github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 16:11:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #5708] erorr loading models x3 7900 XTX #29317

New Issue
Closed
opened 2026-04-22 08:04:47 -05:00 by GiteaMirror · 7 comments
GiteaMirror commented 2026-04-22 08:04:47 -05:00
Owner
Copy Link

Originally created by @darwinvelez58 on GitHub (Jul 15, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/5708

Originally assigned to: @dhiltgen on GitHub.

What is the issue?

Error while loading models since version 1.44 models like mixtral, llama3 70b, etc,. I have 72GB Memory with my 3 gpus and can not load this models.

This is the second week I report this issue.


[root@507c16dbc63d /]# ollama list
NAME                                    ID              SIZE    MODIFIED   
llama3:70b                              786f3184aec0    39 GB   4 days ago
mixtral:8x7b-instruct-v0.1-q4_K_S       ced521c222ac    26 GB   4 days ago
mixtral:8x7b-instruct-v0.1-q3_K_S       5ef4f821a2bb    20 GB   4 days ago
mixtral:8x7b-instruct-v0.1-q4_K_M       94bdff7afa0f    26 GB   4 days ago
phi3:latest                             d184c916657e    2.2 GB  4 days ago
mixtral:8x7b-instruct-v0.1-q6_K         ff4517993407    38 GB   5 days ago
mixtral:8x7b-instruct-v0.1-q5_K_M       b70847cf3431    32 GB   5 days ago

[root@507c16dbc63d /]# ollama run mixtral:8x7b-instruct-v0.1-q4_K_M
Error: llama runner process has terminated: signal: segmentation fault (core dumped) 
[root@507c16dbc63d /]# 


[GIN] 2024/07/15 - 20:04:42 | 200 |     788.824µs |       127.0.0.1 | GET      "/api/tags"
2024/07/15 20:05:01 routes.go:940: INFO server config env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: OLLAMA_DEBUG:false OLLAMA_FLASH_ATTENTION:false OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_LLM_LIBRARY: OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_VRAM:0 OLLAMA_MODELS:/root/.ollama/models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 OLLAMA_ORIGINS:[http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://*] OLLAMA_RUNNERS_DIR: OLLAMA_SCHED_SPREAD:false OLLAMA_TMPDIR: ROCR_VISIBLE_DEVICES:]"
time=2024-07-15T20:05:01.056Z level=INFO source=images.go:760 msg="total blobs: 23"
time=2024-07-15T20:05:01.056Z level=INFO source=images.go:767 msg="total unused blobs removed: 0"
time=2024-07-15T20:05:01.056Z level=INFO source=routes.go:987 msg="Listening on [::]:11434 (version 0.2.3)"
time=2024-07-15T20:05:01.057Z level=INFO source=payload.go:30 msg="extracting embedded files" dir=/tmp/ollama164743434/runners
time=2024-07-15T20:05:02.797Z level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cuda_v11 rocm_v60101 cpu cpu_avx cpu_avx2]"
time=2024-07-15T20:05:02.797Z level=INFO source=gpu.go:205 msg="looking for compatible GPUs"
time=2024-07-15T20:05:02.802Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-07-15T20:05:02.802Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=1 gpu_type=gfx1100
time=2024-07-15T20:05:02.803Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=2 gpu_type=gfx1100
time=2024-07-15T20:05:02.804Z level=INFO source=amd_linux.go:259 msg="unsupported Radeon iGPU detected skipping" id=3 total="512.0 MiB"
time=2024-07-15T20:05:02.804Z level=INFO source=types.go:105 msg="inference compute" id=0 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB"
time=2024-07-15T20:05:02.804Z level=INFO source=types.go:105 msg="inference compute" id=1 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB"
time=2024-07-15T20:05:02.804Z level=INFO source=types.go:105 msg="inference compute" id=2 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB"
[GIN] 2024/07/15 - 20:05:30 | 200 |      30.567µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/07/15 - 20:05:30 | 200 |    4.834722ms |       127.0.0.1 | POST     "/api/show"
time=2024-07-15T20:05:30.543Z level=INFO source=sched.go:717 msg="new model will fit in available VRAM, loading" model=/root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 library=rocm parallel=4 required="33.2 GiB"
time=2024-07-15T20:05:30.544Z level=INFO source=memory.go:309 msg="offload to rocm" layers.requested=-1 layers.model=33 layers.offload=33 layers.split=11,11,11 memory.available="[24.0 GiB 24.0 GiB 24.0 GiB]" memory.required.full="33.2 GiB" memory.required.partial="33.2 GiB" memory.required.kv="1.0 GiB" memory.required.allocations="[11.3 GiB 11.3 GiB 10.6 GiB]" memory.weights.total="25.5 GiB" memory.weights.repeating="25.4 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="1.3 GiB" memory.graph.partial="1.3 GiB"
time=2024-07-15T20:05:30.545Z level=INFO source=server.go:383 msg="starting llama server" cmd="/tmp/ollama164743434/runners/rocm_v60101/ollama_llama_server --model /root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 --ctx-size 8192 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 4 --tensor-split 11,11,11 --port 40079"
time=2024-07-15T20:05:30.545Z level=INFO source=sched.go:437 msg="loaded runners" count=1
time=2024-07-15T20:05:30.545Z level=INFO source=server.go:571 msg="waiting for llama runner to start responding"
time=2024-07-15T20:05:30.545Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server error"
INFO [main] build info | build=1 commit="a8db2a9" tid="125816579203904" timestamp=1721073930
INFO [main] system info | n_threads=16 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 0 | " tid="125816579203904" timestamp=1721073930 total_threads=32
INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="31" port="40079" tid="125816579203904" timestamp=1721073930
llama_model_loader: loaded meta data with 26 key-value pairs and 995 tensors from /root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = llama
llama_model_loader: - kv   1:                               general.name str              = mistralai
llama_model_loader: - kv   2:                       llama.context_length u32              = 32768
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   4:                          llama.block_count u32              = 32
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   9:                         llama.expert_count u32              = 8
llama_model_loader: - kv  10:                    llama.expert_used_count u32              = 2
llama_model_loader: - kv  11:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  12:                       llama.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  13:                          general.file_type u32              = 15
llama_model_loader: - kv  14:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,32000]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  16:                      tokenizer.ggml.scores arr[f32,32000]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  17:                  tokenizer.ggml.token_type arr[i32,32000]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  18:                      tokenizer.ggml.merges arr[str,58980]   = ["▁ t", "i n", "e r", "▁ a", "h e...
llama_model_loader: - kv  19:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  20:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  21:            tokenizer.ggml.unknown_token_id u32              = 0
llama_model_loader: - kv  22:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  23:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  24:                    tokenizer.chat_template str              = {{ bos_token }}{% for message in mess...
llama_model_loader: - kv  25:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type  f16:   32 tensors
llama_model_loader: - type q8_0:   64 tensors
llama_model_loader: - type q4_K:  833 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: special tokens cache size = 259
llm_load_vocab: token to piece cache size = 0.1637 MB
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32000
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: vocab_only       = 0
llm_load_print_meta: n_ctx_train      = 32768
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_layer          = 32
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_swa            = 0
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 4
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
llm_load_print_meta: f_clamp_kqv      = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: f_logit_scale    = 0.0e+00
llm_load_print_meta: n_ff             = 14336
llm_load_print_meta: n_expert         = 8
llm_load_print_meta: n_expert_used    = 2
llm_load_print_meta: causal attn      = 1
llm_load_print_meta: pooling type     = 0
llm_load_print_meta: rope type        = 0
llm_load_print_meta: rope scaling     = linear
llm_load_print_meta: freq_base_train  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 32768
llm_load_print_meta: rope_finetuned   = unknown
llm_load_print_meta: ssm_d_conv       = 0
llm_load_print_meta: ssm_d_inner      = 0
llm_load_print_meta: ssm_d_state      = 0
llm_load_print_meta: ssm_dt_rank      = 0
llm_load_print_meta: model type       = 8x7B
llm_load_print_meta: model ftype      = Q4_K - Medium
llm_load_print_meta: model params     = 46.70 B
llm_load_print_meta: model size       = 24.62 GiB (4.53 BPW) 
llm_load_print_meta: general.name     = mistralai
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 2 '</s>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: LF token         = 13 '<0x0A>'
llm_load_print_meta: max token length = 48
time=2024-07-15T20:05:30.797Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model"
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 3 ROCm devices:
  Device 0: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
  Device 1: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
  Device 2: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
llm_load_tensors: ggml ctx size =    1.53 MiB
llm_load_tensors: offloading 32 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 33/33 layers to GPU
llm_load_tensors:      ROCm0 buffer size =  8608.53 MiB
llm_load_tensors:      ROCm1 buffer size =  8608.53 MiB
llm_load_tensors:      ROCm2 buffer size =  7928.49 MiB
llm_load_tensors:  ROCm_Host buffer size =    70.31 MiB
time=2024-07-15T20:05:42.357Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server not responding"
time=2024-07-15T20:05:42.717Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model"
llama_new_context_with_model: n_ctx      = 8192
llama_new_context_with_model: n_batch    = 512
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: flash_attn = 0
llama_new_context_with_model: freq_base  = 1000000.0
llama_new_context_with_model: freq_scale = 1
time=2024-07-15T20:05:44.171Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server not responding"
time=2024-07-15T20:05:44.534Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model"
llama_kv_cache_init:      ROCm0 KV buffer size =   352.00 MiB
llama_kv_cache_init:      ROCm1 KV buffer size =   352.00 MiB
llama_kv_cache_init:      ROCm2 KV buffer size =   320.00 MiB
llama_new_context_with_model: KV self size  = 1024.00 MiB, K (f16):  512.00 MiB, V (f16):  512.00 MiB
llama_new_context_with_model:  ROCm_Host  output buffer size =     0.55 MiB
llama_new_context_with_model: pipeline parallelism enabled (n_copies=4)
llama_new_context_with_model:      ROCm0 compute buffer size =   640.01 MiB
llama_new_context_with_model:      ROCm1 compute buffer size =   640.01 MiB
llama_new_context_with_model:      ROCm2 compute buffer size =   640.02 MiB
llama_new_context_with_model:  ROCm_Host compute buffer size =    72.02 MiB
llama_new_context_with_model: graph nodes  = 1510
llama_new_context_with_model: graph splits = 4
time=2024-07-15T20:05:46.122Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server error"
time=2024-07-15T20:05:46.372Z level=ERROR source=sched.go:443 msg="error loading llama server" error="llama runner process has terminated: signal: segmentation fault (core dumped) "
[GIN] 2024/07/15 - 20:05:46 | 500 | 15.846603317s |       127.0.0.1 | POST     "/api/chat"
time=2024-07-15T20:05:51.373Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.000554258 model=/root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8
time=2024-07-15T20:05:51.624Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.251135129 model=/root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8
time=2024-07-15T20:05:51.874Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.500996565 model=/root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8

OS

Linux

GPU

AMD

CPU

AMD

Ollama version

0.2.3-rocm

Originally created by @darwinvelez58 on GitHub (Jul 15, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/5708 Originally assigned to: @dhiltgen on GitHub. ### What is the issue? Error while loading models since version 1.44 models like mixtral, llama3 70b, etc,. I have 72GB Memory with my 3 gpus and can not load this models. This is the second week I report this issue. ``` [root@507c16dbc63d /]# ollama list NAME ID SIZE MODIFIED llama3:70b 786f3184aec0 39 GB 4 days ago mixtral:8x7b-instruct-v0.1-q4_K_S ced521c222ac 26 GB 4 days ago mixtral:8x7b-instruct-v0.1-q3_K_S 5ef4f821a2bb 20 GB 4 days ago mixtral:8x7b-instruct-v0.1-q4_K_M 94bdff7afa0f 26 GB 4 days ago phi3:latest d184c916657e 2.2 GB 4 days ago mixtral:8x7b-instruct-v0.1-q6_K ff4517993407 38 GB 5 days ago mixtral:8x7b-instruct-v0.1-q5_K_M b70847cf3431 32 GB 5 days ago [root@507c16dbc63d /]# ollama run mixtral:8x7b-instruct-v0.1-q4_K_M Error: llama runner process has terminated: signal: segmentation fault (core dumped) [root@507c16dbc63d /]# [GIN] 2024/07/15 - 20:04:42 | 200 | 788.824µs | 127.0.0.1 | GET "/api/tags" 2024/07/15 20:05:01 routes.go:940: INFO server config env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: OLLAMA_DEBUG:false OLLAMA_FLASH_ATTENTION:false OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_LLM_LIBRARY: OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_VRAM:0 OLLAMA_MODELS:/root/.ollama/models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 OLLAMA_ORIGINS:[http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://*] OLLAMA_RUNNERS_DIR: OLLAMA_SCHED_SPREAD:false OLLAMA_TMPDIR: ROCR_VISIBLE_DEVICES:]" time=2024-07-15T20:05:01.056Z level=INFO source=images.go:760 msg="total blobs: 23" time=2024-07-15T20:05:01.056Z level=INFO source=images.go:767 msg="total unused blobs removed: 0" time=2024-07-15T20:05:01.056Z level=INFO source=routes.go:987 msg="Listening on [::]:11434 (version 0.2.3)" time=2024-07-15T20:05:01.057Z level=INFO source=payload.go:30 msg="extracting embedded files" dir=/tmp/ollama164743434/runners time=2024-07-15T20:05:02.797Z level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cuda_v11 rocm_v60101 cpu cpu_avx cpu_avx2]" time=2024-07-15T20:05:02.797Z level=INFO source=gpu.go:205 msg="looking for compatible GPUs" time=2024-07-15T20:05:02.802Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100 time=2024-07-15T20:05:02.802Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=1 gpu_type=gfx1100 time=2024-07-15T20:05:02.803Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=2 gpu_type=gfx1100 time=2024-07-15T20:05:02.804Z level=INFO source=amd_linux.go:259 msg="unsupported Radeon iGPU detected skipping" id=3 total="512.0 MiB" time=2024-07-15T20:05:02.804Z level=INFO source=types.go:105 msg="inference compute" id=0 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB" time=2024-07-15T20:05:02.804Z level=INFO source=types.go:105 msg="inference compute" id=1 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB" time=2024-07-15T20:05:02.804Z level=INFO source=types.go:105 msg="inference compute" id=2 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB" [GIN] 2024/07/15 - 20:05:30 | 200 | 30.567µs | 127.0.0.1 | HEAD "/" [GIN] 2024/07/15 - 20:05:30 | 200 | 4.834722ms | 127.0.0.1 | POST "/api/show" time=2024-07-15T20:05:30.543Z level=INFO source=sched.go:717 msg="new model will fit in available VRAM, loading" model=/root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 library=rocm parallel=4 required="33.2 GiB" time=2024-07-15T20:05:30.544Z level=INFO source=memory.go:309 msg="offload to rocm" layers.requested=-1 layers.model=33 layers.offload=33 layers.split=11,11,11 memory.available="[24.0 GiB 24.0 GiB 24.0 GiB]" memory.required.full="33.2 GiB" memory.required.partial="33.2 GiB" memory.required.kv="1.0 GiB" memory.required.allocations="[11.3 GiB 11.3 GiB 10.6 GiB]" memory.weights.total="25.5 GiB" memory.weights.repeating="25.4 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="1.3 GiB" memory.graph.partial="1.3 GiB" time=2024-07-15T20:05:30.545Z level=INFO source=server.go:383 msg="starting llama server" cmd="/tmp/ollama164743434/runners/rocm_v60101/ollama_llama_server --model /root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 --ctx-size 8192 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 4 --tensor-split 11,11,11 --port 40079" time=2024-07-15T20:05:30.545Z level=INFO source=sched.go:437 msg="loaded runners" count=1 time=2024-07-15T20:05:30.545Z level=INFO source=server.go:571 msg="waiting for llama runner to start responding" time=2024-07-15T20:05:30.545Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server error" INFO [main] build info | build=1 commit="a8db2a9" tid="125816579203904" timestamp=1721073930 INFO [main] system info | n_threads=16 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 0 | " tid="125816579203904" timestamp=1721073930 total_threads=32 INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="31" port="40079" tid="125816579203904" timestamp=1721073930 llama_model_loader: loaded meta data with 26 key-value pairs and 995 tensors from /root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = llama llama_model_loader: - kv 1: general.name str = mistralai llama_model_loader: - kv 2: llama.context_length u32 = 32768 llama_model_loader: - kv 3: llama.embedding_length u32 = 4096 llama_model_loader: - kv 4: llama.block_count u32 = 32 llama_model_loader: - kv 5: llama.feed_forward_length u32 = 14336 llama_model_loader: - kv 6: llama.rope.dimension_count u32 = 128 llama_model_loader: - kv 7: llama.attention.head_count u32 = 32 llama_model_loader: - kv 8: llama.attention.head_count_kv u32 = 8 llama_model_loader: - kv 9: llama.expert_count u32 = 8 llama_model_loader: - kv 10: llama.expert_used_count u32 = 2 llama_model_loader: - kv 11: llama.attention.layer_norm_rms_epsilon f32 = 0.000010 llama_model_loader: - kv 12: llama.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 13: general.file_type u32 = 15 llama_model_loader: - kv 14: tokenizer.ggml.model str = llama llama_model_loader: - kv 15: tokenizer.ggml.tokens arr[str,32000] = ["<unk>", "<s>", "</s>", "<0x00>", "<... llama_model_loader: - kv 16: tokenizer.ggml.scores arr[f32,32000] = [0.000000, 0.000000, 0.000000, 0.0000... llama_model_loader: - kv 17: tokenizer.ggml.token_type arr[i32,32000] = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ... llama_model_loader: - kv 18: tokenizer.ggml.merges arr[str,58980] = ["▁ t", "i n", "e r", "▁ a", "h e... llama_model_loader: - kv 19: tokenizer.ggml.bos_token_id u32 = 1 llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 2 llama_model_loader: - kv 21: tokenizer.ggml.unknown_token_id u32 = 0 llama_model_loader: - kv 22: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 23: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 24: tokenizer.chat_template str = {{ bos_token }}{% for message in mess... llama_model_loader: - kv 25: general.quantization_version u32 = 2 llama_model_loader: - type f32: 65 tensors llama_model_loader: - type f16: 32 tensors llama_model_loader: - type q8_0: 64 tensors llama_model_loader: - type q4_K: 833 tensors llama_model_loader: - type q6_K: 1 tensors llm_load_vocab: special tokens cache size = 259 llm_load_vocab: token to piece cache size = 0.1637 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = llama llm_load_print_meta: vocab type = SPM llm_load_print_meta: n_vocab = 32000 llm_load_print_meta: n_merges = 0 llm_load_print_meta: vocab_only = 0 llm_load_print_meta: n_ctx_train = 32768 llm_load_print_meta: n_embd = 4096 llm_load_print_meta: n_layer = 32 llm_load_print_meta: n_head = 32 llm_load_print_meta: n_head_kv = 8 llm_load_print_meta: n_rot = 128 llm_load_print_meta: n_swa = 0 llm_load_print_meta: n_embd_head_k = 128 llm_load_print_meta: n_embd_head_v = 128 llm_load_print_meta: n_gqa = 4 llm_load_print_meta: n_embd_k_gqa = 1024 llm_load_print_meta: n_embd_v_gqa = 1024 llm_load_print_meta: f_norm_eps = 0.0e+00 llm_load_print_meta: f_norm_rms_eps = 1.0e-05 llm_load_print_meta: f_clamp_kqv = 0.0e+00 llm_load_print_meta: f_max_alibi_bias = 0.0e+00 llm_load_print_meta: f_logit_scale = 0.0e+00 llm_load_print_meta: n_ff = 14336 llm_load_print_meta: n_expert = 8 llm_load_print_meta: n_expert_used = 2 llm_load_print_meta: causal attn = 1 llm_load_print_meta: pooling type = 0 llm_load_print_meta: rope type = 0 llm_load_print_meta: rope scaling = linear llm_load_print_meta: freq_base_train = 1000000.0 llm_load_print_meta: freq_scale_train = 1 llm_load_print_meta: n_ctx_orig_yarn = 32768 llm_load_print_meta: rope_finetuned = unknown llm_load_print_meta: ssm_d_conv = 0 llm_load_print_meta: ssm_d_inner = 0 llm_load_print_meta: ssm_d_state = 0 llm_load_print_meta: ssm_dt_rank = 0 llm_load_print_meta: model type = 8x7B llm_load_print_meta: model ftype = Q4_K - Medium llm_load_print_meta: model params = 46.70 B llm_load_print_meta: model size = 24.62 GiB (4.53 BPW) llm_load_print_meta: general.name = mistralai llm_load_print_meta: BOS token = 1 '<s>' llm_load_print_meta: EOS token = 2 '</s>' llm_load_print_meta: UNK token = 0 '<unk>' llm_load_print_meta: LF token = 13 '<0x0A>' llm_load_print_meta: max token length = 48 time=2024-07-15T20:05:30.797Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model" ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 3 ROCm devices: Device 0: Radeon RX 7900 XTX, compute capability 11.0, VMM: no Device 1: Radeon RX 7900 XTX, compute capability 11.0, VMM: no Device 2: Radeon RX 7900 XTX, compute capability 11.0, VMM: no llm_load_tensors: ggml ctx size = 1.53 MiB llm_load_tensors: offloading 32 repeating layers to GPU llm_load_tensors: offloading non-repeating layers to GPU llm_load_tensors: offloaded 33/33 layers to GPU llm_load_tensors: ROCm0 buffer size = 8608.53 MiB llm_load_tensors: ROCm1 buffer size = 8608.53 MiB llm_load_tensors: ROCm2 buffer size = 7928.49 MiB llm_load_tensors: ROCm_Host buffer size = 70.31 MiB time=2024-07-15T20:05:42.357Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server not responding" time=2024-07-15T20:05:42.717Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model" llama_new_context_with_model: n_ctx = 8192 llama_new_context_with_model: n_batch = 512 llama_new_context_with_model: n_ubatch = 512 llama_new_context_with_model: flash_attn = 0 llama_new_context_with_model: freq_base = 1000000.0 llama_new_context_with_model: freq_scale = 1 time=2024-07-15T20:05:44.171Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server not responding" time=2024-07-15T20:05:44.534Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model" llama_kv_cache_init: ROCm0 KV buffer size = 352.00 MiB llama_kv_cache_init: ROCm1 KV buffer size = 352.00 MiB llama_kv_cache_init: ROCm2 KV buffer size = 320.00 MiB llama_new_context_with_model: KV self size = 1024.00 MiB, K (f16): 512.00 MiB, V (f16): 512.00 MiB llama_new_context_with_model: ROCm_Host output buffer size = 0.55 MiB llama_new_context_with_model: pipeline parallelism enabled (n_copies=4) llama_new_context_with_model: ROCm0 compute buffer size = 640.01 MiB llama_new_context_with_model: ROCm1 compute buffer size = 640.01 MiB llama_new_context_with_model: ROCm2 compute buffer size = 640.02 MiB llama_new_context_with_model: ROCm_Host compute buffer size = 72.02 MiB llama_new_context_with_model: graph nodes = 1510 llama_new_context_with_model: graph splits = 4 time=2024-07-15T20:05:46.122Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server error" time=2024-07-15T20:05:46.372Z level=ERROR source=sched.go:443 msg="error loading llama server" error="llama runner process has terminated: signal: segmentation fault (core dumped) " [GIN] 2024/07/15 - 20:05:46 | 500 | 15.846603317s | 127.0.0.1 | POST "/api/chat" time=2024-07-15T20:05:51.373Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.000554258 model=/root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 time=2024-07-15T20:05:51.624Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.251135129 model=/root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 time=2024-07-15T20:05:51.874Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.500996565 model=/root/.ollama/models/blobs/sha256-3a17f7cde150070bbc815645693fb93c311cc42e7deaf198364acadcf08458f8 ``` ### OS Linux ### GPU AMD ### CPU AMD ### Ollama version 0.2.3-rocm
GiteaMirror added the gpuamdbug labels 2026-04-22 08:04:47 -05:00
GiteaMirror commented 2026-04-22 08:04:52 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Jul 15, 2024):

@darwinvelez58 can you confirm 0.1.43 does work on your setup? (or clarify which version you were running before upgrading to 0.1.44) My suspicion is this is a regression in llama.cpp, but narrowing down to when it regressed may help finding the root cause.

<!-- gh-comment-id:2229340692 --> @dhiltgen commented on GitHub (Jul 15, 2024): @darwinvelez58 can you confirm 0.1.43 does work on your setup? (or clarify which version you were running before upgrading to 0.1.44) My suspicion is this is a regression in llama.cpp, but narrowing down to when it regressed may help finding the root cause.
GiteaMirror commented 2026-04-22 08:04:52 -05:00
Author
Owner
Copy Link

@darwinvelez58 commented on GitHub (Jul 15, 2024):

On version 0.1.44 I can load the model in version >1.44 I can not.

[root@df83bcc21230 /]# ollama list
NAME                                    ID              SIZE    MODIFIED   
llama3:70b                              786f3184aec0    39 GB   4 days ago
mixtral:8x7b-instruct-v0.1-q4_K_S       ced521c222ac    26 GB   4 days ago
mixtral:8x7b-instruct-v0.1-q3_K_S       5ef4f821a2bb    20 GB   4 days ago
mixtral:8x7b-instruct-v0.1-q4_K_M       94bdff7afa0f    26 GB   4 days ago
phi3:latest                             d184c916657e    2.2 GB  5 days ago
mixtral:8x7b-instruct-v0.1-q6_K         ff4517993407    38 GB   5 days ago
mixtral:8x7b-instruct-v0.1-q5_K_M       b70847cf3431    32 GB   5 days ago
[root@df83bcc21230 /]# mixtral:8x7b-instruct-v0.1-q4_K_S^C
[root@df83bcc21230 /]# ollama run mixtral:8x7b-instruct-v0.1-q4_K_S
>>> Send a message (/? for help)

GIN] 2024/07/15 - 21:21:30 | 200 |     364.856µs |       127.0.0.1 | POST     "/api/show"
[GIN] 2024/07/15 - 21:21:30 | 200 |     275.637µs |       127.0.0.1 | POST     "/api/show"
time=2024-07-15T21:21:30.425Z level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-07-15T21:21:30.425Z level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=1 gpu_type=gfx1100
time=2024-07-15T21:21:30.425Z level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=2 gpu_type=gfx1100
time=2024-07-15T21:21:30.425Z level=INFO source=amd_linux.go:233 msg="unsupported Radeon iGPU detected skipping" id=3 total="512.0 MiB"
time=2024-07-15T21:21:30.641Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=28 memory.available="24.0 GiB" memory.required.full="26.2 GiB" memory.required.partial="23.7 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="184.0 MiB" memory.graph.partial="935.0 MiB"
time=2024-07-15T21:21:30.642Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=28 memory.available="24.0 GiB" memory.required.full="26.2 GiB" memory.required.partial="23.7 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="184.0 MiB" memory.graph.partial="935.0 MiB"
time=2024-07-15T21:21:30.644Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=28 memory.available="24.0 GiB" memory.required.full="26.2 GiB" memory.required.partial="23.7 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="184.0 MiB" memory.graph.partial="935.0 MiB"
time=2024-07-15T21:21:30.645Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=33 memory.available="71.9 GiB" memory.required.full="26.6 GiB" memory.required.partial="26.6 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="552.1 MiB" memory.graph.partial="2.7 GiB"
time=2024-07-15T21:21:30.646Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=33 memory.available="71.9 GiB" memory.required.full="26.6 GiB" memory.required.partial="26.6 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="552.1 MiB" memory.graph.partial="2.7 GiB"
time=2024-07-15T21:21:30.646Z level=INFO source=server.go:341 msg="starting llama server" cmd="/tmp/ollama1056339174/runners/rocm_v60002/ollama_llama_server --model /root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 --ctx-size 2048 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 1 --port 45657"
time=2024-07-15T21:21:30.646Z level=INFO source=sched.go:338 msg="loaded runners" count=1
time=2024-07-15T21:21:30.646Z level=INFO source=server.go:529 msg="waiting for llama runner to start responding"
time=2024-07-15T21:21:30.646Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server error"
INFO [main] build info | build=1 commit="5921b8f" tid="123317090335808" timestamp=1721078490
INFO [main] system info | n_threads=16 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | " tid="123317090335808" timestamp=1721078490 total_threads=32
INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="31" port="45657" tid="123317090335808" timestamp=1721078490
llama_model_loader: loaded meta data with 26 key-value pairs and 995 tensors from /root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = llama
llama_model_loader: - kv   1:                               general.name str              = mistralai
llama_model_loader: - kv   2:                       llama.context_length u32              = 32768
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   4:                          llama.block_count u32              = 32
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   9:                         llama.expert_count u32              = 8
llama_model_loader: - kv  10:                    llama.expert_used_count u32              = 2
llama_model_loader: - kv  11:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  12:                       llama.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  13:                          general.file_type u32              = 14
llama_model_loader: - kv  14:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,32000]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  16:                      tokenizer.ggml.scores arr[f32,32000]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  17:                  tokenizer.ggml.token_type arr[i32,32000]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  18:                      tokenizer.ggml.merges arr[str,58980]   = ["▁ t", "i n", "e r", "▁ a", "h e...
llama_model_loader: - kv  19:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  20:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  21:            tokenizer.ggml.unknown_token_id u32              = 0
llama_model_loader: - kv  22:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  23:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  24:                    tokenizer.chat_template str              = {{ bos_token }}{% for message in mess...
llama_model_loader: - kv  25:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type  f16:   32 tensors
llama_model_loader: - type q8_0:   64 tensors
llama_model_loader: - type q4_K:  833 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: special tokens cache size = 259
llm_load_vocab: token to piece cache size = 0.3274 MB
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32000
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 32768
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 32
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 4
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
llm_load_print_meta: f_clamp_kqv      = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: f_logit_scale    = 0.0e+00
llm_load_print_meta: n_ff             = 14336
llm_load_print_meta: n_expert         = 8
llm_load_print_meta: n_expert_used    = 2
llm_load_print_meta: causal attn      = 1
llm_load_print_meta: pooling type     = 0
llm_load_print_meta: rope type        = 0
llm_load_print_meta: rope scaling     = linear
llm_load_print_meta: freq_base_train  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 32768
llm_load_print_meta: rope_finetuned   = unknown
llm_load_print_meta: ssm_d_conv       = 0
llm_load_print_meta: ssm_d_inner      = 0
llm_load_print_meta: ssm_d_state      = 0
llm_load_print_meta: ssm_dt_rank      = 0
llm_load_print_meta: model type       = 8x7B
llm_load_print_meta: model ftype      = Q4_K - Small
llm_load_print_meta: model params     = 46.70 B
llm_load_print_meta: model size       = 24.62 GiB (4.53 BPW) 
llm_load_print_meta: general.name     = mistralai
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 2 '</s>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: LF token         = 13 '<0x0A>'
time=2024-07-15T21:21:30.897Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server loading model"
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:   no
ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes
ggml_cuda_init: found 3 ROCm devices:
  Device 0: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
  Device 1: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
  Device 2: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
llm_load_tensors: ggml ctx size =    1.67 MiB
time=2024-07-15T21:21:33.605Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server not responding"
llm_load_tensors: offloading 32 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 33/33 layers to GPU
llm_load_tensors:      ROCm0 buffer size =  8608.53 MiB
llm_load_tensors:      ROCm1 buffer size =  8608.53 MiB
llm_load_tensors:      ROCm2 buffer size =  7928.49 MiB
llm_load_tensors:  ROCm_Host buffer size =    70.31 MiB
time=2024-07-15T21:21:33.856Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server loading model"
time=2024-07-15T21:21:41.582Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server not responding"
llama_new_context_with_model: n_ctx      = 2048
llama_new_context_with_model: n_batch    = 512
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: flash_attn = 0
llama_new_context_with_model: freq_base  = 1000000.0
llama_new_context_with_model: freq_scale = 1
time=2024-07-15T21:21:42.320Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server loading model"
llama_kv_cache_init:      ROCm0 KV buffer size =    88.00 MiB
llama_kv_cache_init:      ROCm1 KV buffer size =    88.00 MiB
llama_kv_cache_init:      ROCm2 KV buffer size =    80.00 MiB
llama_new_context_with_model: KV self size  =  256.00 MiB, K (f16):  128.00 MiB, V (f16):  128.00 MiB
llama_new_context_with_model:  ROCm_Host  output buffer size =     0.14 MiB
llama_new_context_with_model: pipeline parallelism enabled (n_copies=4)
llama_new_context_with_model:      ROCm0 compute buffer size =   208.01 MiB
llama_new_context_with_model:      ROCm1 compute buffer size =   208.01 MiB
llama_new_context_with_model:      ROCm2 compute buffer size =   208.02 MiB
llama_new_context_with_model:  ROCm_Host compute buffer size =    24.02 MiB
llama_new_context_with_model: graph nodes  = 1510
llama_new_context_with_model: graph splits = 4
INFO [main] model loaded | tid="123317090335808" timestamp=1721078506
time=2024-07-15T21:21:46.942Z level=INFO source=server.go:572 msg="llama runner started in 16.30 seconds"
[GIN] 2024/07/15 - 21:21:46 | 200 | 16.521933151s |       127.0.0.1 | POST     "/api/chat"



Every 0.1s: /opt/rocm/bin/rocm-smi                                                                                                             ai1: Mon Jul 15 23:22:44 2024



=========================================== ROCm System Management Interface ===========================================
===================================================== Concise Info =====================================================
Device  Node  IDs              Temp    Power    Partitions          SCLK  MCLK     Fan  Perf  PwrCap       VRAM%  GPU%
^[3m              (DID,     GUID)  (Edge)  (Avg)    (Mem, Compute, ID)                                                      ^[0m
========================================================================================================================
0       1     0x744c,   48669  35.0°C  18.0W    N/A, N/A, 0         0Mhz  96Mhz    0%   auto  315.0W       39%    0%
1       2     0x744c,   34500  34.0°C  16.0W    N/A, N/A, 0         0Mhz  96Mhz    0%   auto  339.0W       39%    0%
2       3     0x744c,   11926  34.0°C  17.0W    N/A, N/A, 0         0Mhz  96Mhz    0%   auto  339.0W       36%    0%
3       4     0x164e,   27864  48.0°C  32.042W  N/A, N/A, 0         None  1800Mhz  0%   auto  Unsupported  7%     0%
========================================================================================================================
================================================= End of ROCm SMI Log ==================================================



With 1.45>= I got  "Error: llama runner process has terminated: signal: segmentation fault (core dumped)"
<!-- gh-comment-id:2229466035 --> @darwinvelez58 commented on GitHub (Jul 15, 2024): On version 0.1.44 I can load the model in version >1.44 I can not. ``` [root@df83bcc21230 /]# ollama list NAME ID SIZE MODIFIED llama3:70b 786f3184aec0 39 GB 4 days ago mixtral:8x7b-instruct-v0.1-q4_K_S ced521c222ac 26 GB 4 days ago mixtral:8x7b-instruct-v0.1-q3_K_S 5ef4f821a2bb 20 GB 4 days ago mixtral:8x7b-instruct-v0.1-q4_K_M 94bdff7afa0f 26 GB 4 days ago phi3:latest d184c916657e 2.2 GB 5 days ago mixtral:8x7b-instruct-v0.1-q6_K ff4517993407 38 GB 5 days ago mixtral:8x7b-instruct-v0.1-q5_K_M b70847cf3431 32 GB 5 days ago [root@df83bcc21230 /]# mixtral:8x7b-instruct-v0.1-q4_K_S^C [root@df83bcc21230 /]# ollama run mixtral:8x7b-instruct-v0.1-q4_K_S >>> Send a message (/? for help) GIN] 2024/07/15 - 21:21:30 | 200 | 364.856µs | 127.0.0.1 | POST "/api/show" [GIN] 2024/07/15 - 21:21:30 | 200 | 275.637µs | 127.0.0.1 | POST "/api/show" time=2024-07-15T21:21:30.425Z level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100 time=2024-07-15T21:21:30.425Z level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=1 gpu_type=gfx1100 time=2024-07-15T21:21:30.425Z level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=2 gpu_type=gfx1100 time=2024-07-15T21:21:30.425Z level=INFO source=amd_linux.go:233 msg="unsupported Radeon iGPU detected skipping" id=3 total="512.0 MiB" time=2024-07-15T21:21:30.641Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=28 memory.available="24.0 GiB" memory.required.full="26.2 GiB" memory.required.partial="23.7 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="184.0 MiB" memory.graph.partial="935.0 MiB" time=2024-07-15T21:21:30.642Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=28 memory.available="24.0 GiB" memory.required.full="26.2 GiB" memory.required.partial="23.7 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="184.0 MiB" memory.graph.partial="935.0 MiB" time=2024-07-15T21:21:30.644Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=28 memory.available="24.0 GiB" memory.required.full="26.2 GiB" memory.required.partial="23.7 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="184.0 MiB" memory.graph.partial="935.0 MiB" time=2024-07-15T21:21:30.645Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=33 memory.available="71.9 GiB" memory.required.full="26.6 GiB" memory.required.partial="26.6 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="552.1 MiB" memory.graph.partial="2.7 GiB" time=2024-07-15T21:21:30.646Z level=INFO source=memory.go:133 msg="offload to gpu" layers.requested=-1 layers.real=33 memory.available="71.9 GiB" memory.required.full="26.6 GiB" memory.required.partial="26.6 GiB" memory.required.kv="256.0 MiB" memory.weights.total="24.6 GiB" memory.weights.repeating="24.5 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="552.1 MiB" memory.graph.partial="2.7 GiB" time=2024-07-15T21:21:30.646Z level=INFO source=server.go:341 msg="starting llama server" cmd="/tmp/ollama1056339174/runners/rocm_v60002/ollama_llama_server --model /root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 --ctx-size 2048 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 1 --port 45657" time=2024-07-15T21:21:30.646Z level=INFO source=sched.go:338 msg="loaded runners" count=1 time=2024-07-15T21:21:30.646Z level=INFO source=server.go:529 msg="waiting for llama runner to start responding" time=2024-07-15T21:21:30.646Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server error" INFO [main] build info | build=1 commit="5921b8f" tid="123317090335808" timestamp=1721078490 INFO [main] system info | n_threads=16 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | " tid="123317090335808" timestamp=1721078490 total_threads=32 INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="31" port="45657" tid="123317090335808" timestamp=1721078490 llama_model_loader: loaded meta data with 26 key-value pairs and 995 tensors from /root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = llama llama_model_loader: - kv 1: general.name str = mistralai llama_model_loader: - kv 2: llama.context_length u32 = 32768 llama_model_loader: - kv 3: llama.embedding_length u32 = 4096 llama_model_loader: - kv 4: llama.block_count u32 = 32 llama_model_loader: - kv 5: llama.feed_forward_length u32 = 14336 llama_model_loader: - kv 6: llama.rope.dimension_count u32 = 128 llama_model_loader: - kv 7: llama.attention.head_count u32 = 32 llama_model_loader: - kv 8: llama.attention.head_count_kv u32 = 8 llama_model_loader: - kv 9: llama.expert_count u32 = 8 llama_model_loader: - kv 10: llama.expert_used_count u32 = 2 llama_model_loader: - kv 11: llama.attention.layer_norm_rms_epsilon f32 = 0.000010 llama_model_loader: - kv 12: llama.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 13: general.file_type u32 = 14 llama_model_loader: - kv 14: tokenizer.ggml.model str = llama llama_model_loader: - kv 15: tokenizer.ggml.tokens arr[str,32000] = ["<unk>", "<s>", "</s>", "<0x00>", "<... llama_model_loader: - kv 16: tokenizer.ggml.scores arr[f32,32000] = [0.000000, 0.000000, 0.000000, 0.0000... llama_model_loader: - kv 17: tokenizer.ggml.token_type arr[i32,32000] = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ... llama_model_loader: - kv 18: tokenizer.ggml.merges arr[str,58980] = ["▁ t", "i n", "e r", "▁ a", "h e... llama_model_loader: - kv 19: tokenizer.ggml.bos_token_id u32 = 1 llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 2 llama_model_loader: - kv 21: tokenizer.ggml.unknown_token_id u32 = 0 llama_model_loader: - kv 22: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 23: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 24: tokenizer.chat_template str = {{ bos_token }}{% for message in mess... llama_model_loader: - kv 25: general.quantization_version u32 = 2 llama_model_loader: - type f32: 65 tensors llama_model_loader: - type f16: 32 tensors llama_model_loader: - type q8_0: 64 tensors llama_model_loader: - type q4_K: 833 tensors llama_model_loader: - type q6_K: 1 tensors llm_load_vocab: special tokens cache size = 259 llm_load_vocab: token to piece cache size = 0.3274 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = llama llm_load_print_meta: vocab type = SPM llm_load_print_meta: n_vocab = 32000 llm_load_print_meta: n_merges = 0 llm_load_print_meta: n_ctx_train = 32768 llm_load_print_meta: n_embd = 4096 llm_load_print_meta: n_head = 32 llm_load_print_meta: n_head_kv = 8 llm_load_print_meta: n_layer = 32 llm_load_print_meta: n_rot = 128 llm_load_print_meta: n_embd_head_k = 128 llm_load_print_meta: n_embd_head_v = 128 llm_load_print_meta: n_gqa = 4 llm_load_print_meta: n_embd_k_gqa = 1024 llm_load_print_meta: n_embd_v_gqa = 1024 llm_load_print_meta: f_norm_eps = 0.0e+00 llm_load_print_meta: f_norm_rms_eps = 1.0e-05 llm_load_print_meta: f_clamp_kqv = 0.0e+00 llm_load_print_meta: f_max_alibi_bias = 0.0e+00 llm_load_print_meta: f_logit_scale = 0.0e+00 llm_load_print_meta: n_ff = 14336 llm_load_print_meta: n_expert = 8 llm_load_print_meta: n_expert_used = 2 llm_load_print_meta: causal attn = 1 llm_load_print_meta: pooling type = 0 llm_load_print_meta: rope type = 0 llm_load_print_meta: rope scaling = linear llm_load_print_meta: freq_base_train = 1000000.0 llm_load_print_meta: freq_scale_train = 1 llm_load_print_meta: n_yarn_orig_ctx = 32768 llm_load_print_meta: rope_finetuned = unknown llm_load_print_meta: ssm_d_conv = 0 llm_load_print_meta: ssm_d_inner = 0 llm_load_print_meta: ssm_d_state = 0 llm_load_print_meta: ssm_dt_rank = 0 llm_load_print_meta: model type = 8x7B llm_load_print_meta: model ftype = Q4_K - Small llm_load_print_meta: model params = 46.70 B llm_load_print_meta: model size = 24.62 GiB (4.53 BPW) llm_load_print_meta: general.name = mistralai llm_load_print_meta: BOS token = 1 '<s>' llm_load_print_meta: EOS token = 2 '</s>' llm_load_print_meta: UNK token = 0 '<unk>' llm_load_print_meta: LF token = 13 '<0x0A>' time=2024-07-15T21:21:30.897Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server loading model" ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes ggml_cuda_init: found 3 ROCm devices: Device 0: Radeon RX 7900 XTX, compute capability 11.0, VMM: no Device 1: Radeon RX 7900 XTX, compute capability 11.0, VMM: no Device 2: Radeon RX 7900 XTX, compute capability 11.0, VMM: no llm_load_tensors: ggml ctx size = 1.67 MiB time=2024-07-15T21:21:33.605Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server not responding" llm_load_tensors: offloading 32 repeating layers to GPU llm_load_tensors: offloading non-repeating layers to GPU llm_load_tensors: offloaded 33/33 layers to GPU llm_load_tensors: ROCm0 buffer size = 8608.53 MiB llm_load_tensors: ROCm1 buffer size = 8608.53 MiB llm_load_tensors: ROCm2 buffer size = 7928.49 MiB llm_load_tensors: ROCm_Host buffer size = 70.31 MiB time=2024-07-15T21:21:33.856Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server loading model" time=2024-07-15T21:21:41.582Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server not responding" llama_new_context_with_model: n_ctx = 2048 llama_new_context_with_model: n_batch = 512 llama_new_context_with_model: n_ubatch = 512 llama_new_context_with_model: flash_attn = 0 llama_new_context_with_model: freq_base = 1000000.0 llama_new_context_with_model: freq_scale = 1 time=2024-07-15T21:21:42.320Z level=INFO source=server.go:567 msg="waiting for server to become available" status="llm server loading model" llama_kv_cache_init: ROCm0 KV buffer size = 88.00 MiB llama_kv_cache_init: ROCm1 KV buffer size = 88.00 MiB llama_kv_cache_init: ROCm2 KV buffer size = 80.00 MiB llama_new_context_with_model: KV self size = 256.00 MiB, K (f16): 128.00 MiB, V (f16): 128.00 MiB llama_new_context_with_model: ROCm_Host output buffer size = 0.14 MiB llama_new_context_with_model: pipeline parallelism enabled (n_copies=4) llama_new_context_with_model: ROCm0 compute buffer size = 208.01 MiB llama_new_context_with_model: ROCm1 compute buffer size = 208.01 MiB llama_new_context_with_model: ROCm2 compute buffer size = 208.02 MiB llama_new_context_with_model: ROCm_Host compute buffer size = 24.02 MiB llama_new_context_with_model: graph nodes = 1510 llama_new_context_with_model: graph splits = 4 INFO [main] model loaded | tid="123317090335808" timestamp=1721078506 time=2024-07-15T21:21:46.942Z level=INFO source=server.go:572 msg="llama runner started in 16.30 seconds" [GIN] 2024/07/15 - 21:21:46 | 200 | 16.521933151s | 127.0.0.1 | POST "/api/chat" Every 0.1s: /opt/rocm/bin/rocm-smi ai1: Mon Jul 15 23:22:44 2024 =========================================== ROCm System Management Interface =========================================== ===================================================== Concise Info ===================================================== Device Node IDs Temp Power Partitions SCLK MCLK Fan Perf PwrCap VRAM% GPU% ^[3m (DID, GUID) (Edge) (Avg) (Mem, Compute, ID) ^[0m ======================================================================================================================== 0 1 0x744c, 48669 35.0°C 18.0W N/A, N/A, 0 0Mhz 96Mhz 0% auto 315.0W 39% 0% 1 2 0x744c, 34500 34.0°C 16.0W N/A, N/A, 0 0Mhz 96Mhz 0% auto 339.0W 39% 0% 2 3 0x744c, 11926 34.0°C 17.0W N/A, N/A, 0 0Mhz 96Mhz 0% auto 339.0W 36% 0% 3 4 0x164e, 27864 48.0°C 32.042W N/A, N/A, 0 None 1800Mhz 0% auto Unsupported 7% 0% ======================================================================================================================== ================================================= End of ROCm SMI Log ================================================== With 1.45>= I got "Error: llama runner process has terminated: signal: segmentation fault (core dumped)" ```
GiteaMirror commented 2026-04-22 08:04:53 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Jul 15, 2024):

Thanks for clarifying. It's either a bug/regression somewhere between llama.cpp tag b3051 and b3171 based on our update in 0.1.45, or could be the result of us bumping to ROCm v6.1.1

I haven't been able to reproduce the failure, although my setup isn't identical. On a dual Radeon RX 6800 setup (2x16G) I'm able to load mixtral:8x7b-instruct-v0.1-q4_K_M

I'm not sure if the defect is GPU model specific, or requires 3 GPUs. One possible workaround to try until we fix this is set HIP_VISIBLE_DEVICES to expose only 2 GPUs.

Also please make sure you're running the latest amdgpu driver.

<!-- gh-comment-id:2229504468 --> @dhiltgen commented on GitHub (Jul 15, 2024): Thanks for clarifying. It's either a bug/regression somewhere between llama.cpp tag b3051 and b3171 based on our update in 0.1.45, or could be the result of us bumping to ROCm v6.1.1 I haven't been able to reproduce the failure, although my setup isn't identical. On a dual Radeon RX 6800 setup (2x16G) I'm able to load `mixtral:8x7b-instruct-v0.1-q4_K_M` I'm not sure if the defect is GPU model specific, or requires 3 GPUs. One possible workaround to try until we fix this is set `HIP_VISIBLE_DEVICES` to expose only 2 GPUs. Also please make sure you're running the latest amdgpu driver.
GiteaMirror commented 2026-04-22 08:04:53 -05:00
Author
Owner
Copy Link

@darwinvelez58 commented on GitHub (Jul 15, 2024):

I don't know if this helps, my packages:

ii  rocm-cmake                                 0.12.0.60101-90~22.04                             amd64        rocm-cmake built using CMake
ii  rocm-core                                  6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-dbgapi                                0.71.0.60101-90~22.04                             amd64        Library to provide AMD GPU debugger API
ii  rocm-debug-agent                           2.0.3.60101-90~22.04                              amd64        Radeon Open Compute Debug Agent (ROCdebug-agent)
ii  rocm-developer-tools                       6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-device-libs                           1.0.0.60101-90~22.04                              amd64        Radeon Open Compute - device libraries
ii  rocm-gdb                                   14.1.60101-90~22.04                               amd64        ROCgdb
ii  rocm-hip-libraries                         6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-hip-runtime                           6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-hip-runtime-dev                       6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-hip-sdk                               6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-language-runtime                      6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-llvm                                  17.0.0.24154.60101-90~22.04                       amd64        ROCm core compiler
ii  rocm-ml-libraries                          6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-ml-sdk                                6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-opencl                                2.0.0.60101-90~22.04                              amd64        clr built using CMake
ii  rocm-opencl-dev                            2.0.0.60101-90~22.04                              amd64        clr built using CMake
ii  rocm-opencl-icd-loader                     1.2.60101-90~22.04                                amd64        OpenCL-ICD-Loader built using CMake
ii  rocm-opencl-runtime                        6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-opencl-sdk                            6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocm-openmp-sdk                            6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) OpenMP Software development Kit.
ii  rocm-smi-lib                               7.0.0.60101-90~22.04                              amd64        AMD System Management libraries
ii  rocm-utils                                 6.1.1.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime software stack
ii  rocminfo                                   1.0.0.60101-90~22.04                              amd64        Radeon Open Compute (ROCm) Runtime rocminfo tool
<!-- gh-comment-id:2229518453 --> @darwinvelez58 commented on GitHub (Jul 15, 2024): I don't know if this helps, my packages: ``` ii rocm-cmake 0.12.0.60101-90~22.04 amd64 rocm-cmake built using CMake ii rocm-core 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-dbgapi 0.71.0.60101-90~22.04 amd64 Library to provide AMD GPU debugger API ii rocm-debug-agent 2.0.3.60101-90~22.04 amd64 Radeon Open Compute Debug Agent (ROCdebug-agent) ii rocm-developer-tools 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-device-libs 1.0.0.60101-90~22.04 amd64 Radeon Open Compute - device libraries ii rocm-gdb 14.1.60101-90~22.04 amd64 ROCgdb ii rocm-hip-libraries 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-hip-runtime 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-hip-runtime-dev 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-hip-sdk 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-language-runtime 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-llvm 17.0.0.24154.60101-90~22.04 amd64 ROCm core compiler ii rocm-ml-libraries 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-ml-sdk 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-opencl 2.0.0.60101-90~22.04 amd64 clr built using CMake ii rocm-opencl-dev 2.0.0.60101-90~22.04 amd64 clr built using CMake ii rocm-opencl-icd-loader 1.2.60101-90~22.04 amd64 OpenCL-ICD-Loader built using CMake ii rocm-opencl-runtime 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-opencl-sdk 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocm-openmp-sdk 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) OpenMP Software development Kit. ii rocm-smi-lib 7.0.0.60101-90~22.04 amd64 AMD System Management libraries ii rocm-utils 6.1.1.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime software stack ii rocminfo 1.0.0.60101-90~22.04 amd64 Radeon Open Compute (ROCm) Runtime rocminfo tool ```
GiteaMirror commented 2026-04-22 08:04:53 -05:00
Author
Owner
Copy Link

@darwinvelez58 commented on GitHub (Jul 17, 2024):

Still same issue on ollama rocm 2.6

[root@83ea2d0ac04b /]# ollama run mixtral:8x7b-instruct-v0.1-q4_K_S
Error: llama runner process has terminated: signal: segmentation fault (core dumped) 
[root@83ea2d0ac04b /]# 



2024/07/17 12:15:49 routes.go:1101: INFO server config env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: OLLAMA_DEBUG:false OLLAMA_FLASH_ATTENTION:false OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_LLM_LIBRARY: OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_VRAM:0 OLLAMA_MODELS:/root/.ollama/models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 OLLAMA_ORIGINS:[http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://*] OLLAMA_RUNNERS_DIR: OLLAMA_SCHED_SPREAD:false OLLAMA_TMPDIR: ROCR_VISIBLE_DEVICES:]"
time=2024-07-17T12:15:49.555Z level=INFO source=images.go:778 msg="total blobs: 23"
time=2024-07-17T12:15:49.555Z level=INFO source=images.go:785 msg="total unused blobs removed: 0"
time=2024-07-17T12:15:49.555Z level=INFO source=routes.go:1148 msg="Listening on [::]:11434 (version 0.2.6)"
time=2024-07-17T12:15:49.556Z level=INFO source=payload.go:30 msg="extracting embedded files" dir=/tmp/ollama2393277633/runners
time=2024-07-17T12:15:51.333Z level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cpu cpu_avx cpu_avx2 cuda_v11 rocm_v60102]"
time=2024-07-17T12:15:51.333Z level=INFO source=gpu.go:205 msg="looking for compatible GPUs"
time=2024-07-17T12:15:51.338Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-07-17T12:15:51.339Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=1 gpu_type=gfx1100
time=2024-07-17T12:15:51.340Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=2 gpu_type=gfx1100
time=2024-07-17T12:15:51.340Z level=INFO source=amd_linux.go:259 msg="unsupported Radeon iGPU detected skipping" id=3 total="512.0 MiB"
time=2024-07-17T12:15:51.340Z level=INFO source=types.go:105 msg="inference compute" id=0 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB"
time=2024-07-17T12:15:51.340Z level=INFO source=types.go:105 msg="inference compute" id=1 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB"
time=2024-07-17T12:15:51.340Z level=INFO source=types.go:105 msg="inference compute" id=2 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB"
[GIN] 2024/07/17 - 12:17:01 | 200 |      39.074µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/07/17 - 12:17:01 | 200 |    2.293986ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2024/07/17 - 12:17:10 | 200 |       15.83µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/07/17 - 12:17:10 | 200 |   15.895513ms |       127.0.0.1 | POST     "/api/show"
time=2024-07-17T12:17:10.801Z level=INFO source=sched.go:717 msg="new model will fit in available VRAM, loading" model=/root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 library=rocm parallel=4 required="33.2 GiB"
time=2024-07-17T12:17:10.802Z level=INFO source=memory.go:309 msg="offload to rocm" layers.requested=-1 layers.model=33 layers.offload=33 layers.split=11,11,11 memory.available="[24.0 GiB 24.0 GiB 24.0 GiB]" memory.required.full="33.2 GiB" memory.required.partial="33.2 GiB" memory.required.kv="1.0 GiB" memory.required.allocations="[11.3 GiB 11.3 GiB 10.6 GiB]" memory.weights.total="25.5 GiB" memory.weights.repeating="25.4 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="1.3 GiB" memory.graph.partial="1.3 GiB"
time=2024-07-17T12:17:10.803Z level=INFO source=server.go:383 msg="starting llama server" cmd="/tmp/ollama2393277633/runners/rocm_v60102/ollama_llama_server --model /root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 --ctx-size 8192 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 4 --tensor-split 11,11,11 --port 37197"
time=2024-07-17T12:17:10.803Z level=INFO source=sched.go:437 msg="loaded runners" count=1
time=2024-07-17T12:17:10.803Z level=INFO source=server.go:571 msg="waiting for llama runner to start responding"
time=2024-07-17T12:17:10.803Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server error"
INFO [main] build info | build=1 commit="a8db2a9" tid="136873938252608" timestamp=1721218630
INFO [main] system info | n_threads=16 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 0 | " tid="136873938252608" timestamp=1721218630 total_threads=32
INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="31" port="37197" tid="136873938252608" timestamp=1721218630
llama_model_loader: loaded meta data with 26 key-value pairs and 995 tensors from /root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = llama
llama_model_loader: - kv   1:                               general.name str              = mistralai
llama_model_loader: - kv   2:                       llama.context_length u32              = 32768
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   4:                          llama.block_count u32              = 32
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   9:                         llama.expert_count u32              = 8
llama_model_loader: - kv  10:                    llama.expert_used_count u32              = 2
llama_model_loader: - kv  11:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  12:                       llama.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  13:                          general.file_type u32              = 14
llama_model_loader: - kv  14:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,32000]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  16:                      tokenizer.ggml.scores arr[f32,32000]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  17:                  tokenizer.ggml.token_type arr[i32,32000]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  18:                      tokenizer.ggml.merges arr[str,58980]   = ["▁ t", "i n", "e r", "▁ a", "h e...
llama_model_loader: - kv  19:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  20:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  21:            tokenizer.ggml.unknown_token_id u32              = 0
llama_model_loader: - kv  22:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  23:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  24:                    tokenizer.chat_template str              = {{ bos_token }}{% for message in mess...
llama_model_loader: - kv  25:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type  f16:   32 tensors
llama_model_loader: - type q8_0:   64 tensors
llama_model_loader: - type q4_K:  833 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: special tokens cache size = 259
llm_load_vocab: token to piece cache size = 0.1637 MB
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32000
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: vocab_only       = 0
llm_load_print_meta: n_ctx_train      = 32768
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_layer          = 32
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_swa            = 0
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 4
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
llm_load_print_meta: f_clamp_kqv      = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: f_logit_scale    = 0.0e+00
llm_load_print_meta: n_ff             = 14336
llm_load_print_meta: n_expert         = 8
llm_load_print_meta: n_expert_used    = 2
llm_load_print_meta: causal attn      = 1
llm_load_print_meta: pooling type     = 0
llm_load_print_meta: rope type        = 0
llm_load_print_meta: rope scaling     = linear
llm_load_print_meta: freq_base_train  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 32768
llm_load_print_meta: rope_finetuned   = unknown
llm_load_print_meta: ssm_d_conv       = 0
llm_load_print_meta: ssm_d_inner      = 0
llm_load_print_meta: ssm_d_state      = 0
llm_load_print_meta: ssm_dt_rank      = 0
llm_load_print_meta: model type       = 8x7B
llm_load_print_meta: model ftype      = Q4_K - Small
llm_load_print_meta: model params     = 46.70 B
llm_load_print_meta: model size       = 24.62 GiB (4.53 BPW) 
llm_load_print_meta: general.name     = mistralai
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 2 '</s>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: LF token         = 13 '<0x0A>'
llm_load_print_meta: max token length = 48
time=2024-07-17T12:17:11.054Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model"
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 3 ROCm devices:
  Device 0: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
  Device 1: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
  Device 2: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
llm_load_tensors: ggml ctx size =    1.53 MiB
llm_load_tensors: offloading 32 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 33/33 layers to GPU
llm_load_tensors:      ROCm0 buffer size =  8608.53 MiB
llm_load_tensors:      ROCm1 buffer size =  8608.53 MiB
llm_load_tensors:      ROCm2 buffer size =  7928.49 MiB
llm_load_tensors:  ROCm_Host buffer size =    70.31 MiB
time=2024-07-17T12:17:31.309Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server not responding"
llama_new_context_with_model: n_ctx      = 8192
llama_new_context_with_model: n_batch    = 512
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: flash_attn = 0
llama_new_context_with_model: freq_base  = 1000000.0
llama_new_context_with_model: freq_scale = 1
time=2024-07-17T12:17:32.140Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model"
llama_kv_cache_init:      ROCm0 KV buffer size =   352.00 MiB
llama_kv_cache_init:      ROCm1 KV buffer size =   352.00 MiB
llama_kv_cache_init:      ROCm2 KV buffer size =   320.00 MiB
llama_new_context_with_model: KV self size  = 1024.00 MiB, K (f16):  512.00 MiB, V (f16):  512.00 MiB
llama_new_context_with_model:  ROCm_Host  output buffer size =     0.55 MiB
llama_new_context_with_model: pipeline parallelism enabled (n_copies=4)
llama_new_context_with_model:      ROCm0 compute buffer size =   640.01 MiB
llama_new_context_with_model:      ROCm1 compute buffer size =   640.01 MiB
llama_new_context_with_model:      ROCm2 compute buffer size =   640.02 MiB
llama_new_context_with_model:  ROCm_Host compute buffer size =    72.02 MiB
llama_new_context_with_model: graph nodes  = 1510
llama_new_context_with_model: graph splits = 4
time=2024-07-17T12:17:34.714Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server error"
time=2024-07-17T12:17:34.965Z level=ERROR source=sched.go:443 msg="error loading llama server" error="llama runner process has terminated: signal: segmentation fault (core dumped) "
[GIN] 2024/07/17 - 12:17:34 | 500 | 24.179477976s |       127.0.0.1 | POST     "/api/chat"
time=2024-07-17T12:17:39.965Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.000331042 model=/root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215
time=2024-07-17T12:17:40.215Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.250300816 model=/root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215
time=2024-07-17T12:17:40.466Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.500866415 model=/root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215
<!-- gh-comment-id:2233187095 --> @darwinvelez58 commented on GitHub (Jul 17, 2024): Still same issue on ollama rocm 2.6 ``` [root@83ea2d0ac04b /]# ollama run mixtral:8x7b-instruct-v0.1-q4_K_S Error: llama runner process has terminated: signal: segmentation fault (core dumped) [root@83ea2d0ac04b /]# 2024/07/17 12:15:49 routes.go:1101: INFO server config env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: OLLAMA_DEBUG:false OLLAMA_FLASH_ATTENTION:false OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_LLM_LIBRARY: OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_VRAM:0 OLLAMA_MODELS:/root/.ollama/models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:0 OLLAMA_ORIGINS:[http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://*] OLLAMA_RUNNERS_DIR: OLLAMA_SCHED_SPREAD:false OLLAMA_TMPDIR: ROCR_VISIBLE_DEVICES:]" time=2024-07-17T12:15:49.555Z level=INFO source=images.go:778 msg="total blobs: 23" time=2024-07-17T12:15:49.555Z level=INFO source=images.go:785 msg="total unused blobs removed: 0" time=2024-07-17T12:15:49.555Z level=INFO source=routes.go:1148 msg="Listening on [::]:11434 (version 0.2.6)" time=2024-07-17T12:15:49.556Z level=INFO source=payload.go:30 msg="extracting embedded files" dir=/tmp/ollama2393277633/runners time=2024-07-17T12:15:51.333Z level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cpu cpu_avx cpu_avx2 cuda_v11 rocm_v60102]" time=2024-07-17T12:15:51.333Z level=INFO source=gpu.go:205 msg="looking for compatible GPUs" time=2024-07-17T12:15:51.338Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100 time=2024-07-17T12:15:51.339Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=1 gpu_type=gfx1100 time=2024-07-17T12:15:51.340Z level=INFO source=amd_linux.go:330 msg="amdgpu is supported" gpu=2 gpu_type=gfx1100 time=2024-07-17T12:15:51.340Z level=INFO source=amd_linux.go:259 msg="unsupported Radeon iGPU detected skipping" id=3 total="512.0 MiB" time=2024-07-17T12:15:51.340Z level=INFO source=types.go:105 msg="inference compute" id=0 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB" time=2024-07-17T12:15:51.340Z level=INFO source=types.go:105 msg="inference compute" id=1 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB" time=2024-07-17T12:15:51.340Z level=INFO source=types.go:105 msg="inference compute" id=2 library=rocm compute=gfx1100 driver=6.7 name=1002:744c total="24.0 GiB" available="24.0 GiB" [GIN] 2024/07/17 - 12:17:01 | 200 | 39.074µs | 127.0.0.1 | HEAD "/" [GIN] 2024/07/17 - 12:17:01 | 200 | 2.293986ms | 127.0.0.1 | GET "/api/tags" [GIN] 2024/07/17 - 12:17:10 | 200 | 15.83µs | 127.0.0.1 | HEAD "/" [GIN] 2024/07/17 - 12:17:10 | 200 | 15.895513ms | 127.0.0.1 | POST "/api/show" time=2024-07-17T12:17:10.801Z level=INFO source=sched.go:717 msg="new model will fit in available VRAM, loading" model=/root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 library=rocm parallel=4 required="33.2 GiB" time=2024-07-17T12:17:10.802Z level=INFO source=memory.go:309 msg="offload to rocm" layers.requested=-1 layers.model=33 layers.offload=33 layers.split=11,11,11 memory.available="[24.0 GiB 24.0 GiB 24.0 GiB]" memory.required.full="33.2 GiB" memory.required.partial="33.2 GiB" memory.required.kv="1.0 GiB" memory.required.allocations="[11.3 GiB 11.3 GiB 10.6 GiB]" memory.weights.total="25.5 GiB" memory.weights.repeating="25.4 GiB" memory.weights.nonrepeating="102.6 MiB" memory.graph.full="1.3 GiB" memory.graph.partial="1.3 GiB" time=2024-07-17T12:17:10.803Z level=INFO source=server.go:383 msg="starting llama server" cmd="/tmp/ollama2393277633/runners/rocm_v60102/ollama_llama_server --model /root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 --ctx-size 8192 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 4 --tensor-split 11,11,11 --port 37197" time=2024-07-17T12:17:10.803Z level=INFO source=sched.go:437 msg="loaded runners" count=1 time=2024-07-17T12:17:10.803Z level=INFO source=server.go:571 msg="waiting for llama runner to start responding" time=2024-07-17T12:17:10.803Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server error" INFO [main] build info | build=1 commit="a8db2a9" tid="136873938252608" timestamp=1721218630 INFO [main] system info | n_threads=16 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 0 | " tid="136873938252608" timestamp=1721218630 total_threads=32 INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="31" port="37197" tid="136873938252608" timestamp=1721218630 llama_model_loader: loaded meta data with 26 key-value pairs and 995 tensors from /root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = llama llama_model_loader: - kv 1: general.name str = mistralai llama_model_loader: - kv 2: llama.context_length u32 = 32768 llama_model_loader: - kv 3: llama.embedding_length u32 = 4096 llama_model_loader: - kv 4: llama.block_count u32 = 32 llama_model_loader: - kv 5: llama.feed_forward_length u32 = 14336 llama_model_loader: - kv 6: llama.rope.dimension_count u32 = 128 llama_model_loader: - kv 7: llama.attention.head_count u32 = 32 llama_model_loader: - kv 8: llama.attention.head_count_kv u32 = 8 llama_model_loader: - kv 9: llama.expert_count u32 = 8 llama_model_loader: - kv 10: llama.expert_used_count u32 = 2 llama_model_loader: - kv 11: llama.attention.layer_norm_rms_epsilon f32 = 0.000010 llama_model_loader: - kv 12: llama.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 13: general.file_type u32 = 14 llama_model_loader: - kv 14: tokenizer.ggml.model str = llama llama_model_loader: - kv 15: tokenizer.ggml.tokens arr[str,32000] = ["<unk>", "<s>", "</s>", "<0x00>", "<... llama_model_loader: - kv 16: tokenizer.ggml.scores arr[f32,32000] = [0.000000, 0.000000, 0.000000, 0.0000... llama_model_loader: - kv 17: tokenizer.ggml.token_type arr[i32,32000] = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ... llama_model_loader: - kv 18: tokenizer.ggml.merges arr[str,58980] = ["▁ t", "i n", "e r", "▁ a", "h e... llama_model_loader: - kv 19: tokenizer.ggml.bos_token_id u32 = 1 llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 2 llama_model_loader: - kv 21: tokenizer.ggml.unknown_token_id u32 = 0 llama_model_loader: - kv 22: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 23: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 24: tokenizer.chat_template str = {{ bos_token }}{% for message in mess... llama_model_loader: - kv 25: general.quantization_version u32 = 2 llama_model_loader: - type f32: 65 tensors llama_model_loader: - type f16: 32 tensors llama_model_loader: - type q8_0: 64 tensors llama_model_loader: - type q4_K: 833 tensors llama_model_loader: - type q6_K: 1 tensors llm_load_vocab: special tokens cache size = 259 llm_load_vocab: token to piece cache size = 0.1637 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = llama llm_load_print_meta: vocab type = SPM llm_load_print_meta: n_vocab = 32000 llm_load_print_meta: n_merges = 0 llm_load_print_meta: vocab_only = 0 llm_load_print_meta: n_ctx_train = 32768 llm_load_print_meta: n_embd = 4096 llm_load_print_meta: n_layer = 32 llm_load_print_meta: n_head = 32 llm_load_print_meta: n_head_kv = 8 llm_load_print_meta: n_rot = 128 llm_load_print_meta: n_swa = 0 llm_load_print_meta: n_embd_head_k = 128 llm_load_print_meta: n_embd_head_v = 128 llm_load_print_meta: n_gqa = 4 llm_load_print_meta: n_embd_k_gqa = 1024 llm_load_print_meta: n_embd_v_gqa = 1024 llm_load_print_meta: f_norm_eps = 0.0e+00 llm_load_print_meta: f_norm_rms_eps = 1.0e-05 llm_load_print_meta: f_clamp_kqv = 0.0e+00 llm_load_print_meta: f_max_alibi_bias = 0.0e+00 llm_load_print_meta: f_logit_scale = 0.0e+00 llm_load_print_meta: n_ff = 14336 llm_load_print_meta: n_expert = 8 llm_load_print_meta: n_expert_used = 2 llm_load_print_meta: causal attn = 1 llm_load_print_meta: pooling type = 0 llm_load_print_meta: rope type = 0 llm_load_print_meta: rope scaling = linear llm_load_print_meta: freq_base_train = 1000000.0 llm_load_print_meta: freq_scale_train = 1 llm_load_print_meta: n_ctx_orig_yarn = 32768 llm_load_print_meta: rope_finetuned = unknown llm_load_print_meta: ssm_d_conv = 0 llm_load_print_meta: ssm_d_inner = 0 llm_load_print_meta: ssm_d_state = 0 llm_load_print_meta: ssm_dt_rank = 0 llm_load_print_meta: model type = 8x7B llm_load_print_meta: model ftype = Q4_K - Small llm_load_print_meta: model params = 46.70 B llm_load_print_meta: model size = 24.62 GiB (4.53 BPW) llm_load_print_meta: general.name = mistralai llm_load_print_meta: BOS token = 1 '<s>' llm_load_print_meta: EOS token = 2 '</s>' llm_load_print_meta: UNK token = 0 '<unk>' llm_load_print_meta: LF token = 13 '<0x0A>' llm_load_print_meta: max token length = 48 time=2024-07-17T12:17:11.054Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model" ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 3 ROCm devices: Device 0: Radeon RX 7900 XTX, compute capability 11.0, VMM: no Device 1: Radeon RX 7900 XTX, compute capability 11.0, VMM: no Device 2: Radeon RX 7900 XTX, compute capability 11.0, VMM: no llm_load_tensors: ggml ctx size = 1.53 MiB llm_load_tensors: offloading 32 repeating layers to GPU llm_load_tensors: offloading non-repeating layers to GPU llm_load_tensors: offloaded 33/33 layers to GPU llm_load_tensors: ROCm0 buffer size = 8608.53 MiB llm_load_tensors: ROCm1 buffer size = 8608.53 MiB llm_load_tensors: ROCm2 buffer size = 7928.49 MiB llm_load_tensors: ROCm_Host buffer size = 70.31 MiB time=2024-07-17T12:17:31.309Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server not responding" llama_new_context_with_model: n_ctx = 8192 llama_new_context_with_model: n_batch = 512 llama_new_context_with_model: n_ubatch = 512 llama_new_context_with_model: flash_attn = 0 llama_new_context_with_model: freq_base = 1000000.0 llama_new_context_with_model: freq_scale = 1 time=2024-07-17T12:17:32.140Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server loading model" llama_kv_cache_init: ROCm0 KV buffer size = 352.00 MiB llama_kv_cache_init: ROCm1 KV buffer size = 352.00 MiB llama_kv_cache_init: ROCm2 KV buffer size = 320.00 MiB llama_new_context_with_model: KV self size = 1024.00 MiB, K (f16): 512.00 MiB, V (f16): 512.00 MiB llama_new_context_with_model: ROCm_Host output buffer size = 0.55 MiB llama_new_context_with_model: pipeline parallelism enabled (n_copies=4) llama_new_context_with_model: ROCm0 compute buffer size = 640.01 MiB llama_new_context_with_model: ROCm1 compute buffer size = 640.01 MiB llama_new_context_with_model: ROCm2 compute buffer size = 640.02 MiB llama_new_context_with_model: ROCm_Host compute buffer size = 72.02 MiB llama_new_context_with_model: graph nodes = 1510 llama_new_context_with_model: graph splits = 4 time=2024-07-17T12:17:34.714Z level=INFO source=server.go:612 msg="waiting for server to become available" status="llm server error" time=2024-07-17T12:17:34.965Z level=ERROR source=sched.go:443 msg="error loading llama server" error="llama runner process has terminated: signal: segmentation fault (core dumped) " [GIN] 2024/07/17 - 12:17:34 | 500 | 24.179477976s | 127.0.0.1 | POST "/api/chat" time=2024-07-17T12:17:39.965Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.000331042 model=/root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 time=2024-07-17T12:17:40.215Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.250300816 model=/root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 time=2024-07-17T12:17:40.466Z level=WARN source=sched.go:634 msg="gpu VRAM usage didn't recover within timeout" seconds=5.500866415 model=/root/.ollama/models/blobs/sha256-728969cf2d06e54ae8e8bec04eccb52c3db919587800c563917e2729b7172215 ```
GiteaMirror commented 2026-04-22 08:04:54 -05:00
Author
Owner
Copy Link

@OliverStutz commented on GitHub (Jul 18, 2024):

@dhiltgen we are happy to provide a resolution bounty if we can load and run the mixtral on the 3x 7900 XTX

<!-- gh-comment-id:2237740745 --> @OliverStutz commented on GitHub (Jul 18, 2024): @dhiltgen we are happy to provide a resolution bounty if we can load and run the mixtral on the 3x 7900 XTX
GiteaMirror commented 2026-04-22 08:04:54 -05:00
Author
Owner
Copy Link

@dhiltgen commented on GitHub (Jul 23, 2024):

Just noticed we appear to have 2 issues tracking this. Lets dup this to the older issue #5629

<!-- gh-comment-id:2246412655 --> @dhiltgen commented on GitHub (Jul 23, 2024): Just noticed we appear to have 2 issues tracking this. Lets dup this to the older issue #5629
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label amd bug gpu
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#29317
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?