[User] Bus error (core dumped) on a 65B model

[mastr-ch13f]

I tried running the a 65B model that was converted using the unversioned .py conversion script then migrated from an 8-file ggml .bin to a single-file ggjt .bin. Tried to run the model and I get a Bus error then the program ends.

user@ubuntu: ~/Desktop/llama.cpp$ ./main -m ./models/ggjt-model-model-q4_0.bin -t 7 -i
main: seed = 1680306291
llama_model_load: loading model from './models/ggjt-model-model-q4_0.bin' - please wait ...
llama_model_load: n_vocab = 32000
llama_model_load: n_ctx   = 512
llama_model_load: n_embd  = 8192
llama_model_load: n_mult  = 256
llama_model_load: n_head  = 64
llama_model_load: n_layer = 80
llama_model_load: n_rot   = 128
llama_model_load: f16     = 2
llama_model_load: n_ff    = 22016
llama_model_load: n_parts = 8
llama_model_load: type    = 4
llama_model_load: ggml map size = 38917.99 MB
llama_model_load: ggml ctx size = 201.25 KB
llama_model_load: mem required  = 41478.18 MB (+ 5120.00 MB per state)
llama_model_load: loading tensors from './models/ggjt-model-model-q4_0.bin'
llama_model_load: model size = 38917.53 MB / num tensors = 723
llama_init_from_file: kv self size  = 1280.00 MB

system_info: n_threads = 7 / 8 | AVX = 1 | AVX2 = 1 | AVX512 = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | VSX = 0 | 
main: interactive mode on.
sampling: temp = 0.800000, top_k = 40, top_p = 0.950000, repeat_last_n = 64, repeat_penalty = 1.100000
generate: n_ctx = 512, n_batch = 8, n_predict = 128, n_keep = 0


== Running in interactive mode. ==
 - Press Ctrl+C to interject at any time.
 - Press Return to return control to LLaMa.
 - If you want to submit another line, end your input in '\'.

 Bus error (core dumped)

[FNsi]

Maybe should Try with -t 8.
I think it's because hyper threading.
Try any even may work.

[Martin-HZK]

Same problem! But I encounter this when I try to reproduce the SYCL backend sample for Ubuntu Linux 24.04. I do not know why