Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Ollama crashes after idle and can't process new requests #4492

Open
artem-zinnatullin opened this issue May 17, 2024 · 3 comments
Open

Ollama crashes after idle and can't process new requests #4492

artem-zinnatullin opened this issue May 17, 2024 · 3 comments
Assignees
@dhiltgen
Labels
amd Issues relating to AMD GPUs and ROCm bug Something isn't working

Comments

@artem-zinnatullin
Copy link

What is the issue?

When I keep ollama running idling for some time it then crashes and stops responding to requests:

llm_load_print_meta: EOT token        = 128009 '<|eot_id|>'

rocBLAS error: Could not initialize Tensile host: No devices found
[GIN] 2024/05/17 - 01:05:34 | 500 |  1.472888689s |   192.168.1.112 | POST     "/api/chat"
time=2024-05-17T01:05:34.549-06:00 level=ERROR source=sched.go:339 msg="error loading llama server" error="llama runner process has terminated: signal: aborted (core dumped) error:Could not initialize Tensile host: No devices found"
time=2024-05-17T01:05:34.553-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100

Ollama ollama/ollama:0.1.37-rocm is running in Docker (actually K8S) on Ubuntu Server with AMD 7900XTX GPU

Full log from start to crash: 
2024/05/16 22:16:57 routes.go:1006: INFO server config env="map[OLLAMA_DEBUG:false OLLAMA_LLM_LIBRARY: OLLAMA_MAX_LOADED_MODELS:1 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_VRAM:0 OLLAMA_NOPRUNE:false OLLAMA_NUM_PARALLEL:8 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:*] OLLAMA_RUNNERS_DIR: OLLAMA_TMPDIR:]"
time=2024-05-16T22:16:57.384-06:00 level=INFO source=images.go:704 msg="total blobs: 75"
time=2024-05-16T22:16:57.387-06:00 level=INFO source=images.go:711 msg="total unused blobs removed: 0"
time=2024-05-16T22:16:57.388-06:00 level=INFO source=routes.go:1052 msg="Listening on [::]:11434 (version 0.1.37)"
time=2024-05-16T22:16:57.389-06:00 level=INFO source=payload.go:30 msg="extracting embedded files" dir=/tmp/ollama1104229860/runners
time=2024-05-16T22:16:58.936-06:00 level=INFO source=payload.go:44 msg="Dynamic LLM libraries [cpu cpu_avx cpu_avx2 cuda_v11 rocm_v60002]"
time=2024-05-16T22:16:58.942-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-16T22:16:58.942-06:00 level=INFO source=types.go:71 msg="inference compute" id=0 library=rocm compute=gfx1100 driver=6.3 name=1002:744c total="24.0 GiB" available="24.0 GiB"
[GIN] 2024/05/16 - 22:17:08 | 200 |    2.540133ms |    10.244.0.173 | GET      "/api/tags"
[GIN] 2024/05/16 - 22:17:08 | 200 |    1.311845ms |    10.244.0.173 | GET      "/api/tags"
[GIN] 2024/05/16 - 22:17:09 | 200 |    1.439971ms |    10.244.0.173 | GET      "/api/tags"
[GIN] 2024/05/16 - 22:17:09 | 200 |    1.218168ms |    10.244.0.173 | GET      "/api/tags"
[GIN] 2024/05/16 - 22:17:19 | 200 |      41.469µs |    10.244.0.173 | GET      "/api/version"
time=2024-05-16T22:17:21.319-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-16T22:17:22.017-06:00 level=INFO source=memory.go:127 msg="offload to gpu" layers.real=-1 layers.estimate=33 memory.available="24.0 GiB" memory.required.full="7.7 GiB" memory.required.partial="7.7 GiB" memory.required.kv="2.0 GiB" memory.weights.total="4.2 GiB" memory.weights.repeating="3.8 GiB" memory.weights.nonrepeating="411.0 MiB" memory.graph.full="1.1 GiB" memory.graph.partial="1.1 GiB"
time=2024-05-16T22:17:22.017-06:00 level=INFO source=memory.go:127 msg="offload to gpu" layers.real=-1 layers.estimate=33 memory.available="24.0 GiB" memory.required.full="7.7 GiB" memory.required.partial="7.7 GiB" memory.required.kv="2.0 GiB" memory.weights.total="4.2 GiB" memory.weights.repeating="3.8 GiB" memory.weights.nonrepeating="411.0 MiB" memory.graph.full="1.1 GiB" memory.graph.partial="1.1 GiB"
time=2024-05-16T22:17:22.018-06:00 level=INFO source=server.go:318 msg="starting llama server" cmd="/tmp/ollama1104229860/runners/rocm_v60002/ollama_llama_server --model /root/.ollama/models/blobs/sha256-00e1317cbf74d901080d7100f57580ba8dd8de57203072dc6f668324ba545f29 --ctx-size 16384 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 8 --port 34471"
time=2024-05-16T22:17:22.018-06:00 level=INFO source=sched.go:333 msg="loaded runners" count=1
time=2024-05-16T22:17:22.018-06:00 level=INFO source=server.go:488 msg="waiting for llama runner to start responding"
time=2024-05-16T22:17:22.018-06:00 level=INFO source=server.go:524 msg="waiting for server to become available" status="llm server error"
INFO [main] build info | build=1 commit="952d03d" tid="133496113871936" timestamp=1715919442
INFO [main] system info | n_threads=12 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 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="133496113871936" timestamp=1715919442 total_threads=24
INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="23" port="34471" tid="133496113871936" timestamp=1715919442
llama_model_loader: loaded meta data with 21 key-value pairs and 291 tensors from /root/.ollama/models/blobs/sha256-00e1317cbf74d901080d7100f57580ba8dd8de57203072dc6f668324ba545f29 (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              = Meta-Llama-3-8B-Instruct
llama_model_loader: - kv   2:                          llama.block_count u32              = 32
llama_model_loader: - kv   3:                       llama.context_length u32              = 8192
llama_model_loader: - kv   4:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   7:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   8:                       llama.rope.freq_base f32              = 500000.000000
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                          general.file_type u32              = 2
llama_model_loader: - kv  11:                           llama.vocab_size u32              = 128256
llama_model_loader: - kv  12:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv  13:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  14:                      tokenizer.ggml.tokens arr[str,128256]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  15:                  tokenizer.ggml.token_type arr[i32,128256]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  16:                      tokenizer.ggml.merges arr[str,280147]  = ["Ġ Ġ", "Ġ ĠĠĠ", "ĠĠ ĠĠ", "...
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 128000
llama_model_loader: - kv  18:                tokenizer.ggml.eos_token_id u32              = 128001
llama_model_loader: - kv  19:                    tokenizer.chat_template str              = {% set loop_messages = messages %}{% ...
llama_model_loader: - kv  20:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type q4_0:  225 tensors
llama_model_loader: - type q6_K:    1 tensors
time=2024-05-16T22:17:22.269-06:00 level=INFO source=server.go:524 msg="waiting for server to become available" status="llm server loading model"
llm_load_vocab: missing pre-tokenizer type, using: 'default'
llm_load_vocab:
llm_load_vocab: ************************************
llm_load_vocab: GENERATION QUALITY WILL BE DEGRADED!
llm_load_vocab: CONSIDER REGENERATING THE MODEL
llm_load_vocab: ************************************
llm_load_vocab:
llm_load_vocab: special tokens definition check successful ( 256/128256 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = BPE
llm_load_print_meta: n_vocab          = 128256
llm_load_print_meta: n_merges         = 280147
llm_load_print_meta: n_ctx_train      = 8192
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         = 0
llm_load_print_meta: n_expert_used    = 0
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  = 500000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 8192
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       = 8B
llm_load_print_meta: model ftype      = Q4_0
llm_load_print_meta: model params     = 8.03 B
llm_load_print_meta: model size       = 4.33 GiB (4.64 BPW)
llm_load_print_meta: general.name     = Meta-Llama-3-8B-Instruct
llm_load_print_meta: BOS token        = 128000 '<|begin_of_text|>'
llm_load_print_meta: EOS token        = 128001 '<|end_of_text|>'
llm_load_print_meta: LF token         = 128 'Ä'
llm_load_print_meta: EOT token        = 128009 '<|eot_id|>'
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:   no
ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes
ggml_cuda_init: found 1 ROCm devices:
  Device 0: Radeon RX 7900 XTX, compute capability 11.0, VMM: no
llm_load_tensors: ggml ctx size =    0.30 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 =  4155.99 MiB
llm_load_tensors:        CPU buffer size =   281.81 MiB
......................................................................................
llama_new_context_with_model: n_ctx      = 16384
llama_new_context_with_model: n_batch    = 512
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: freq_base  = 500000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:      ROCm0 KV buffer size =  2048.00 MiB
llama_new_context_with_model: KV self size  = 2048.00 MiB, K (f16): 1024.00 MiB, V (f16): 1024.00 MiB
llama_new_context_with_model:  ROCm_Host  output buffer size =     4.04 MiB
llama_new_context_with_model:      ROCm0 compute buffer size =  1088.00 MiB
llama_new_context_with_model:  ROCm_Host compute buffer size =    40.01 MiB
llama_new_context_with_model: graph nodes  = 1030
llama_new_context_with_model: graph splits = 2
INFO [main] model loaded | tid="133496113871936" timestamp=1715919446
time=2024-05-16T22:17:26.783-06:00 level=INFO source=server.go:529 msg="llama runner started in 4.77 seconds"
[GIN] 2024/05/16 - 22:17:28 | 200 |  7.031066272s |    10.244.0.173 | POST     "/api/chat"
[GIN] 2024/05/16 - 22:17:30 | 200 |  253.996638ms |    10.244.0.173 | POST     "/v1/chat/completions"
[GIN] 2024/05/16 - 22:17:39 | 200 |    1.163486ms |   192.168.1.112 | GET      "/api/tags"
[GIN] 2024/05/16 - 22:19:07 | 200 |   3.71407638s |   192.168.1.112 | POST     "/api/chat"
[GIN] 2024/05/16 - 22:20:57 | 200 |  4.096279991s |   192.168.1.112 | POST     "/api/chat"
time=2024-05-16T22:25:57.760-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-16T22:25:58.014-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:04:51.795-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:04:52.504-06:00 level=INFO source=memory.go:127 msg="offload to gpu" layers.real=-1 layers.estimate=33 memory.available="24.0 GiB" memory.required.full="7.7 GiB" memory.required.partial="7.7 GiB" memory.required.kv="2.0 GiB" memory.weights.total="4.2 GiB" memory.weights.repeating="3.8 GiB" memory.weights.nonrepeating="411.0 MiB" memory.graph.full="1.1 GiB" memory.graph.partial="1.1 GiB"
time=2024-05-17T01:04:52.504-06:00 level=INFO source=memory.go:127 msg="offload to gpu" layers.real=-1 layers.estimate=33 memory.available="24.0 GiB" memory.required.full="7.7 GiB" memory.required.partial="7.7 GiB" memory.required.kv="2.0 GiB" memory.weights.total="4.2 GiB" memory.weights.repeating="3.8 GiB" memory.weights.nonrepeating="411.0 MiB" memory.graph.full="1.1 GiB" memory.graph.partial="1.1 GiB"
time=2024-05-17T01:04:52.504-06:00 level=INFO source=server.go:318 msg="starting llama server" cmd="/tmp/ollama1104229860/runners/rocm_v60002/ollama_llama_server --model /root/.ollama/models/blobs/sha256-00e1317cbf74d901080d7100f57580ba8dd8de57203072dc6f668324ba545f29 --ctx-size 16384 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 8 --port 37839"
time=2024-05-17T01:04:52.505-06:00 level=INFO source=sched.go:333 msg="loaded runners" count=1
time=2024-05-17T01:04:52.505-06:00 level=INFO source=server.go:488 msg="waiting for llama runner to start responding"
time=2024-05-17T01:04:52.505-06:00 level=INFO source=server.go:524 msg="waiting for server to become available" status="llm server error"
INFO [main] build info | build=1 commit="952d03d" tid="127493904571456" timestamp=1715929492
INFO [main] system info | n_threads=12 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 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="127493904571456" timestamp=1715929492 total_threads=24
INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="23" port="37839" tid="127493904571456" timestamp=1715929492
llama_model_loader: loaded meta data with 21 key-value pairs and 291 tensors from /root/.ollama/models/blobs/sha256-00e1317cbf74d901080d7100f57580ba8dd8de57203072dc6f668324ba545f29 (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              = Meta-Llama-3-8B-Instruct
llama_model_loader: - kv   2:                          llama.block_count u32              = 32
llama_model_loader: - kv   3:                       llama.context_length u32              = 8192
llama_model_loader: - kv   4:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   7:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   8:                       llama.rope.freq_base f32              = 500000.000000
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                          general.file_type u32              = 2
llama_model_loader: - kv  11:                           llama.vocab_size u32              = 128256
llama_model_loader: - kv  12:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv  13:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  14:                      tokenizer.ggml.tokens arr[str,128256]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  15:                  tokenizer.ggml.token_type arr[i32,128256]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  16:                      tokenizer.ggml.merges arr[str,280147]  = ["Ġ Ġ", "Ġ ĠĠĠ", "ĠĠ ĠĠ", "...
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 128000
llama_model_loader: - kv  18:                tokenizer.ggml.eos_token_id u32              = 128001
llama_model_loader: - kv  19:                    tokenizer.chat_template str              = {% set loop_messages = messages %}{% ...
llama_model_loader: - kv  20:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type q4_0:  225 tensors
llama_model_loader: - type q6_K:    1 tensors
time=2024-05-17T01:04:52.756-06:00 level=INFO source=server.go:524 msg="waiting for server to become available" status="llm server loading model"
llm_load_vocab: missing pre-tokenizer type, using: 'default'
llm_load_vocab:
llm_load_vocab: ************************************
llm_load_vocab: GENERATION QUALITY WILL BE DEGRADED!
llm_load_vocab: CONSIDER REGENERATING THE MODEL
llm_load_vocab: ************************************
llm_load_vocab:
llm_load_vocab: special tokens definition check successful ( 256/128256 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = BPE
llm_load_print_meta: n_vocab          = 128256
llm_load_print_meta: n_merges         = 280147
llm_load_print_meta: n_ctx_train      = 8192
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         = 0
llm_load_print_meta: n_expert_used    = 0
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  = 500000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 8192
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       = 8B
llm_load_print_meta: model ftype      = Q4_0
llm_load_print_meta: model params     = 8.03 B
llm_load_print_meta: model size       = 4.33 GiB (4.64 BPW)
llm_load_print_meta: general.name     = Meta-Llama-3-8B-Instruct
llm_load_print_meta: BOS token        = 128000 '<|begin_of_text|>'
llm_load_print_meta: EOS token        = 128001 '<|end_of_text|>'
llm_load_print_meta: LF token         = 128 'Ä'
llm_load_print_meta: EOT token        = 128009 '<|eot_id|>'

rocBLAS error: Could not initialize Tensile host: No devices found
time=2024-05-17T01:05:11.060-06:00 level=ERROR source=sched.go:339 msg="error loading llama server" error="llama runner process has terminated: signal: aborted (core dumped) error:Could not initialize Tensile host: No devices found"
[GIN] 2024/05/17 - 01:05:11 | 500 | 19.269265911s |   192.168.1.112 | POST     "/api/chat"
time=2024-05-17T01:05:11.064-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:11.319-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:11.568-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:11.818-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:12.068-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:12.318-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:12.568-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:12.819-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:13.069-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:13.319-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:13.569-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:13.819-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:14.068-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:14.319-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:14.569-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:14.819-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:15.069-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:15.319-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:15.569-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:15.819-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:16.065-06:00 level=WARN source=sched.go:507 msg="gpu VRAM usage didn't recover within timeout" seconds=5.00548985
time=2024-05-17T01:05:16.069-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:16.315-06:00 level=WARN source=sched.go:507 msg="gpu VRAM usage didn't recover within timeout" seconds=5.255538881
time=2024-05-17T01:05:16.319-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:16.565-06:00 level=WARN source=sched.go:507 msg="gpu VRAM usage didn't recover within timeout" seconds=5.505473591
time=2024-05-17T01:05:33.080-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:33.795-06:00 level=INFO source=memory.go:127 msg="offload to gpu" layers.real=-1 layers.estimate=33 memory.available="24.0 GiB" memory.required.full="7.7 GiB" memory.required.partial="7.7 GiB" memory.required.kv="2.0 GiB" memory.weights.total="4.2 GiB" memory.weights.repeating="3.8 GiB" memory.weights.nonrepeating="411.0 MiB" memory.graph.full="1.1 GiB" memory.graph.partial="1.1 GiB"
time=2024-05-17T01:05:33.796-06:00 level=INFO source=memory.go:127 msg="offload to gpu" layers.real=-1 layers.estimate=33 memory.available="24.0 GiB" memory.required.full="7.7 GiB" memory.required.partial="7.7 GiB" memory.required.kv="2.0 GiB" memory.weights.total="4.2 GiB" memory.weights.repeating="3.8 GiB" memory.weights.nonrepeating="411.0 MiB" memory.graph.full="1.1 GiB" memory.graph.partial="1.1 GiB"
time=2024-05-17T01:05:33.796-06:00 level=INFO source=server.go:318 msg="starting llama server" cmd="/tmp/ollama1104229860/runners/rocm_v60002/ollama_llama_server --model /root/.ollama/models/blobs/sha256-00e1317cbf74d901080d7100f57580ba8dd8de57203072dc6f668324ba545f29 --ctx-size 16384 --batch-size 512 --embedding --log-disable --n-gpu-layers 33 --parallel 8 --port 46195"
time=2024-05-17T01:05:33.796-06:00 level=INFO source=sched.go:333 msg="loaded runners" count=1
time=2024-05-17T01:05:33.796-06:00 level=INFO source=server.go:488 msg="waiting for llama runner to start responding"
time=2024-05-17T01:05:33.796-06:00 level=INFO source=server.go:524 msg="waiting for server to become available" status="llm server error"
INFO [main] build info | build=1 commit="952d03d" tid="133936011103296" timestamp=1715929533
INFO [main] system info | n_threads=12 n_threads_batch=-1 system_info="AVX = 1 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 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="133936011103296" timestamp=1715929533 total_threads=24
INFO [main] HTTP server listening | hostname="127.0.0.1" n_threads_http="23" port="46195" tid="133936011103296" timestamp=1715929533
llama_model_loader: loaded meta data with 21 key-value pairs and 291 tensors from /root/.ollama/models/blobs/sha256-00e1317cbf74d901080d7100f57580ba8dd8de57203072dc6f668324ba545f29 (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              = Meta-Llama-3-8B-Instruct
llama_model_loader: - kv   2:                          llama.block_count u32              = 32
llama_model_loader: - kv   3:                       llama.context_length u32              = 8192
llama_model_loader: - kv   4:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   7:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   8:                       llama.rope.freq_base f32              = 500000.000000
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                          general.file_type u32              = 2
llama_model_loader: - kv  11:                           llama.vocab_size u32              = 128256
llama_model_loader: - kv  12:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv  13:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  14:                      tokenizer.ggml.tokens arr[str,128256]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  15:                  tokenizer.ggml.token_type arr[i32,128256]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  16:                      tokenizer.ggml.merges arr[str,280147]  = ["Ġ Ġ", "Ġ ĠĠĠ", "ĠĠ ĠĠ", "...
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 128000
llama_model_loader: - kv  18:                tokenizer.ggml.eos_token_id u32              = 128001
llama_model_loader: - kv  19:                    tokenizer.chat_template str              = {% set loop_messages = messages %}{% ...
llama_model_loader: - kv  20:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type q4_0:  225 tensors
llama_model_loader: - type q6_K:    1 tensors
time=2024-05-17T01:05:34.048-06:00 level=INFO source=server.go:524 msg="waiting for server to become available" status="llm server loading model"
llm_load_vocab: missing pre-tokenizer type, using: 'default'
llm_load_vocab:
llm_load_vocab: ************************************
llm_load_vocab: GENERATION QUALITY WILL BE DEGRADED!
llm_load_vocab: CONSIDER REGENERATING THE MODEL
llm_load_vocab: ************************************
llm_load_vocab:
llm_load_vocab: special tokens definition check successful ( 256/128256 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = BPE
llm_load_print_meta: n_vocab          = 128256
llm_load_print_meta: n_merges         = 280147
llm_load_print_meta: n_ctx_train      = 8192
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         = 0
llm_load_print_meta: n_expert_used    = 0
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  = 500000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 8192
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       = 8B
llm_load_print_meta: model ftype      = Q4_0
llm_load_print_meta: model params     = 8.03 B
llm_load_print_meta: model size       = 4.33 GiB (4.64 BPW)
llm_load_print_meta: general.name     = Meta-Llama-3-8B-Instruct
llm_load_print_meta: BOS token        = 128000 '<|begin_of_text|>'
llm_load_print_meta: EOS token        = 128001 '<|end_of_text|>'
llm_load_print_meta: LF token         = 128 'Ä'
llm_load_print_meta: EOT token        = 128009 '<|eot_id|>'

rocBLAS error: Could not initialize Tensile host: No devices found
[GIN] 2024/05/17 - 01:05:34 | 500 |  1.472888689s |   192.168.1.112 | POST     "/api/chat"
time=2024-05-17T01:05:34.549-06:00 level=ERROR source=sched.go:339 msg="error loading llama server" error="llama runner process has terminated: signal: aborted (core dumped) error:Could not initialize Tensile host: No devices found"
time=2024-05-17T01:05:34.553-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:34.807-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:35.057-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:35.307-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:35.557-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:35.807-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:36.057-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:36.307-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:36.557-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:36.807-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:37.057-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:37.307-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:37.557-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:37.807-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:38.057-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:38.308-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:38.557-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:38.807-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:39.057-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:39.307-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:39.553-06:00 level=WARN source=sched.go:507 msg="gpu VRAM usage didn't recover within timeout" seconds=5.003790879
time=2024-05-17T01:05:39.557-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:39.803-06:00 level=WARN source=sched.go:507 msg="gpu VRAM usage didn't recover within timeout" seconds=5.253690762
time=2024-05-17T01:05:39.807-06:00 level=INFO source=amd_linux.go:301 msg="amdgpu is supported" gpu=0 gpu_type=gfx1100
time=2024-05-17T01:05:40.053-06:00 level=WARN source=sched.go:507 msg="gpu VRAM usage didn't recover within timeout" seconds=5.50376608

OS

Linux, Docker

GPU

AMD

CPU

AMD

Ollama version

ollama/ollama:0.1.37-rocm
@artem-zinnatullin artem-zinnatullin added the bug Something isn't working label May 17, 2024
@artem-zinnatullin
Copy link
Author

Another user @cyai reported what seems like an exact same error few days ago #4358 (comment)

@artem-zinnatullin
Copy link
Author

Actually, not only restarting ollama didn't help, amdgpu_top failed to find the device! Full computer reboot helped. Crazy.

I do have GPU_MAX_HW_QUEUES=1 set for ollama because otherwise my AMD 7900XTX GPU stays on ~100W IDLE consumption instead of normal 29W, not sure if this is related tho.

@dhiltgen dhiltgen added the amd Issues relating to AMD GPUs and ROCm label May 21, 2024
@dhiltgen dhiltgen self-assigned this May 21, 2024
@dhiltgen
Copy link
Collaborator

@artem-zinnatullin it seems plausible it may be related to that workaround and issue ROCm/ROCm#2625

If you don't use the workaround and let it run at full power, does it still hang?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@dhiltgen dhiltgen

Labels
amd Issues relating to AMD GPUs and ROCm bug Something isn't working
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@artem-zinnatullin @dhiltgen