github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #6348] Mistral 7B, running on CPU only - can't fix it #50493

New Issue
Closed
opened 2026-04-28 16:06:03 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-28 16:06:03 -05:00
Owner
Copy Link

Originally created by @openSourcerer9000 on GitHub (Aug 13, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/6348

What is the issue?

Running Mistral 7B instruct, simple prompts take tens of minutes. Task manager shows CPU is in heavy use and GPU is doing nothing. I can run it with quantization normally without ollama. How to force ollama to use GPU?

This is my code:

from langchain_ollama import ChatOllama

misty = ChatOllama(
    model = "mistral",
    temperature = 0.2,
    num_predict = 5200,
)

time=2024-08-12T20:24:12.353-06:00 level=INFO source=server.go:392 msg="starting llama server" cmd="C:\\Users\\user\\AppData\\Local\\Programs\\Ollama\\ollama_runners\\cuda_v11.3\\ollama_llama_server.exe --model C:\\Users\\user\\.ollama\\models\\blobs\\sha256-ff82381e2bea77d91c1b824c7afb83f6fb73e9f7de9dda631bcdbca564aa5435 --ctx-size 2048 --batch-size 512 --embedding --log-disable --n-gpu-layers 12 --no-mmap --parallel 1 --port 62988"
time=2024-08-12T20:24:12.620-06:00 level=INFO source=sched.go:445 msg="loaded runners" count=1
time=2024-08-12T20:24:12.620-06:00 level=INFO source=server.go:592 msg="waiting for llama runner to start responding"
time=2024-08-12T20:24:12.621-06:00 level=INFO source=server.go:626 msg="waiting for server to become available" status="llm server error"
INFO [wmain] build info | build=3535 commit="1e6f6554" tid="13432" timestamp=1723515852
INFO [wmain] system info | n_threads=6 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="13432" timestamp=1723515852 total_threads=12
INFO [wmain] HTTP server listening | hostname="127.0.0.1" n_threads_http="11" port="62988" tid="13432" timestamp=1723515852
llama_model_loader: loaded meta data with 25 key-value pairs and 291 tensors from C:\Users\user\.ollama\models\blobs\sha256-ff82381e2bea77d91c1b824c7afb83f6fb73e9f7de9dda631bcdbca564aa5435 (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              = Mistral-7B-Instruct-v0.3
llama_model_loader: - kv   2:                          llama.block_count u32              = 32
llama_model_loader: - kv   3:                       llama.context_length u32              = 32768
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              = 1000000.000000
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              = 32768
llama_model_loader: - kv  12:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv  13:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  14:                         tokenizer.ggml.pre str              = default
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,32768]   = ["<unk>", "<s>", "</s>", "[INST]", "[...
llama_model_loader: - kv  16:                      tokenizer.ggml.scores arr[f32,32768]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  17:                  tokenizer.ggml.token_type arr[i32,32768]   = [2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, ...
llama_model_loader: - kv  18:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  19:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  20:            tokenizer.ggml.unknown_token_id u32              = 0
llama_model_loader: - kv  21:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  22:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  23:                    tokenizer.chat_template str              = {{ bos_token }}{% for message in mess...
llama_model_loader: - kv  24:               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
llm_load_vocab: special tokens cache size = 771
llm_load_vocab: token to piece cache size = 0.1731 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          = 32768
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         = 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  = 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       = 7B
llm_load_print_meta: model ftype      = Q4_0
llm_load_print_meta: model params     = 7.25 B
llm_load_print_meta: model size       = 3.83 GiB (4.54 BPW) 
llm_load_print_meta: general.name     = Mistral-7B-Instruct-v0.3
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         = 781 '<0x0A>'
llm_load_print_meta: max token length = 48
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 2080, compute capability 7.5, VMM: yes
time=2024-08-12T20:24:12.884-06:00 level=INFO source=server.go:626 msg="waiting for server to become available" status="llm server loading model"
llm_load_tensors: ggml ctx size =    0.27 MiB
llm_load_tensors: offloading 12 repeating layers to GPU
llm_load_tensors: offloaded 12/33 layers to GPU
llm_load_tensors:  CUDA_Host buffer size =  2517.64 MiB
llm_load_tensors:      CUDA0 buffer size =  1404.38 MiB
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
llama_kv_cache_init:  CUDA_Host KV buffer size =   160.00 MiB
llama_kv_cache_init:      CUDA0 KV buffer size =    96.00 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:  CUDA_Host  output buffer size =     0.14 MiB
llama_new_context_with_model:      CUDA0 compute buffer size =   185.00 MiB
llama_new_context_with_model:  CUDA_Host compute buffer size =    12.01 MiB
llama_new_context_with_model: graph nodes  = 1030
llama_new_context_with_model: graph splits = 224
INFO [wmain] model loaded | tid="13432" timestamp=1723515859
time=2024-08-12T20:24:19.236-06:00 level=INFO source=server.go:631 msg="llama runner started in 6.62 seconds"
[GIN] 2024/08/12 - 20:24:57 | 200 |   44.8684308s |       127.0.0.1 | POST     "/api/chat"
[GIN] 2024/08/12 - 20:29:27 | 200 |   38.6899925s |       127.0.0.1 | POST     "/api/chat"```

### OS

Windows

### GPU

Nvidia

### CPU

Intel

### Ollama version

_No response_
Originally created by @openSourcerer9000 on GitHub (Aug 13, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/6348 ### What is the issue? Running Mistral 7B instruct, simple prompts take tens of minutes. Task manager shows CPU is in heavy use and GPU is doing nothing. I can run it with quantization normally without ollama. How to force ollama to use GPU? This is my code: ``` from langchain_ollama import ChatOllama misty = ChatOllama( model = "mistral", temperature = 0.2, num_predict = 5200, ) ``` ```time=2024-08-12T20:24:12.346-06:00 level=INFO source=memory.go:309 msg="offload to cuda" layers.requested=-1 layers.model=33 layers.offload=12 layers.split="" memory.available="[2.2 GiB]" memory.required.full="4.8 GiB" memory.required.partial="2.2 GiB" memory.required.kv="256.0 MiB" memory.required.allocations="[2.2 GiB]" memory.weights.total="3.9 GiB" memory.weights.repeating="3.8 GiB" memory.weights.nonrepeating="105.0 MiB" memory.graph.full="164.0 MiB" memory.graph.partial="185.0 MiB" time=2024-08-12T20:24:12.353-06:00 level=INFO source=server.go:392 msg="starting llama server" cmd="C:\\Users\\user\\AppData\\Local\\Programs\\Ollama\\ollama_runners\\cuda_v11.3\\ollama_llama_server.exe --model C:\\Users\\user\\.ollama\\models\\blobs\\sha256-ff82381e2bea77d91c1b824c7afb83f6fb73e9f7de9dda631bcdbca564aa5435 --ctx-size 2048 --batch-size 512 --embedding --log-disable --n-gpu-layers 12 --no-mmap --parallel 1 --port 62988" time=2024-08-12T20:24:12.620-06:00 level=INFO source=sched.go:445 msg="loaded runners" count=1 time=2024-08-12T20:24:12.620-06:00 level=INFO source=server.go:592 msg="waiting for llama runner to start responding" time=2024-08-12T20:24:12.621-06:00 level=INFO source=server.go:626 msg="waiting for server to become available" status="llm server error" INFO [wmain] build info | build=3535 commit="1e6f6554" tid="13432" timestamp=1723515852 INFO [wmain] system info | n_threads=6 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="13432" timestamp=1723515852 total_threads=12 INFO [wmain] HTTP server listening | hostname="127.0.0.1" n_threads_http="11" port="62988" tid="13432" timestamp=1723515852 llama_model_loader: loaded meta data with 25 key-value pairs and 291 tensors from C:\Users\user\.ollama\models\blobs\sha256-ff82381e2bea77d91c1b824c7afb83f6fb73e9f7de9dda631bcdbca564aa5435 (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 = Mistral-7B-Instruct-v0.3 llama_model_loader: - kv 2: llama.block_count u32 = 32 llama_model_loader: - kv 3: llama.context_length u32 = 32768 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 = 1000000.000000 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 = 32768 llama_model_loader: - kv 12: llama.rope.dimension_count u32 = 128 llama_model_loader: - kv 13: tokenizer.ggml.model str = llama llama_model_loader: - kv 14: tokenizer.ggml.pre str = default llama_model_loader: - kv 15: tokenizer.ggml.tokens arr[str,32768] = ["<unk>", "<s>", "</s>", "[INST]", "[... llama_model_loader: - kv 16: tokenizer.ggml.scores arr[f32,32768] = [0.000000, 0.000000, 0.000000, 0.0000... llama_model_loader: - kv 17: tokenizer.ggml.token_type arr[i32,32768] = [2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, ... llama_model_loader: - kv 18: tokenizer.ggml.bos_token_id u32 = 1 llama_model_loader: - kv 19: tokenizer.ggml.eos_token_id u32 = 2 llama_model_loader: - kv 20: tokenizer.ggml.unknown_token_id u32 = 0 llama_model_loader: - kv 21: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 22: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 23: tokenizer.chat_template str = {{ bos_token }}{% for message in mess... llama_model_loader: - kv 24: 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 llm_load_vocab: special tokens cache size = 771 llm_load_vocab: token to piece cache size = 0.1731 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 = 32768 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 = 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 = 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 = 7B llm_load_print_meta: model ftype = Q4_0 llm_load_print_meta: model params = 7.25 B llm_load_print_meta: model size = 3.83 GiB (4.54 BPW) llm_load_print_meta: general.name = Mistral-7B-Instruct-v0.3 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 = 781 '<0x0A>' llm_load_print_meta: max token length = 48 ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 2080, compute capability 7.5, VMM: yes time=2024-08-12T20:24:12.884-06:00 level=INFO source=server.go:626 msg="waiting for server to become available" status="llm server loading model" llm_load_tensors: ggml ctx size = 0.27 MiB llm_load_tensors: offloading 12 repeating layers to GPU llm_load_tensors: offloaded 12/33 layers to GPU llm_load_tensors: CUDA_Host buffer size = 2517.64 MiB llm_load_tensors: CUDA0 buffer size = 1404.38 MiB 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 llama_kv_cache_init: CUDA_Host KV buffer size = 160.00 MiB llama_kv_cache_init: CUDA0 KV buffer size = 96.00 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: CUDA_Host output buffer size = 0.14 MiB llama_new_context_with_model: CUDA0 compute buffer size = 185.00 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 12.01 MiB llama_new_context_with_model: graph nodes = 1030 llama_new_context_with_model: graph splits = 224 INFO [wmain] model loaded | tid="13432" timestamp=1723515859 time=2024-08-12T20:24:19.236-06:00 level=INFO source=server.go:631 msg="llama runner started in 6.62 seconds" [GIN] 2024/08/12 - 20:24:57 | 200 | 44.8684308s | 127.0.0.1 | POST "/api/chat" [GIN] 2024/08/12 - 20:29:27 | 200 | 38.6899925s | 127.0.0.1 | POST "/api/chat"``` ### OS Windows ### GPU Nvidia ### CPU Intel ### Ollama version _No response_
GiteaMirror added the bug label 2026-04-28 16:06:03 -05:00
GiteaMirror commented 2026-04-28 16:06:05 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Aug 13, 2024):

Hi @openSourcerer9000 - it seems only ~1/3 of the model fits on your GPU, meaning some of it will have to run on CPU. In that case, the CPU quickly becomes the bottleneck. May I ask which GPU you have (and how much VRAM?)

<!-- gh-comment-id:2287388582 --> @jmorganca commented on GitHub (Aug 13, 2024): Hi @openSourcerer9000 - it seems only ~1/3 of the model fits on your GPU, meaning some of it will have to run on CPU. In that case, the CPU quickly becomes the bottleneck. May I ask which GPU you have (and how much VRAM?)
GiteaMirror commented 2026-04-28 16:06:06 -05:00
Author
Owner
Copy Link

@openSourcerer9000 commented on GitHub (Aug 15, 2024):

image
It's using heavy quantization out of the box no? With 8GB dedicated VRAM I assume it should be able to handle mistral 7B. Is it possible ollama is loading the model twice?

<!-- gh-comment-id:2291316103 --> @openSourcerer9000 commented on GitHub (Aug 15, 2024): ![image](https://github.com/user-attachments/assets/4b5168f4-a87d-4675-94d1-98b485eb3b2e) It's using heavy quantization out of the box no? With 8GB dedicated VRAM I assume it should be able to handle mistral 7B. Is it possible ollama is loading the model twice?
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 bug
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#50493
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?