print_timing: id 3 | task 332 | eval time = 4159.59 ms / 78 tokens ( 53.33 ms per token, 18.75 tokens per second) 0.37.728.075 I slot print_timing: id 3 | task 332 | total time = 4814.26 ms / 164 tokens 0.37.728.075 I slot print_timing: id 3 | task 332 | graphs reused = 393 0.37.728.124 I slot release: id 3 | task 332 | stop processing: n_tokens = 462, truncated = 0 0.37.728.134 I srv update_slots: all slots are idle 0.37.745.353 I srv operator(): Chat format: peg-native 0.37.746.126 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.760 (> 0.100 thold), f_keep = 0.829 0.37.746.191 I slot launch_slot_: id 3 | task 412 | processing task, is_child = 0 0.37.746.194 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 0.37.746.194 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 0.37.746.221 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 0.37.746.222 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 0.37.746.254 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 0.37.746.254 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 0.37.746.285 I slot update_slots: id 3 | task 412 | Checking checkpoint with [298, 298] against 383... 0.37.781.682 W slot update_slots: id 3 | task 412 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 0.44.140.509 I slot print_timing: id 3 | task 412 | n_decoded = 100, tg = 18.73 t/s, tg_3s = 18.73 t/s 0.47.191.774 I slot print_timing: id 3 | task 412 | n_decoded = 157, tg = 18.71 t/s, tg_3s = 18.68 t/s 0.49.341.276 I slot print_timing: id 3 | task 412 | prompt eval time = 1054.92 ms / 205 tokens ( 5.15 ms per token, 194.33 tokens per second) 0.49.341.279 I slot print_timing: id 3 | task 412 | eval time = 10540.07 ms / 197 tokens ( 53.50 ms per token, 18.69 tokens per second) 0.49.341.280 I slot print_timing: id 3 | task 412 | total time = 11594.99 ms / 402 tokens 0.49.341.281 I slot print_timing: id 3 | task 412 | graphs reused = 587 0.49.341.341 I slot release: id 3 | task 412 | stop processing: n_tokens = 700, truncated = 0 0.49.341.350 I srv update_slots: all slots are idle 0.49.360.107 I srv operator(): Chat format: peg-native 0.49.361.179 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.783 (> 0.100 thold), f_keep = 0.717 0.49.361.217 I slot launch_slot_: id 3 | task 612 | processing task, is_child = 0 0.49.361.219 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 0.49.361.219 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 0.49.361.242 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 0.49.361.242 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 0.49.361.259 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 0.49.361.260 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 0.49.361.279 I slot update_slots: id 3 | task 612 | Checking checkpoint with [298, 298] against 502... 0.49.396.824 W slot update_slots: id 3 | task 612 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 0.50.972.892 I slot create_check: id 3 | task 612 | created context checkpoint 2 of 32 (pos_min = 615, pos_max = 615, n_tokens = 616, size = 149.626 MiB) 0.53.596.975 I slot print_timing: id 3 | task 612 | prompt eval time = 1963.24 ms / 342 tokens ( 5.74 ms per token, 174.20 tokens per second) 0.53.596.981 I slot print_timing: id 3 | task 612 | eval time = 2272.45 ms / 43 tokens ( 52.85 ms per token, 18.92 tokens per second) 0.53.596.981 I slot print_timing: id 3 | task 612 | total time = 4235.69 ms / 385 tokens 0.53.596.982 I slot print_timing: id 3 | task 612 | graphs reused = 628 0.53.597.050 I slot release: id 3 | task 612 | stop processing: n_tokens = 683, truncated = 0 0.53.597.059 I srv update_slots: all slots are idle 0.53.630.778 I srv operator(): Chat format: peg-native 0.53.631.956 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.931 (> 0.100 thold), f_keep = 0.936 0.53.632.009 I slot launch_slot_: id 3 | task 658 | processing task, is_child = 0 0.53.632.010 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 0.53.632.011 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 0.53.632.032 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 0.53.632.033 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 0.53.632.050 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 0.53.632.050 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 0.53.632.071 I slot update_slots: id 3 | task 658 | Checking checkpoint with [615, 615] against 639... 0.53.667.369 W slot update_slots: id 3 | task 658 | restored context checkpoint (pos_min = 615, pos_max = 615, n_tokens = 616, n_past = 616, size = 149.626 MiB) 0.59.162.978 I slot print_timing: id 3 | task 658 | prompt eval time = 561.36 ms / 70 tokens ( 8.02 ms per token, 124.70 tokens per second) 0.59.162.982 I slot print_timing: id 3 | task 658 | eval time = 4969.54 ms / 93 tokens ( 53.44 ms per token, 18.71 tokens per second) 0.59.162.982 I slot print_timing: id 3 | task 658 | total time = 5530.90 ms / 163 tokens 0.59.162.983 I slot print_timing: id 3 | task 658 | graphs reused = 718 0.59.163.042 I slot release: id 3 | task 658 | stop processing: n_tokens = 778, truncated = 0 0.59.163.050 I srv update_slots: all slots are idle 0.59.196.703 I srv operator(): Chat format: peg-native 0.59.197.372 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.770 (> 0.100 thold), f_keep = 0.391 0.59.197.374 I srv get_availabl: updating prompt cache 0.59.197.437 W srv prompt_save: - saving prompt with length 778, total state size = 198.266 MiB (draft: 0.000 MiB) 0.59.485.485 I srv load: - looking for better prompt, base f_keep = 0.391, sim = 0.770 0.59.485.491 I srv update: - cache state: 2 prompts, 993.662 MiB (limits: 8192.000 MiB, 32768 tokens, 32768 est) 0.59.485.492 I srv update: - prompt 0x63ea542b1aa0: 756 tokens, checkpoints: 2, 496.143 MiB 0.59.485.493 I srv update: - prompt 0x63ea4cf25900: 778 tokens, checkpoints: 2, 497.519 MiB 0.59.485.494 I srv get_availabl: prompt cache update took 288.12 ms 0.59.485.575 I slot launch_slot_: id 3 | task 754 | processing task, is_child = 0 0.59.485.576 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 0.59.485.576 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 0.59.485.598 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 0.59.485.599 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 0.59.485.616 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 0.59.485.616 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 0.59.485.641 I slot update_slots: id 3 | task 754 | Checking checkpoint with [615, 615] against 304... 0.59.485.641 I slot update_slots: id 3 | task 754 | Checking checkpoint with [298, 298] against 304... 0.59.520.698 W slot update_slots: id 3 | task 754 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 0.59.520.700 W slot update_slots: id 3 | task 754 | erased invalidated context checkpoint (pos_min = 615, pos_max = 615, n_tokens = 616, n_swa = 0, pos_next = 299, size = 149.626 MiB) 1.05.480.020 I slot print_timing: id 3 | task 754 | n_decoded = 100, tg = 18.75 t/s, tg_3s = 18.75 t/s 1.08.289.037 I slot print_timing: id 3 | task 754 | prompt eval time = 661.39 ms / 96 tokens ( 6.89 ms per token, 145.15 tokens per second) 1.08.289.041 I slot print_timing: id 3 | task 754 | eval time = 8142.00 ms / 152 tokens ( 53.57 ms per token, 18.67 tokens per second) 1.08.289.042 I slot print_timing: id 3 | task 754 | total time = 8803.39 ms / 248 tokens 1.08.289.042 I slot print_timing: id 3 | task 754 | graphs reused = 867 1.08.289.101 I slot release: id 3 | task 754 | stop processing: n_tokens = 546, truncated = 0 1.08.289.110 I srv update_slots: all slots are idle 1.08.306.356 I srv operator(): Chat format: peg-native 1.08.307.181 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.822 (> 0.100 thold), f_keep = 0.720 1.08.307.218 I slot launch_slot_: id 3 | task 908 | processing task, is_child = 0 1.08.307.218 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 1.08.307.219 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 1.08.307.242 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 1.08.307.242 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 1.08.307.259 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 1.08.307.259 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 1.08.307.279 I slot update_slots: id 3 | task 908 | Checking checkpoint with [298, 298] against 393... 1.08.342.589 W slot update_slots: id 3 | task 908 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 1.11.067.062 I slot print_timing: id 3 | task 908 | prompt eval time = 1027.93 ms / 179 tokens ( 5.74 ms per token, 174.14 tokens per second) 1.11.067.066 I slot print_timing: id 3 | task 908 | eval time = 1731.85 ms / 33 tokens ( 52.48 ms per token, 19.05 tokens per second) 1.11.067.067 I slot print_timing: id 3 | task 908 | total time = 2759.78 ms / 212 tokens 1.11.067.067 I slot print_timing: id 3 | task 908 | graphs reused = 898 1.11.067.122 I slot release: id 3 | task 908 | stop processing: n_tokens = 510, truncated = 0 1.11.067.131 I srv update_slots: all slots are idle 1.11.100.055 I srv operator(): Chat format: peg-native 1.11.100.929 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.917 (> 0.100 thold), f_keep = 0.933 1.11.100.982 I slot launch_slot_: id 3 | task 944 | processing task, is_child = 0 1.11.100.984 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 1.11.100.985 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 1.11.101.010 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 1.11.101.010 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 1.11.101.027 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 1.11.101.027 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 1.11.101.049 I slot update_slots: id 3 | task 944 | Checking checkpoint with [298, 298] against 476... 1.11.136.463 W slot update_slots: id 3 | task 944 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 1.16.804.108 I slot print_timing: id 3 | task 944 | prompt eval time = 1321.02 ms / 220 tokens ( 6.00 ms per token, 166.54 tokens per second) 1.16.804.112 I slot print_timing: id 3 | task 944 | eval time = 4382.03 ms / 82 tokens ( 53.44 ms per token, 18.71 tokens per second) 1.16.804.113 I slot print_timing: id 3 | task 944 | total time = 5703.05 ms / 302 tokens 1.16.804.113 I slot print_timing: id 3 | task 944 | graphs reused = 978 1.16.804.180 I slot release: id 3 | task 944 | stop processing: n_tokens = 600, truncated = 0 1.16.804.189 I srv update_slots: all slots are idle 1.16.822.652 I srv operator(): Chat format: peg-native 1.16.823.613 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.860 (> 0.100 thold), f_keep = 0.862 1.16.823.647 I slot launch_slot_: id 3 | task 1029 | processing task, is_child = 0 1.16.823.647 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 1.16.823.648 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 1.16.823.670 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 1.16.823.670 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 1.16.823.687 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 1.16.823.687 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 1.16.823.707 I slot update_slots: id 3 | task 1029 | Checking checkpoint with [298, 298] against 517... 1.16.859.178 W slot update_slots: id 3 | task 1029 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 1.18.269.719 I slot create_check: id 3 | task 1029 | created context checkpoint 2 of 32 (pos_min = 575, pos_max = 575, n_tokens = 576, size = 149.626 MiB) 1.20.574.257 I slot print_timing: id 3 | task 1029 | prompt eval time = 1798.03 ms / 302 tokens ( 5.95 ms per token, 167.96 tokens per second) 1.20.574.261 I slot print_timing: id 3 | task 1029 | eval time = 1952.52 ms / 37 tokens ( 52.77 ms per token, 18.95 tokens per second) 1.20.574.262 I slot print_timing: id 3 | task 1029 | total time = 3750.54 ms / 339 tokens 1.20.574.262 I slot print_timing: id 3 | task 1029 | graphs reused = 1013 1.20.574.334 I slot release: id 3 | task 1029 | stop processing: n_tokens = 637, truncated = 0 1.20.574.345 I srv update_slots: all slots are idle 1.20.608.146 I srv operator(): Chat format: peg-native 1.20.609.164 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.927 (> 0.100 thold), f_keep = 0.940 1.20.609.225 I slot launch_slot_: id 3 | task 1069 | processing task, is_child = 0 1.20.609.227 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 1.20.609.228 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 1.20.609.253 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 1.20.609.254 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 1.20.609.273 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 1.20.609.273 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 1.20.609.298 I slot update_slots: id 3 | task 1069 | Checking checkpoint with [575, 575] against 599... 1.20.644.648 W slot update_slots: id 3 | task 1069 | restored context checkpoint (pos_min = 575, pos_max = 575, n_tokens = 576, n_past = 576, size = 149.626 MiB) 1.25.121.320 I slot print_timing: id 3 | task 1069 | prompt eval time = 561.31 ms / 70 tokens ( 8.02 ms per token, 124.71 tokens per second) 1.25.121.324 I slot print_timing: id 3 | task 1069 | eval time = 3950.71 ms / 74 tokens ( 53.39 ms per token, 18.73 tokens per second) 1.25.121.324 I slot print_timing: id 3 | task 1069 | total time = 4512.02 ms / 144 tokens 1.25.121.325 I slot print_timing: id 3 | task 1069 | graphs reused = 1085 1.25.121.384 I slot release: id 3 | task 1069 | stop processing: n_tokens = 719, truncated = 0 1.25.121.392 I srv update_slots: all slots are idle 1.25.152.817 I srv operator(): Chat format: peg-native 1.25.153.426 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.724 (> 0.100 thold), f_keep = 0.423 1.25.153.428 I srv get_availabl: updating prompt cache 1.25.153.486 W srv prompt_save: - saving prompt with length 719, total state size = 194.578 MiB (draft: 0.000 MiB) 1.25.429.889 I srv load: - looking for better prompt, base f_keep = 0.423, sim = 0.724 1.25.429.898 I srv update: - cache state: 3 prompts, 1487.492 MiB (limits: 8192.000 MiB, 32768 tokens, 32768 est) 1.25.429.899 I srv update: - prompt 0x63ea542b1aa0: 756 tokens, checkpoints: 2, 496.143 MiB 1.25.429.899 I srv update: - prompt 0x63ea4cf25900: 778 tokens, checkpoints: 2, 497.519 MiB 1.25.429.900 I srv update: - prompt 0x63ea5015b160: 719 tokens, checkpoints: 2, 493.830 MiB 1.25.429.901 I srv get_availabl: prompt cache update took 276.47 ms 1.25.429.972 I slot launch_slot_: id 3 | task 1146 | processing task, is_child = 0 1.25.429.973 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 1.25.429.974 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 1.25.429.998 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 1.25.429.998 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 1.25.430.038 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 1.25.430.040 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 1.25.430.065 I slot update_slots: id 3 | task 1146 | Checking checkpoint with [575, 575] against 304... 1.25.430.066 I slot update_slots: id 3 | task 1146 | Checking checkpoint with [298, 298] against 304... 1.25.465.348 W slot update_slots: id 3 | task 1146 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 1.25.465.353 W slot update_slots: id 3 | task 1146 | erased invalidated context checkpoint (pos_min = 575, pos_max = 575, n_tokens = 576, n_swa = 0, pos_next = 299, size = 149.626 MiB) 1.31.581.804 I slot print_timing: id 3 | task 1146 | n_decoded = 100, tg = 18.73 t/s, tg_3s = 18.73 t/s 1.34.587.801 I slot print_timing: id 3 | task 1146 | n_decoded = 156, tg = 18.70 t/s, tg_3s = 18.63 t/s 1.37.589.761 I slot print_timing: id 3 | task 1146 | n_decoded = 212, tg = 18.68 t/s, tg_3s = 18.65 t/s 1.40.602.748 I slot print_timing: id 3 | task 1146 | n_decoded = 268, tg = 18.66 t/s, tg_3s = 18.59 t/s 1.43.613.842 I slot print_timing: id 3 | task 1146 | n_decoded = 324, tg = 18.65 t/s, tg_3s = 18.60 t/s 1.46.646.857 I slot print_timing: id 3 | task 1146 | n_decoded = 380, tg = 18.62 t/s, tg_3s = 18.46 t/s 1.48.804.913 I slot print_timing: id 3 | task 1146 | prompt eval time = 813.67 ms / 121 tokens ( 6.72 ms per token, 148.71 tokens per second) 1.48.804.918 I slot print_timing: id 3 | task 1146 | eval time = 22561.17 ms / 420 tokens ( 53.72 ms per token, 18.62 tokens per second) 1.48.804.918 I slot print_timing: id 3 | task 1146 | total time = 23374.84 ms / 541 tokens 1.48.804.919 I slot print_timing: id 3 | task 1146 | graphs reused = 1501 1.48.804.980 I slot release: id 3 | task 1146 | stop processing: n_tokens = 839, truncated = 0 1.48.804.989 I srv update_slots: all slots are idle 1.48.822.291 I srv operator(): Chat format: peg-native 1.48.823.178 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.722 (> 0.100 thold), f_keep = 0.498 1.48.823.180 I srv get_availabl: updating prompt cache 1.48.823.241 W srv prompt_save: - saving prompt with length 839, total state size = 202.080 MiB (draft: 0.000 MiB) 1.49.042.054 I srv load: - looking for better prompt, base f_keep = 0.498, sim = 0.722 1.49.042.063 I srv update: - cache state: 4 prompts, 1839.198 MiB (limits: 8192.000 MiB, 32768 tokens, 32768 est) 1.49.042.064 I srv update: - prompt 0x63ea542b1aa0: 756 tokens, checkpoints: 2, 496.143 MiB 1.49.042.064 I srv update: - prompt 0x63ea4cf25900: 778 tokens, checkpoints: 2, 497.519 MiB 1.49.042.065 I srv update: - prompt 0x63ea5015b160: 719 tokens, checkpoints: 2, 493.830 MiB 1.49.042.065 I srv update: - prompt 0x63ea4cadf8a0: 839 tokens, checkpoints: 1, 351.706 MiB 1.49.042.067 I srv get_availabl: prompt cache update took 218.89 ms 1.49.042.136 I slot launch_slot_: id 3 | task 1568 | processing task, is_child = 0 1.49.042.136 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 1.49.042.137 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 1.49.042.167 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 1.49.042.167 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 1.49.042.208 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 1.49.042.208 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 1.49.042.232 I slot update_slots: id 3 | task 1568 | Checking checkpoint with [298, 298] against 418... 1.49.077.530 W slot update_slots: id 3 | task 1568 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 1.50.527.077 I slot create_check: id 3 | task 1568 | created context checkpoint 2 of 32 (pos_min = 574, pos_max = 574, n_tokens = 575, size = 149.626 MiB) 1.56.021.098 I slot print_timing: id 3 | task 1568 | n_decoded = 100, tg = 18.68 t/s, tg_3s = 18.68 t/s 1.58.558.194 I slot print_timing: id 3 | task 1568 | prompt eval time = 1625.39 ms / 280 tokens ( 5.80 ms per token, 172.27 tokens per second) 1.58.558.197 I slot print_timing: id 3 | task 1568 | eval time = 7890.57 ms / 147 tokens ( 53.68 ms per token, 18.63 tokens per second) 1.58.558.198 I slot print_timing: id 3 | task 1568 | total time = 9515.95 ms / 427 tokens 1.58.558.199 I slot print_timing: id 3 | task 1568 | graphs reused = 1646 1.58.558.259 I slot release: id 3 | task 1568 | stop processing: n_tokens = 725, truncated = 0 1.58.558.268 I srv update_slots: all slots are idle 1.58.576.553 I srv operator(): Chat format: peg-native 1.58.577.611 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.809 (> 0.100 thold), f_keep = 0.796 1.58.577.665 I slot launch_slot_: id 3 | task 1718 | processing task, is_child = 0 1.58.577.667 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 1.58.577.667 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 1.58.577.689 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 1.58.577.689 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 1.58.577.706 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 1.58.577.706 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 1.58.577.727 I slot update_slots: id 3 | task 1718 | Checking checkpoint with [574, 574] against 577... 1.58.612.904 W slot update_slots: id 3 | task 1718 | restored context checkpoint (pos_min = 574, pos_max = 574, n_tokens = 575, n_past = 575, size = 149.626 MiB) 2.01.298.271 I slot print_timing: id 3 | task 1718 | prompt eval time = 930.99 ms / 138 tokens ( 6.75 ms per token, 148.23 tokens per second) 2.01.298.274 I slot print_timing: id 3 | task 1718 | eval time = 1789.55 ms / 34 tokens ( 52.63 ms per token, 19.00 tokens per second) 2.01.298.275 I slot print_timing: id 3 | task 1718 | total time = 2720.54 ms / 172 tokens 2.01.298.276 I slot print_timing: id 3 | task 1718 | graphs reused = 1678 2.01.298.347 I slot release: id 3 | task 1718 | stop processing: n_tokens = 746, truncated = 0 2.01.298.356 I srv update_slots: all slots are idle 2.01.332.106 I srv operator(): Chat format: peg-native 2.01.333.257 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.934 (> 0.100 thold), f_keep = 0.953 2.01.333.310 I slot launch_slot_: id 3 | task 1755 | processing task, is_child = 0 2.01.333.312 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 2.01.333.313 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 2.01.333.342 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 2.01.333.342 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 2.01.333.365 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 2.01.333.366 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 2.01.333.396 I slot update_slots: id 3 | task 1755 | Checking checkpoint with [574, 574] against 711... 2.01.368.967 W slot update_slots: id 3 | task 1755 | restored context checkpoint (pos_min = 574, pos_max = 574, n_tokens = 575, n_past = 575, size = 149.626 MiB) 2.05.648.080 I slot print_timing: id 3 | task 1755 | prompt eval time = 1056.84 ms / 186 tokens ( 5.68 ms per token, 176.00 tokens per second) 2.05.648.084 I slot print_timing: id 3 | task 1755 | eval time = 3257.84 ms / 61 tokens ( 53.41 ms per token, 18.72 tokens per second) 2.05.648.084 I slot print_timing: id 3 | task 1755 | total time = 4314.68 ms / 247 tokens 2.05.648.085 I slot print_timing: id 3 | task 1755 | graphs reused = 1736 2.05.648.159 I slot release: id 3 | task 1755 | stop processing: n_tokens = 821, truncated = 0 2.05.648.168 I srv update_slots: all slots are idle 2.05.681.502 I srv operator(): Chat format: peg-native 2.05.682.110 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.727 (> 0.100 thold), f_keep = 0.373 2.05.682.112 I srv get_availabl: updating prompt cache 2.05.682.194 W srv prompt_save: - saving prompt with length 821, total state size = 200.955 MiB (draft: 0.000 MiB) 2.05.969.387 I srv load: - looking for better prompt, base f_keep = 0.373, sim = 0.727 2.05.969.396 I srv update: - cache state: 5 prompts, 2339.405 MiB (limits: 8192.000 MiB, 32768 tokens, 32768 est) 2.05.969.397 I srv update: - prompt 0x63ea542b1aa0: 756 tokens, checkpoints: 2, 496.143 MiB 2.05.969.400 I srv update: - prompt 0x63ea4cf25900: 778 tokens, checkpoints: 2, 497.519 MiB 2.05.969.401 I srv update: - prompt 0x63ea5015b160: 719 tokens, checkpoints: 2, 493.830 MiB 2.05.969.401 I srv update: - prompt 0x63ea4cadf8a0: 839 tokens, checkpoints: 1, 351.706 MiB 2.05.969.402 I srv update: - prompt 0x63ea4ca8bf60: 821 tokens, checkpoints: 2, 500.207 MiB 2.05.969.404 I srv get_availabl: prompt cache update took 287.29 ms 2.05.969.476 I slot launch_slot_: id 3 | task 1819 | processing task, is_child = 0 2.05.969.477 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 2.05.969.478 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 2.05.969.502 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 2.05.969.502 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 2.05.969.519 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 2.05.969.519 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 2.05.969.545 I slot update_slots: id 3 | task 1819 | Checking checkpoint with [574, 574] against 306... 2.05.969.545 I slot update_slots: id 3 | task 1819 | Checking checkpoint with [298, 298] against 306... 2.06.004.854 W slot update_slots: id 3 | task 1819 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 2.06.004.856 W slot update_slots: id 3 | task 1819 | erased invalidated context checkpoint (pos_min = 574, pos_max = 574, n_tokens = 575, n_swa = 0, pos_next = 299, size = 149.626 MiB) 2.12.148.082 I slot print_timing: id 3 | task 1819 | n_decoded = 100, tg = 18.65 t/s, tg_3s = 18.65 t/s 2.15.172.394 I slot print_timing: id 3 | task 1819 | n_decoded = 156, tg = 18.60 t/s, tg_3s = 18.52 t/s 2.18.196.165 I slot print_timing: id 3 | task 1819 | n_decoded = 212, tg = 18.58 t/s, tg_3s = 18.52 t/s 2.21.221.309 I slot print_timing: id 3 | task 1819 | n_decoded = 268, tg = 18.56 t/s, tg_3s = 18.51 t/s 2.21.868.820 I slot print_timing: id 3 | task 1819 | prompt eval time = 815.19 ms / 122 tokens ( 6.68 ms per token, 149.66 tokens per second) 2.21.868.824 I slot print_timing: id 3 | task 1819 | eval time = 15084.08 ms / 280 tokens ( 53.87 ms per token, 18.56 tokens per second) 2.21.868.824 I slot print_timing: id 3 | task 1819 | total time = 15899.27 ms / 402 tokens 2.21.868.825 I slot print_timing: id 3 | task 1819 | graphs reused = 2013 2.21.868.890 I slot release: id 3 | task 1819 | stop processing: n_tokens = 700, truncated = 0 2.21.868.901 I srv update_slots: all slots are idle 2.21.886.512 I srv operator(): Chat format: peg-native 2.21.887.517 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.693 (> 0.100 thold), f_keep = 0.599 2.21.887.572 I slot launch_slot_: id 3 | task 2101 | processing task, is_child = 0 2.21.887.573 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 2.21.887.574 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 2.21.887.595 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 2.21.887.595 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 2.21.887.612 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 2.21.887.612 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 2.21.887.633 I slot update_slots: id 3 | task 2101 | Checking checkpoint with [298, 298] against 419... 2.21.923.149 W slot update_slots: id 3 | task 2101 | restored context checkpoint (pos_min = 298, pos_max = 298, n_tokens = 299, n_past = 299, size = 149.626 MiB) 2.23.463.848 I slot create_check: id 3 | task 2101 | created context checkpoint 2 of 32 (pos_min = 579, pos_max = 579, n_tokens = 580, size = 149.626 MiB) 2.25.869.926 I slot print_timing: id 3 | task 2101 | prompt eval time = 1928.18 ms / 306 tokens ( 6.30 ms per token, 158.70 tokens per second) 2.25.869.930 I slot print_timing: id 3 | task 2101 | eval time = 2054.11 ms / 39 tokens ( 52.67 ms per token, 18.99 tokens per second) 2.25.869.931 I slot print_timing: id 3 | task 2101 | total time = 3982.29 ms / 345 tokens 2.25.869.932 I slot print_timing: id 3 | task 2101 | graphs reused = 2050 2.25.870.031 I slot release: id 3 | task 2101 | stop processing: n_tokens = 643, truncated = 0 2.25.870.043 I srv update_slots: all slots are idle 2.25.903.442 I srv operator(): Chat format: peg-native 2.25.904.517 I slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 0.928 (> 0.100 thold), f_keep = 0.938 2.25.904.562 I slot launch_slot_: id 3 | task 2143 | processing task, is_child = 0 2.25.904.564 I slot process_sing: id 0 | task -1 | saving idle slot to prompt cache 2.25.904.565 I slot prompt_clear: id 0 | task -1 | clearing prompt with 0 tokens 2.25.904.601 I slot process_sing: id 1 | task -1 | saving idle slot to prompt cache 2.25.904.603 I slot prompt_clear: id 1 | task -1 | clearing prompt with 0 tokens 2.25.904.622 I slot process_sing: id 2 | task -1 | saving idle slot to prompt cache 2.25.904.622 I slot prompt_clear: id 2 | task -1 | clearing prompt with 0 tokens 2.25.904.647 I slot update_slots: id 3 | task 2143 | Checking checkpoint with [579, 579] against 603... 2.25.940.265 W slot update_slots: id 3 | task 2143 | restored context checkpoint (pos_min = 579, pos_max = 579, n_tokens = 580, n_past = 580, size = 149.626 MiB) 2.31.808.057 I slot print_timing: id 3 | task 2143 | n_decoded = 100, tg = 18.72 t/s, tg_3s = 18.72 t/s 2.32.510.029 I slot print_timing: id 3 | task 2143 | prompt eval time = 562.11 ms / 70 tokens ( 8.03 ms per token, 124.53 tokens per second) 2.32.510.033 I slot print_timing: id 3 | task 2143 | eval time = 6043.27 ms / 113 tokens ( 53.48 ms per token, 18.70 tokens per second) 2.32.510.034 I slot print_timing: id 3 | task 2143 | total time = 6605.38 ms / 183 tokens 2.32.510.035 I slot print_timing: id 3 | task 2143 | graphs reused = 2161 2.32.510.102 I slot release: id 3 | task 2143 | stop processing: n_tokens = 762, truncated = 0 2.32.510.111 I srv update_slots: all slots are idle 2.32.526.836 I srv operator(): operator(): cleaning up before exit...