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 #8884] 0.5.8-rc7: ROCm Not Loading on to GPU #5759

New Issue
Closed
opened 2026-04-12 17:04:48 -05:00 by GiteaMirror · 9 comments
GiteaMirror commented 2026-04-12 17:04:48 -05:00
Owner
Copy Link

Originally created by @ProjectMoon on GitHub (Feb 6, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/8884

Originally assigned to: @mxyng on GitHub.

What is the issue?

I tried out 0.5.8-rc7, and ollama reports that the model will be offloaded to the GPU, but it appears to only run on CPU.

I found https://github.com/ollama/ollama/issues/8828, but this seems to be happening even with libraries in the right place.

I tried two scenarios:

  1. Download ollama-linux and ollama-linux-rocm .tar.gz files, and untar to /opt/ollama. Run with the dist/linux/amd64 directory structure.
  2. Copy the libraries into the old directory structure: /opt/ollama/bin/ollama and /opt/ollama/lib/ with all new separate lib directories merged together.

In either case, ollama reports offloading to ROCm, but runs on CPU.

Relevant log output

Expand for logs

Here are the logs:

2025/02/06 12:30:37 routes.go:1186: INFO server config env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_DEBUG:true OLLAMA_FLASH_ATTENTION:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE:q8_0 OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/ollama OLLAMA_MULTIUSER_CACHE:false 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://* vscode-webview://*] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]"
time=2025-02-06T12:30:37.908+01:00 level=INFO source=images.go:432 msg="total blobs: 118"
time=2025-02-06T12:30:37.909+01:00 level=INFO source=images.go:439 msg="total unused blobs removed: 0"
time=2025-02-06T12:30:37.910+01:00 level=INFO source=routes.go:1237 msg="Listening on [::]:11434 (version 0.5.8-rc7)"
time=2025-02-06T12:30:37.911+01:00 level=DEBUG source=sched.go:105 msg="starting llm scheduler"
time=2025-02-06T12:30:37.911+01:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs"
time=2025-02-06T12:30:37.912+01:00 level=DEBUG source=gpu.go:98 msg="searching for GPU discovery libraries for NVIDIA"
time=2025-02-06T12:30:37.912+01:00 level=DEBUG source=gpu.go:501 msg="Searching for GPU library" name=libcuda.so*
time=2025-02-06T12:30:37.912+01:00 level=DEBUG source=gpu.go:525 msg="gpu library search" globs="[/opt/ollama/bin/libcuda.so* /libcuda.so* /usr/local/cuda*/targets/*/lib/libcuda.so* /usr/lib/*-linux-gnu/nvidia/current/libcuda.so* /usr/lib/*-linux-gnu/libcuda.so* /usr/lib/wsl/lib/libcuda.so* /usr/lib/wsl/drivers/*/libcuda.so* /opt/cuda/lib*/libcuda.so* /usr/local/cuda/lib*/libcuda.so* /usr/lib*/libcuda.so* /usr/local/lib*/libcuda.so*]"
time=2025-02-06T12:30:37.924+01:00 level=DEBUG source=gpu.go:558 msg="discovered GPU libraries" paths="[/usr/lib/libcuda.so.565.77 /usr/lib64/libcuda.so.565.77]"
initializing /usr/lib/libcuda.so.565.77
library /usr/lib/libcuda.so.565.77 load err: /usr/lib/libcuda.so.565.77: wrong ELF class: ELFCLASS32
time=2025-02-06T12:30:37.924+01:00 level=DEBUG source=gpu.go:609 msg="skipping 32bit library" library=/usr/lib/libcuda.so.565.77
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:37.931+01:00 level=DEBUG source=gpu.go:125 msg="detected GPUs" count=1 library=/usr/lib64/libcuda.so.565.77
[GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7] CUDA totalMem 4030 mb
[GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7] CUDA freeMem 3963 mb
[GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7] Compute Capability 5.2
time=2025-02-06T12:30:38.019+01:00 level=WARN source=amd_linux.go:61 msg="ollama recommends running the https://www.amd.com/en/support/linux-drivers" error="amdgpu version file missing: /sys/module/amdgpu/version stat /sys/module/amdgpu/version: no such file or directory"
time=2025-02-06T12:30:38.019+01:00 level=DEBUG source=amd_linux.go:101 msg="evaluating amdgpu node /sys/class/kfd/kfd/topology/nodes/0/properties"
time=2025-02-06T12:30:38.019+01:00 level=DEBUG source=amd_linux.go:121 msg="detected CPU /sys/class/kfd/kfd/topology/nodes/0/properties"
time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:101 msg="evaluating amdgpu node /sys/class/kfd/kfd/topology/nodes/1/properties"
time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:206 msg="mapping amdgpu to drm sysfs nodes" amdgpu=/sys/class/kfd/kfd/topology/nodes/1/properties vendor=4098 device=29631 unique_id=10870137312548343375
time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:240 msg=matched amdgpu=/sys/class/kfd/kfd/topology/nodes/1/properties drm=/sys/class/drm/card0/device
time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:318 msg="amdgpu memory" gpu=0 total="16.0 GiB"
time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:319 msg="amdgpu memory" gpu=0 available="15.4 GiB"
time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_common.go:16 msg="evaluating potential rocm lib dir /opt/ollama/bin/rocm"
time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_common.go:16 msg="evaluating potential rocm lib dir /opt/rocm/lib"
time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_common.go:16 msg="evaluating potential rocm lib dir /usr/lib64"
time=2025-02-06T12:30:38.025+01:00 level=DEBUG source=amd_linux.go:371 msg="rocm supported GPUs" types="[gfx1030 gfx1100 gfx906 gfx908 gfx90a gfx942]"
time=2025-02-06T12:30:38.025+01:00 level=INFO source=amd_linux.go:386 msg="amdgpu is supported" gpu=GPU-96da7c4b1629ce4f gpu_type=gfx1030
releasing cuda driver library
time=2025-02-06T12:30:38.025+01:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 library=cuda variant=v11 compute=5.2 driver=12.7 name="NVIDIA GeForce GTX 970" total="3.9 GiB" available="3.9 GiB"
time=2025-02-06T12:30:38.025+01:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-96da7c4b1629ce4f library=rocm variant="" compute=gfx1030 driver=0.0 name=1002:73bf total="16.0 GiB" available="15.4 GiB"
[GIN] 2025/02/06 - 12:30:44 | 200 |      41.835µs |       127.0.0.1 | HEAD     "/"
[GIN] 2025/02/06 - 12:30:44 | 200 |   21.973698ms |       127.0.0.1 | POST     "/api/show"
time=2025-02-06T12:30:44.263+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.2 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.344+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.344+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.344+01:00 level=DEBUG source=sched.go:181 msg="updating default concurrency" OLLAMA_MAX_LOADED_MODELS=0x55e3b630e9c0 gpu_count=2
time=2025-02-06T12:30:44.395+01:00 level=DEBUG source=sched.go:224 msg="loading first model" model=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862
time=2025-02-06T12:30:44.395+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=cuda gpu_count=1 available="[3.9 GiB]"
time=2025-02-06T12:30:44.396+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=memory.go:186 msg="gpu has too little memory to allocate any layers" id=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 library=cuda variant=v11 compute=5.2 driver=12.7 name="NVIDIA GeForce GTX 970" total="3.9 GiB" available="3.9 GiB" minimum_memory=479199232 layer_size="307.4 MiB" gpu_zer_overhead="0 B" partial_offload="5.9 GiB" full_offload="4.7 GiB"
time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=memory.go:330 msg="insufficient VRAM to load any model layers"
time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=cuda gpu_count=1 available="[3.9 GiB]"
time=2025-02-06T12:30:44.471+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.0 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.543+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.543+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.544+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=cuda gpu_count=1 available="[3.9 GiB]"
time=2025-02-06T12:30:44.544+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.0 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.611+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.611+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.612+01:00 level=DEBUG source=memory.go:186 msg="gpu has too little memory to allocate any layers" id=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 library=cuda variant=v11 compute=5.2 driver=12.7 name="NVIDIA GeForce GTX 970" total="3.9 GiB" available="3.9 GiB" minimum_memory=479199232 layer_size="307.4 MiB" gpu_zer_overhead="0 B" partial_offload="5.9 GiB" full_offload="4.7 GiB"
time=2025-02-06T12:30:44.612+01:00 level=DEBUG source=memory.go:330 msg="insufficient VRAM to load any model layers"
time=2025-02-06T12:30:44.612+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=cuda gpu_count=1 available="[3.9 GiB]"
time=2025-02-06T12:30:44.612+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.681+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.681+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.681+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=rocm gpu_count=1 available="[15.4 GiB]"
time=2025-02-06T12:30:44.682+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.750+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.750+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.751+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=rocm gpu_count=1 available="[15.4 GiB]"
time=2025-02-06T12:30:44.751+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.820+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.820+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.820+01:00 level=INFO source=sched.go:714 msg="new model will fit in available VRAM in single GPU, loading" model=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 gpu=GPU-96da7c4b1629ce4f parallel=1 available=16494198784 required="12.5 GiB"
time=2025-02-06T12:30:44.820+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.888+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.888+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.888+01:00 level=INFO source=server.go:100 msg="system memory" total="62.7 GiB" free="36.1 GiB" free_swap="1.0 GiB"
time=2025-02-06T12:30:44.888+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=rocm gpu_count=1 available="[15.4 GiB]"
time=2025-02-06T12:30:44.889+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB"
initializing /usr/lib64/libcuda.so.565.77
dlsym: cuInit - 0x7f758d2d4cc0
dlsym: cuDriverGetVersion - 0x7f758d2d4ce0
dlsym: cuDeviceGetCount - 0x7f758d2d4d20
dlsym: cuDeviceGet - 0x7f758d2d4d00
dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00
dlsym: cuDeviceGetUuid - 0x7f758d2d4d60
dlsym: cuDeviceGetName - 0x7f758d2d4d40
dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0
dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760
dlsym: cuCtxDestroy - 0x7f758d3213a0
calling cuInit
calling cuDriverGetVersion
raw version 0x2f26
CUDA driver version: 12.7
calling cuDeviceGetCount
device count 1
time=2025-02-06T12:30:44.959+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB"
time=2025-02-06T12:30:44.959+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB"
releasing cuda driver library
time=2025-02-06T12:30:44.959+01:00 level=INFO source=memory.go:356 msg="offload to rocm" layers.requested=-1 layers.model=49 layers.offload=49 layers.split="" memory.available="[15.4 GiB]" memory.gpu_overhead="0 B" memory.required.full="12.5 GiB" memory.required.partial="12.5 GiB" memory.required.kv="1.4 GiB" memory.required.allocations="[12.5 GiB]" memory.weights.total="10.1 GiB" memory.weights.repeating="9.5 GiB" memory.weights.nonrepeating="609.1 MiB" memory.graph.full="1.2 GiB" memory.graph.partial="1.5 GiB"
time=2025-02-06T12:30:44.959+01:00 level=INFO source=server.go:185 msg="enabling flash attention"
time=2025-02-06T12:30:44.959+01:00 level=DEBUG source=server.go:262 msg="compatible gpu libraries" compatible=[]
time=2025-02-06T12:30:44.960+01:00 level=DEBUG source=server.go:310 msg="adding gpu dependency paths" paths=[/usr/lib64]
time=2025-02-06T12:30:44.960+01:00 level=INFO source=server.go:381 msg="starting llama server" cmd="/opt/ollama/bin/ollama runner --model /ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 --ctx-size 15000 --batch-size 512 --n-gpu-layers 49 --verbose --threads 6 --flash-attn --kv-cache-type q8_0 --parallel 1 --port 46243"
time=2025-02-06T12:30:44.960+01:00 level=DEBUG source=server.go:399 msg=subprocess environment="[PATH=/opt/ollama/bin:/bin:/sbin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin:/opt/bin:/usr/lib/llvm/19/bin:/usr/lib/llvm/18/bin:/etc/eselect/wine/bin:/opt/cuda/bin LD_LIBRARY_PATH=/usr/lib64:/opt/ollama/bin ROCR_VISIBLE_DEVICES=GPU-96da7c4b1629ce4f]"
time=2025-02-06T12:30:44.960+01:00 level=INFO source=sched.go:449 msg="loaded runners" count=1
time=2025-02-06T12:30:44.961+01:00 level=INFO source=server.go:558 msg="waiting for llama runner to start responding"
time=2025-02-06T12:30:44.961+01:00 level=INFO source=server.go:592 msg="waiting for server to become available" status="llm server error"
time=2025-02-06T12:30:44.982+01:00 level=INFO source=runner.go:936 msg="starting go runner"
time=2025-02-06T12:30:44.982+01:00 level=INFO source=runner.go:937 msg=system info="CPU : LLAMAFILE = 1 | CPU : LLAMAFILE = 1 | cgo(gcc)" threads=6
time=2025-02-06T12:30:44.982+01:00 level=DEBUG source=ggml.go:84 msg="ggml backend load all from path" path=/usr/lib64
time=2025-02-06T12:30:44.982+01:00 level=INFO source=runner.go:995 msg="Server listening on 127.0.0.1:46243"
time=2025-02-06T12:30:45.119+01:00 level=DEBUG source=ggml.go:84 msg="ggml backend load all from path" path=/opt/ollama/bin
llama_model_loader: loaded meta data with 43 key-value pairs and 579 tensors from /ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 (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              = qwen2
llama_model_loader: - kv   1:                               general.type str              = model
llama_model_loader: - kv   2:                               general.name str              = Qwen2.5 14B
llama_model_loader: - kv   3:                       general.organization str              = Qwen
llama_model_loader: - kv   4:                           general.basename str              = Qwen2.5
llama_model_loader: - kv   5:                         general.size_label str              = 14B
llama_model_loader: - kv   6:                   general.base_model.count u32              = 3
llama_model_loader: - kv   7:                  general.base_model.0.name str              = Qwamma 14b Merge v1
llama_model_loader: - kv   8:               general.base_model.0.version str              = v1
llama_model_loader: - kv   9:          general.base_model.0.organization str              = Chargoddard
llama_model_loader: - kv  10:              general.base_model.0.repo_url str              = https://huggingface.co/chargoddard/qw...
llama_model_loader: - kv  11:                  general.base_model.1.name str              = Qwen2.5 14B Instruct_arcee Qwen2 14B ...
llama_model_loader: - kv  12:               general.base_model.1.version str              = v0.2
llama_model_loader: - kv  13:          general.base_model.1.organization str              = Arcee Train
llama_model_loader: - kv  14:              general.base_model.1.repo_url str              = https://huggingface.co/arcee-train/Qw...
llama_model_loader: - kv  15:                  general.base_model.2.name str              = Qwen2.5 14B
llama_model_loader: - kv  16:          general.base_model.2.organization str              = Qwen
llama_model_loader: - kv  17:              general.base_model.2.repo_url str              = https://huggingface.co/Qwen/Qwen2.5-14B
llama_model_loader: - kv  18:                               general.tags arr[str,2]       = ["mergekit", "merge"]
llama_model_loader: - kv  19:                          qwen2.block_count u32              = 48
llama_model_loader: - kv  20:                       qwen2.context_length u32              = 131072
llama_model_loader: - kv  21:                     qwen2.embedding_length u32              = 5120
llama_model_loader: - kv  22:                  qwen2.feed_forward_length u32              = 13824
llama_model_loader: - kv  23:                 qwen2.attention.head_count u32              = 40
llama_model_loader: - kv  24:              qwen2.attention.head_count_kv u32              = 8
llama_model_loader: - kv  25:                       qwen2.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  26:     qwen2.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  27:                          general.file_type u32              = 17
llama_model_loader: - kv  28:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  29:                         tokenizer.ggml.pre str              = qwen2
time=2025-02-06T12:30:45.213+01:00 level=INFO source=server.go:592 msg="waiting for server to become available" status="llm server loading model"
llama_model_loader: - kv  30:                      tokenizer.ggml.tokens arr[str,152064]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  31:                  tokenizer.ggml.token_type arr[i32,152064]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  32:                      tokenizer.ggml.merges arr[str,151387]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv  33:                tokenizer.ggml.eos_token_id u32              = 151645
llama_model_loader: - kv  34:            tokenizer.ggml.padding_token_id u32              = 151643
llama_model_loader: - kv  35:                tokenizer.ggml.bos_token_id u32              = 151643
llama_model_loader: - kv  36:               tokenizer.ggml.add_bos_token bool             = false
llama_model_loader: - kv  37:                    tokenizer.chat_template str              = {%- if tools %}\n    {{- '<|im_start|>...
llama_model_loader: - kv  38:               general.quantization_version u32              = 2
llama_model_loader: - kv  39:                      quantize.imatrix.file str              = /models_out/SuperNova-14B-GGUF/SuperN...
llama_model_loader: - kv  40:                   quantize.imatrix.dataset str              = /training_dir/calibration_datav3.txt
llama_model_loader: - kv  41:             quantize.imatrix.entries_count i32              = 336
llama_model_loader: - kv  42:              quantize.imatrix.chunks_count i32              = 128
llama_model_loader: - type  f32:  241 tensors
llama_model_loader: - type q5_K:  289 tensors
llama_model_loader: - type q6_K:   49 tensors
llm_load_vocab: control token: 151660 '<|fim_middle|>' is not marked as EOG
llm_load_vocab: control token: 151659 '<|fim_prefix|>' is not marked as EOG
llm_load_vocab: control token: 151653 '<|vision_end|>' is not marked as EOG
llm_load_vocab: control token: 151648 '<|box_start|>' is not marked as EOG
llm_load_vocab: control token: 151646 '<|object_ref_start|>' is not marked as EOG
llm_load_vocab: control token: 151649 '<|box_end|>' is not marked as EOG
llm_load_vocab: control token: 151655 '<|image_pad|>' is not marked as EOG
llm_load_vocab: control token: 151651 '<|quad_end|>' is not marked as EOG
llm_load_vocab: control token: 151647 '<|object_ref_end|>' is not marked as EOG
llm_load_vocab: control token: 151652 '<|vision_start|>' is not marked as EOG
llm_load_vocab: control token: 151654 '<|vision_pad|>' is not marked as EOG
llm_load_vocab: control token: 151656 '<|video_pad|>' is not marked as EOG
llm_load_vocab: control token: 151644 '<|im_start|>' is not marked as EOG
llm_load_vocab: control token: 151661 '<|fim_suffix|>' is not marked as EOG
llm_load_vocab: control token: 151650 '<|quad_start|>' is not marked as EOG
llm_load_vocab: special tokens cache size = 22
llm_load_vocab: token to piece cache size = 0.9310 MB
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = qwen2
llm_load_print_meta: vocab type       = BPE
llm_load_print_meta: n_vocab          = 152064
llm_load_print_meta: n_merges         = 151387
llm_load_print_meta: vocab_only       = 0
llm_load_print_meta: n_ctx_train      = 131072
llm_load_print_meta: n_embd           = 5120
llm_load_print_meta: n_layer          = 48
llm_load_print_meta: n_head           = 40
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            = 5
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             = 13824
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        = 2
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  = 131072
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: ssm_dt_b_c_rms   = 0
llm_load_print_meta: model type       = 14B
llm_load_print_meta: model ftype      = Q5_K - Medium
llm_load_print_meta: model params     = 14.77 B
llm_load_print_meta: model size       = 9.78 GiB (5.69 BPW) 
llm_load_print_meta: general.name     = Qwen2.5 14B
llm_load_print_meta: BOS token        = 151643 '<|endoftext|>'
llm_load_print_meta: EOS token        = 151645 '<|im_end|>'
llm_load_print_meta: EOT token        = 151645 '<|im_end|>'
llm_load_print_meta: PAD token        = 151643 '<|endoftext|>'
llm_load_print_meta: LF token         = 148848 'ÄĬ'
llm_load_print_meta: FIM PRE token    = 151659 '<|fim_prefix|>'
llm_load_print_meta: FIM SUF token    = 151661 '<|fim_suffix|>'
llm_load_print_meta: FIM MID token    = 151660 '<|fim_middle|>'
llm_load_print_meta: FIM PAD token    = 151662 '<|fim_pad|>'
llm_load_print_meta: FIM REP token    = 151663 '<|repo_name|>'
llm_load_print_meta: FIM SEP token    = 151664 '<|file_sep|>'
llm_load_print_meta: EOG token        = 151643 '<|endoftext|>'
llm_load_print_meta: EOG token        = 151645 '<|im_end|>'
llm_load_print_meta: EOG token        = 151662 '<|fim_pad|>'
llm_load_print_meta: EOG token        = 151663 '<|repo_name|>'
llm_load_print_meta: EOG token        = 151664 '<|file_sep|>'
llm_load_print_meta: max token length = 256
llm_load_tensors:   CPU_Mapped model buffer size = 10016.35 MiB
llama_new_context_with_model: n_seq_max     = 1
llama_new_context_with_model: n_ctx         = 15104
llama_new_context_with_model: n_ctx_per_seq = 15104
llama_new_context_with_model: n_batch       = 512
llama_new_context_with_model: n_ubatch      = 512
llama_new_context_with_model: flash_attn    = 1
llama_new_context_with_model: freq_base     = 1000000.0
llama_new_context_with_model: freq_scale    = 1
llama_new_context_with_model: n_ctx_per_seq (15104) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
llama_kv_cache_init: kv_size = 15104, offload = 1, type_k = 'q8_0', type_v = 'q8_0', n_layer = 48, can_shift = 1
llama_kv_cache_init: layer 0: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 1: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 2: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 3: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 4: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 5: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 6: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 7: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 8: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 9: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 10: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 11: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 12: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 13: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 14: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 15: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 16: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 17: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 18: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 19: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 20: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 21: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 22: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 23: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 24: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 25: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 26: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 27: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 28: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 29: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 30: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 31: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 32: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 33: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 34: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 35: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 36: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 37: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 38: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 39: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 40: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 41: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 42: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 43: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 44: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 45: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 46: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
llama_kv_cache_init: layer 47: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024
time=2025-02-06T12:30:46.971+01:00 level=DEBUG source=server.go:603 msg="model load progress 1.00"
time=2025-02-06T12:30:47.221+01:00 level=DEBUG source=server.go:606 msg="model load completed, waiting for server to become available" status="llm server loading model"
llama_kv_cache_init:        CPU KV buffer size =  1504.50 MiB
llama_new_context_with_model: KV self size  = 1504.50 MiB, K (q8_0):  752.25 MiB, V (q8_0):  752.25 MiB
llama_new_context_with_model:        CPU  output buffer size =     0.60 MiB
llama_new_context_with_model:        CPU compute buffer size =   317.00 MiB
llama_new_context_with_model: graph nodes  = 1495
llama_new_context_with_model: graph splits = 1
time=2025-02-06T12:30:47.974+01:00 level=INFO source=server.go:597 msg="llama runner started in 3.01 seconds"
time=2025-02-06T12:30:47.974+01:00 level=DEBUG source=sched.go:462 msg="finished setting up runner" model=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862
[GIN] 2025/02/06 - 12:30:47 | 200 |  3.733069514s |       127.0.0.1 | POST     "/api/generate"
time=2025-02-06T12:30:47.974+01:00 level=DEBUG source=sched.go:466 msg="context for request finished"
time=2025-02-06T12:30:47.975+01:00 level=DEBUG source=sched.go:339 msg="runner with non-zero duration has gone idle, adding timer" modelPath=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 duration=5m0s
time=2025-02-06T12:30:47.975+01:00 level=DEBUG source=sched.go:357 msg="after processing request finished event" modelPath=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 refCount=0

OS

Linux

GPU

Nvidia, AMD

CPU

AMD

Ollama version

0.5.8-rc7

Originally created by @ProjectMoon on GitHub (Feb 6, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/8884 Originally assigned to: @mxyng on GitHub. ### What is the issue? I tried out 0.5.8-rc7, and ollama reports that the model will be offloaded to the GPU, but it appears to only run on CPU. I found https://github.com/ollama/ollama/issues/8828, but this seems to be happening even with libraries in the right place. I tried two scenarios: 1. Download ollama-linux and ollama-linux-rocm .tar.gz files, and untar to /opt/ollama. Run with the `dist/linux/amd64` directory structure. 2. Copy the libraries into the old directory structure: `/opt/ollama/bin/ollama` and `/opt/ollama/lib/` with all new separate lib directories merged together. In either case, ollama reports offloading to ROCm, but runs on CPU. ### Relevant log output <details> <summary>Expand for logs</summary> Here are the logs: ```shell 2025/02/06 12:30:37 routes.go:1186: INFO server config env="map[CUDA_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL: HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: NO_PROXY: OLLAMA_DEBUG:true OLLAMA_FLASH_ATTENTION:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11434 OLLAMA_INTEL_GPU:false OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE:q8_0 OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MODELS:/ollama OLLAMA_MULTIUSER_CACHE:false 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://* vscode-webview://*] OLLAMA_SCHED_SPREAD:false ROCR_VISIBLE_DEVICES: http_proxy: https_proxy: no_proxy:]" time=2025-02-06T12:30:37.908+01:00 level=INFO source=images.go:432 msg="total blobs: 118" time=2025-02-06T12:30:37.909+01:00 level=INFO source=images.go:439 msg="total unused blobs removed: 0" time=2025-02-06T12:30:37.910+01:00 level=INFO source=routes.go:1237 msg="Listening on [::]:11434 (version 0.5.8-rc7)" time=2025-02-06T12:30:37.911+01:00 level=DEBUG source=sched.go:105 msg="starting llm scheduler" time=2025-02-06T12:30:37.911+01:00 level=INFO source=gpu.go:217 msg="looking for compatible GPUs" time=2025-02-06T12:30:37.912+01:00 level=DEBUG source=gpu.go:98 msg="searching for GPU discovery libraries for NVIDIA" time=2025-02-06T12:30:37.912+01:00 level=DEBUG source=gpu.go:501 msg="Searching for GPU library" name=libcuda.so* time=2025-02-06T12:30:37.912+01:00 level=DEBUG source=gpu.go:525 msg="gpu library search" globs="[/opt/ollama/bin/libcuda.so* /libcuda.so* /usr/local/cuda*/targets/*/lib/libcuda.so* /usr/lib/*-linux-gnu/nvidia/current/libcuda.so* /usr/lib/*-linux-gnu/libcuda.so* /usr/lib/wsl/lib/libcuda.so* /usr/lib/wsl/drivers/*/libcuda.so* /opt/cuda/lib*/libcuda.so* /usr/local/cuda/lib*/libcuda.so* /usr/lib*/libcuda.so* /usr/local/lib*/libcuda.so*]" time=2025-02-06T12:30:37.924+01:00 level=DEBUG source=gpu.go:558 msg="discovered GPU libraries" paths="[/usr/lib/libcuda.so.565.77 /usr/lib64/libcuda.so.565.77]" initializing /usr/lib/libcuda.so.565.77 library /usr/lib/libcuda.so.565.77 load err: /usr/lib/libcuda.so.565.77: wrong ELF class: ELFCLASS32 time=2025-02-06T12:30:37.924+01:00 level=DEBUG source=gpu.go:609 msg="skipping 32bit library" library=/usr/lib/libcuda.so.565.77 initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:37.931+01:00 level=DEBUG source=gpu.go:125 msg="detected GPUs" count=1 library=/usr/lib64/libcuda.so.565.77 [GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7] CUDA totalMem 4030 mb [GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7] CUDA freeMem 3963 mb [GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7] Compute Capability 5.2 time=2025-02-06T12:30:38.019+01:00 level=WARN source=amd_linux.go:61 msg="ollama recommends running the https://www.amd.com/en/support/linux-drivers" error="amdgpu version file missing: /sys/module/amdgpu/version stat /sys/module/amdgpu/version: no such file or directory" time=2025-02-06T12:30:38.019+01:00 level=DEBUG source=amd_linux.go:101 msg="evaluating amdgpu node /sys/class/kfd/kfd/topology/nodes/0/properties" time=2025-02-06T12:30:38.019+01:00 level=DEBUG source=amd_linux.go:121 msg="detected CPU /sys/class/kfd/kfd/topology/nodes/0/properties" time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:101 msg="evaluating amdgpu node /sys/class/kfd/kfd/topology/nodes/1/properties" time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:206 msg="mapping amdgpu to drm sysfs nodes" amdgpu=/sys/class/kfd/kfd/topology/nodes/1/properties vendor=4098 device=29631 unique_id=10870137312548343375 time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:240 msg=matched amdgpu=/sys/class/kfd/kfd/topology/nodes/1/properties drm=/sys/class/drm/card0/device time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:318 msg="amdgpu memory" gpu=0 total="16.0 GiB" time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_linux.go:319 msg="amdgpu memory" gpu=0 available="15.4 GiB" time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_common.go:16 msg="evaluating potential rocm lib dir /opt/ollama/bin/rocm" time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_common.go:16 msg="evaluating potential rocm lib dir /opt/rocm/lib" time=2025-02-06T12:30:38.020+01:00 level=DEBUG source=amd_common.go:16 msg="evaluating potential rocm lib dir /usr/lib64" time=2025-02-06T12:30:38.025+01:00 level=DEBUG source=amd_linux.go:371 msg="rocm supported GPUs" types="[gfx1030 gfx1100 gfx906 gfx908 gfx90a gfx942]" time=2025-02-06T12:30:38.025+01:00 level=INFO source=amd_linux.go:386 msg="amdgpu is supported" gpu=GPU-96da7c4b1629ce4f gpu_type=gfx1030 releasing cuda driver library time=2025-02-06T12:30:38.025+01:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 library=cuda variant=v11 compute=5.2 driver=12.7 name="NVIDIA GeForce GTX 970" total="3.9 GiB" available="3.9 GiB" time=2025-02-06T12:30:38.025+01:00 level=INFO source=types.go:130 msg="inference compute" id=GPU-96da7c4b1629ce4f library=rocm variant="" compute=gfx1030 driver=0.0 name=1002:73bf total="16.0 GiB" available="15.4 GiB" [GIN] 2025/02/06 - 12:30:44 | 200 | 41.835µs | 127.0.0.1 | HEAD "/" [GIN] 2025/02/06 - 12:30:44 | 200 | 21.973698ms | 127.0.0.1 | POST "/api/show" time=2025-02-06T12:30:44.263+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.2 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.344+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.344+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.344+01:00 level=DEBUG source=sched.go:181 msg="updating default concurrency" OLLAMA_MAX_LOADED_MODELS=0x55e3b630e9c0 gpu_count=2 time=2025-02-06T12:30:44.395+01:00 level=DEBUG source=sched.go:224 msg="loading first model" model=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 time=2025-02-06T12:30:44.395+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=cuda gpu_count=1 available="[3.9 GiB]" time=2025-02-06T12:30:44.396+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=memory.go:186 msg="gpu has too little memory to allocate any layers" id=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 library=cuda variant=v11 compute=5.2 driver=12.7 name="NVIDIA GeForce GTX 970" total="3.9 GiB" available="3.9 GiB" minimum_memory=479199232 layer_size="307.4 MiB" gpu_zer_overhead="0 B" partial_offload="5.9 GiB" full_offload="4.7 GiB" time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=memory.go:330 msg="insufficient VRAM to load any model layers" time=2025-02-06T12:30:44.470+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=cuda gpu_count=1 available="[3.9 GiB]" time=2025-02-06T12:30:44.471+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.0 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.543+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.543+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.544+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=cuda gpu_count=1 available="[3.9 GiB]" time=2025-02-06T12:30:44.544+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.0 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.611+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.611+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.612+01:00 level=DEBUG source=memory.go:186 msg="gpu has too little memory to allocate any layers" id=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 library=cuda variant=v11 compute=5.2 driver=12.7 name="NVIDIA GeForce GTX 970" total="3.9 GiB" available="3.9 GiB" minimum_memory=479199232 layer_size="307.4 MiB" gpu_zer_overhead="0 B" partial_offload="5.9 GiB" full_offload="4.7 GiB" time=2025-02-06T12:30:44.612+01:00 level=DEBUG source=memory.go:330 msg="insufficient VRAM to load any model layers" time=2025-02-06T12:30:44.612+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=cuda gpu_count=1 available="[3.9 GiB]" time=2025-02-06T12:30:44.612+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.681+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.681+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.681+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=rocm gpu_count=1 available="[15.4 GiB]" time=2025-02-06T12:30:44.682+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.750+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.750+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.751+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=rocm gpu_count=1 available="[15.4 GiB]" time=2025-02-06T12:30:44.751+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.820+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.820+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.820+01:00 level=INFO source=sched.go:714 msg="new model will fit in available VRAM in single GPU, loading" model=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 gpu=GPU-96da7c4b1629ce4f parallel=1 available=16494198784 required="12.5 GiB" time=2025-02-06T12:30:44.820+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.888+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.888+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.888+01:00 level=INFO source=server.go:100 msg="system memory" total="62.7 GiB" free="36.1 GiB" free_swap="1.0 GiB" time=2025-02-06T12:30:44.888+01:00 level=DEBUG source=memory.go:107 msg=evaluating library=rocm gpu_count=1 available="[15.4 GiB]" time=2025-02-06T12:30:44.889+01:00 level=DEBUG source=gpu.go:391 msg="updating system memory data" before.total="62.7 GiB" before.free="36.1 GiB" before.free_swap="1.0 GiB" now.total="62.7 GiB" now.free="36.1 GiB" now.free_swap="1.0 GiB" initializing /usr/lib64/libcuda.so.565.77 dlsym: cuInit - 0x7f758d2d4cc0 dlsym: cuDriverGetVersion - 0x7f758d2d4ce0 dlsym: cuDeviceGetCount - 0x7f758d2d4d20 dlsym: cuDeviceGet - 0x7f758d2d4d00 dlsym: cuDeviceGetAttribute - 0x7f758d2d4e00 dlsym: cuDeviceGetUuid - 0x7f758d2d4d60 dlsym: cuDeviceGetName - 0x7f758d2d4d40 dlsym: cuCtxCreate_v3 - 0x7f758d2d4fe0 dlsym: cuMemGetInfo_v2 - 0x7f758d2d5760 dlsym: cuCtxDestroy - 0x7f758d3213a0 calling cuInit calling cuDriverGetVersion raw version 0x2f26 CUDA driver version: 12.7 calling cuDeviceGetCount device count 1 time=2025-02-06T12:30:44.959+01:00 level=DEBUG source=gpu.go:441 msg="updating cuda memory data" gpu=GPU-64fa45ff-fe00-d712-1796-ed74da57bfa7 name="NVIDIA GeForce GTX 970" overhead="0 B" before.total="3.9 GiB" before.free="3.9 GiB" now.total="3.9 GiB" now.free="3.9 GiB" now.used="66.5 MiB" time=2025-02-06T12:30:44.959+01:00 level=DEBUG source=amd_linux.go:488 msg="updating rocm free memory" gpu=GPU-96da7c4b1629ce4f name=1002:73bf before="15.4 GiB" now="15.4 GiB" releasing cuda driver library time=2025-02-06T12:30:44.959+01:00 level=INFO source=memory.go:356 msg="offload to rocm" layers.requested=-1 layers.model=49 layers.offload=49 layers.split="" memory.available="[15.4 GiB]" memory.gpu_overhead="0 B" memory.required.full="12.5 GiB" memory.required.partial="12.5 GiB" memory.required.kv="1.4 GiB" memory.required.allocations="[12.5 GiB]" memory.weights.total="10.1 GiB" memory.weights.repeating="9.5 GiB" memory.weights.nonrepeating="609.1 MiB" memory.graph.full="1.2 GiB" memory.graph.partial="1.5 GiB" time=2025-02-06T12:30:44.959+01:00 level=INFO source=server.go:185 msg="enabling flash attention" time=2025-02-06T12:30:44.959+01:00 level=DEBUG source=server.go:262 msg="compatible gpu libraries" compatible=[] time=2025-02-06T12:30:44.960+01:00 level=DEBUG source=server.go:310 msg="adding gpu dependency paths" paths=[/usr/lib64] time=2025-02-06T12:30:44.960+01:00 level=INFO source=server.go:381 msg="starting llama server" cmd="/opt/ollama/bin/ollama runner --model /ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 --ctx-size 15000 --batch-size 512 --n-gpu-layers 49 --verbose --threads 6 --flash-attn --kv-cache-type q8_0 --parallel 1 --port 46243" time=2025-02-06T12:30:44.960+01:00 level=DEBUG source=server.go:399 msg=subprocess environment="[PATH=/opt/ollama/bin:/bin:/sbin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin:/opt/bin:/usr/lib/llvm/19/bin:/usr/lib/llvm/18/bin:/etc/eselect/wine/bin:/opt/cuda/bin LD_LIBRARY_PATH=/usr/lib64:/opt/ollama/bin ROCR_VISIBLE_DEVICES=GPU-96da7c4b1629ce4f]" time=2025-02-06T12:30:44.960+01:00 level=INFO source=sched.go:449 msg="loaded runners" count=1 time=2025-02-06T12:30:44.961+01:00 level=INFO source=server.go:558 msg="waiting for llama runner to start responding" time=2025-02-06T12:30:44.961+01:00 level=INFO source=server.go:592 msg="waiting for server to become available" status="llm server error" time=2025-02-06T12:30:44.982+01:00 level=INFO source=runner.go:936 msg="starting go runner" time=2025-02-06T12:30:44.982+01:00 level=INFO source=runner.go:937 msg=system info="CPU : LLAMAFILE = 1 | CPU : LLAMAFILE = 1 | cgo(gcc)" threads=6 time=2025-02-06T12:30:44.982+01:00 level=DEBUG source=ggml.go:84 msg="ggml backend load all from path" path=/usr/lib64 time=2025-02-06T12:30:44.982+01:00 level=INFO source=runner.go:995 msg="Server listening on 127.0.0.1:46243" time=2025-02-06T12:30:45.119+01:00 level=DEBUG source=ggml.go:84 msg="ggml backend load all from path" path=/opt/ollama/bin llama_model_loader: loaded meta data with 43 key-value pairs and 579 tensors from /ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 (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 = qwen2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen2.5 14B llama_model_loader: - kv 3: general.organization str = Qwen llama_model_loader: - kv 4: general.basename str = Qwen2.5 llama_model_loader: - kv 5: general.size_label str = 14B llama_model_loader: - kv 6: general.base_model.count u32 = 3 llama_model_loader: - kv 7: general.base_model.0.name str = Qwamma 14b Merge v1 llama_model_loader: - kv 8: general.base_model.0.version str = v1 llama_model_loader: - kv 9: general.base_model.0.organization str = Chargoddard llama_model_loader: - kv 10: general.base_model.0.repo_url str = https://huggingface.co/chargoddard/qw... llama_model_loader: - kv 11: general.base_model.1.name str = Qwen2.5 14B Instruct_arcee Qwen2 14B ... llama_model_loader: - kv 12: general.base_model.1.version str = v0.2 llama_model_loader: - kv 13: general.base_model.1.organization str = Arcee Train llama_model_loader: - kv 14: general.base_model.1.repo_url str = https://huggingface.co/arcee-train/Qw... llama_model_loader: - kv 15: general.base_model.2.name str = Qwen2.5 14B llama_model_loader: - kv 16: general.base_model.2.organization str = Qwen llama_model_loader: - kv 17: general.base_model.2.repo_url str = https://huggingface.co/Qwen/Qwen2.5-14B llama_model_loader: - kv 18: general.tags arr[str,2] = ["mergekit", "merge"] llama_model_loader: - kv 19: qwen2.block_count u32 = 48 llama_model_loader: - kv 20: qwen2.context_length u32 = 131072 llama_model_loader: - kv 21: qwen2.embedding_length u32 = 5120 llama_model_loader: - kv 22: qwen2.feed_forward_length u32 = 13824 llama_model_loader: - kv 23: qwen2.attention.head_count u32 = 40 llama_model_loader: - kv 24: qwen2.attention.head_count_kv u32 = 8 llama_model_loader: - kv 25: qwen2.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 26: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000010 llama_model_loader: - kv 27: general.file_type u32 = 17 llama_model_loader: - kv 28: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 29: tokenizer.ggml.pre str = qwen2 time=2025-02-06T12:30:45.213+01:00 level=INFO source=server.go:592 msg="waiting for server to become available" status="llm server loading model" llama_model_loader: - kv 30: tokenizer.ggml.tokens arr[str,152064] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 31: tokenizer.ggml.token_type arr[i32,152064] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 32: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 33: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 34: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 35: tokenizer.ggml.bos_token_id u32 = 151643 llama_model_loader: - kv 36: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 37: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>... llama_model_loader: - kv 38: general.quantization_version u32 = 2 llama_model_loader: - kv 39: quantize.imatrix.file str = /models_out/SuperNova-14B-GGUF/SuperN... llama_model_loader: - kv 40: quantize.imatrix.dataset str = /training_dir/calibration_datav3.txt llama_model_loader: - kv 41: quantize.imatrix.entries_count i32 = 336 llama_model_loader: - kv 42: quantize.imatrix.chunks_count i32 = 128 llama_model_loader: - type f32: 241 tensors llama_model_loader: - type q5_K: 289 tensors llama_model_loader: - type q6_K: 49 tensors llm_load_vocab: control token: 151660 '<|fim_middle|>' is not marked as EOG llm_load_vocab: control token: 151659 '<|fim_prefix|>' is not marked as EOG llm_load_vocab: control token: 151653 '<|vision_end|>' is not marked as EOG llm_load_vocab: control token: 151648 '<|box_start|>' is not marked as EOG llm_load_vocab: control token: 151646 '<|object_ref_start|>' is not marked as EOG llm_load_vocab: control token: 151649 '<|box_end|>' is not marked as EOG llm_load_vocab: control token: 151655 '<|image_pad|>' is not marked as EOG llm_load_vocab: control token: 151651 '<|quad_end|>' is not marked as EOG llm_load_vocab: control token: 151647 '<|object_ref_end|>' is not marked as EOG llm_load_vocab: control token: 151652 '<|vision_start|>' is not marked as EOG llm_load_vocab: control token: 151654 '<|vision_pad|>' is not marked as EOG llm_load_vocab: control token: 151656 '<|video_pad|>' is not marked as EOG llm_load_vocab: control token: 151644 '<|im_start|>' is not marked as EOG llm_load_vocab: control token: 151661 '<|fim_suffix|>' is not marked as EOG llm_load_vocab: control token: 151650 '<|quad_start|>' is not marked as EOG llm_load_vocab: special tokens cache size = 22 llm_load_vocab: token to piece cache size = 0.9310 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = qwen2 llm_load_print_meta: vocab type = BPE llm_load_print_meta: n_vocab = 152064 llm_load_print_meta: n_merges = 151387 llm_load_print_meta: vocab_only = 0 llm_load_print_meta: n_ctx_train = 131072 llm_load_print_meta: n_embd = 5120 llm_load_print_meta: n_layer = 48 llm_load_print_meta: n_head = 40 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 = 5 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 = 13824 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 = 2 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 = 131072 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: ssm_dt_b_c_rms = 0 llm_load_print_meta: model type = 14B llm_load_print_meta: model ftype = Q5_K - Medium llm_load_print_meta: model params = 14.77 B llm_load_print_meta: model size = 9.78 GiB (5.69 BPW) llm_load_print_meta: general.name = Qwen2.5 14B llm_load_print_meta: BOS token = 151643 '<|endoftext|>' llm_load_print_meta: EOS token = 151645 '<|im_end|>' llm_load_print_meta: EOT token = 151645 '<|im_end|>' llm_load_print_meta: PAD token = 151643 '<|endoftext|>' llm_load_print_meta: LF token = 148848 'ÄĬ' llm_load_print_meta: FIM PRE token = 151659 '<|fim_prefix|>' llm_load_print_meta: FIM SUF token = 151661 '<|fim_suffix|>' llm_load_print_meta: FIM MID token = 151660 '<|fim_middle|>' llm_load_print_meta: FIM PAD token = 151662 '<|fim_pad|>' llm_load_print_meta: FIM REP token = 151663 '<|repo_name|>' llm_load_print_meta: FIM SEP token = 151664 '<|file_sep|>' llm_load_print_meta: EOG token = 151643 '<|endoftext|>' llm_load_print_meta: EOG token = 151645 '<|im_end|>' llm_load_print_meta: EOG token = 151662 '<|fim_pad|>' llm_load_print_meta: EOG token = 151663 '<|repo_name|>' llm_load_print_meta: EOG token = 151664 '<|file_sep|>' llm_load_print_meta: max token length = 256 llm_load_tensors: CPU_Mapped model buffer size = 10016.35 MiB llama_new_context_with_model: n_seq_max = 1 llama_new_context_with_model: n_ctx = 15104 llama_new_context_with_model: n_ctx_per_seq = 15104 llama_new_context_with_model: n_batch = 512 llama_new_context_with_model: n_ubatch = 512 llama_new_context_with_model: flash_attn = 1 llama_new_context_with_model: freq_base = 1000000.0 llama_new_context_with_model: freq_scale = 1 llama_new_context_with_model: n_ctx_per_seq (15104) < n_ctx_train (131072) -- the full capacity of the model will not be utilized llama_kv_cache_init: kv_size = 15104, offload = 1, type_k = 'q8_0', type_v = 'q8_0', n_layer = 48, can_shift = 1 llama_kv_cache_init: layer 0: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 1: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 2: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 3: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 4: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 5: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 6: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 7: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 8: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 9: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 10: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 11: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 12: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 13: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 14: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 15: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 16: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 17: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 18: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 19: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 20: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 21: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 22: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 23: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 24: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 25: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 26: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 27: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 28: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 29: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 30: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 31: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 32: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 33: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 34: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 35: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 36: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 37: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 38: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 39: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 40: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 41: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 42: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 43: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 44: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 45: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 46: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 llama_kv_cache_init: layer 47: n_embd_k_gqa = 1024, n_embd_v_gqa = 1024 time=2025-02-06T12:30:46.971+01:00 level=DEBUG source=server.go:603 msg="model load progress 1.00" time=2025-02-06T12:30:47.221+01:00 level=DEBUG source=server.go:606 msg="model load completed, waiting for server to become available" status="llm server loading model" llama_kv_cache_init: CPU KV buffer size = 1504.50 MiB llama_new_context_with_model: KV self size = 1504.50 MiB, K (q8_0): 752.25 MiB, V (q8_0): 752.25 MiB llama_new_context_with_model: CPU output buffer size = 0.60 MiB llama_new_context_with_model: CPU compute buffer size = 317.00 MiB llama_new_context_with_model: graph nodes = 1495 llama_new_context_with_model: graph splits = 1 time=2025-02-06T12:30:47.974+01:00 level=INFO source=server.go:597 msg="llama runner started in 3.01 seconds" time=2025-02-06T12:30:47.974+01:00 level=DEBUG source=sched.go:462 msg="finished setting up runner" model=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 [GIN] 2025/02/06 - 12:30:47 | 200 | 3.733069514s | 127.0.0.1 | POST "/api/generate" time=2025-02-06T12:30:47.974+01:00 level=DEBUG source=sched.go:466 msg="context for request finished" time=2025-02-06T12:30:47.975+01:00 level=DEBUG source=sched.go:339 msg="runner with non-zero duration has gone idle, adding timer" modelPath=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 duration=5m0s time=2025-02-06T12:30:47.975+01:00 level=DEBUG source=sched.go:357 msg="after processing request finished event" modelPath=/ollama/blobs/sha256-bbfb685133c274407d565c65b1ca806eb1593482b1c9d8524596797b24123862 refCount=0 ``` </details> ### OS Linux ### GPU Nvidia, AMD ### CPU AMD ### Ollama version 0.5.8-rc7
GiteaMirror added the buildamdbug labels 2026-04-12 17:04:48 -05:00
GiteaMirror commented 2026-04-12 17:04:50 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 6, 2025):

time=2025-02-06T12:30:44.959+01:00 level=DEBUG source=server.go:262 msg="compatible gpu libraries" compatible=[]

No runners found. What's the result of find /opt/ollama/?

<!-- gh-comment-id:2639617827 --> @rick-github commented on GitHub (Feb 6, 2025): ``` time=2025-02-06T12:30:44.959+01:00 level=DEBUG source=server.go:262 msg="compatible gpu libraries" compatible=[] ``` No runners found. What's the result of `find /opt/ollama/`?
GiteaMirror commented 2026-04-12 17:04:51 -05:00
Author
Owner
Copy Link

@ProjectMoon commented on GitHub (Feb 6, 2025):

In both scenarios, ollama uses CPU instead of GPU. untar.txt is the directory structure after just un-tarring the linux and linux-rocm.tgz files. merged.txt is after cp -ring all of the separate library directories into one place.

untar.txt
merged.txt

Edit:

My installation process has always been to download the linux and linux-rocm .tgz files from GitHub, untar them into /opt/ollama, and restart the service.

<!-- gh-comment-id:2639799719 --> @ProjectMoon commented on GitHub (Feb 6, 2025): In both scenarios, ollama uses CPU instead of GPU. untar.txt is the directory structure after just un-tarring the linux and linux-rocm.tgz files. merged.txt is after `cp -r`ing all of the separate library directories into one place. [untar.txt](https://github.com/user-attachments/files/18690454/untar.txt) [merged.txt](https://github.com/user-attachments/files/18690453/merged.txt) Edit: My installation process has always been to download the linux and linux-rocm .tgz files from GitHub, untar them into `/opt/ollama`, and restart the service.
GiteaMirror commented 2026-04-12 17:04:52 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 6, 2025):

The bundle for 0.5.8 appears to be missing a bunch of libraries that are in the 0.5.7 bundle.

8,15c6
< ./lib/ollama/libdrm_amdgpu.so.1
< ./lib/ollama/libdrm_amdgpu.so.1.0.0
< ./lib/ollama/libdrm.so.2
< ./lib/ollama/libdrm.so.2.4.0
< ./lib/ollama/libelf-0.176.so
< ./lib/ollama/libelf.so.1
< ./lib/ollama/libhipblaslt.so.0
< ./lib/ollama/libhipblaslt.so.0.7.60102
---
> ./lib/ollama/libggml-hip.so
18,19d8
< ./lib/ollama/libhsa-runtime64.so.1
< ./lib/ollama/libhsa-runtime64.so.1.13.60102
24,31d12
< ./lib/ollama/librocprofiler-register.so.0
< ./lib/ollama/librocprofiler-register.so.0.3.0
< ./lib/ollama/librocsolver.so.0
< ./lib/ollama/librocsolver.so.0.1.60102
< ./lib/ollama/librocsparse.so.1
< ./lib/ollama/librocsparse.so.1.0.60102
< ./lib/ollama/libtinfo.so.5
< ./lib/ollama/libtinfo.so.5.9
<!-- gh-comment-id:2639916084 --> @rick-github commented on GitHub (Feb 6, 2025): The bundle for 0.5.8 appears to be missing a bunch of libraries that are in the 0.5.7 bundle. ```diff 8,15c6 < ./lib/ollama/libdrm_amdgpu.so.1 < ./lib/ollama/libdrm_amdgpu.so.1.0.0 < ./lib/ollama/libdrm.so.2 < ./lib/ollama/libdrm.so.2.4.0 < ./lib/ollama/libelf-0.176.so < ./lib/ollama/libelf.so.1 < ./lib/ollama/libhipblaslt.so.0 < ./lib/ollama/libhipblaslt.so.0.7.60102 --- > ./lib/ollama/libggml-hip.so 18,19d8 < ./lib/ollama/libhsa-runtime64.so.1 < ./lib/ollama/libhsa-runtime64.so.1.13.60102 24,31d12 < ./lib/ollama/librocprofiler-register.so.0 < ./lib/ollama/librocprofiler-register.so.0.3.0 < ./lib/ollama/librocsolver.so.0 < ./lib/ollama/librocsolver.so.0.1.60102 < ./lib/ollama/librocsparse.so.1 < ./lib/ollama/librocsparse.so.1.0.60102 < ./lib/ollama/libtinfo.so.5 < ./lib/ollama/libtinfo.so.5.9 ```
GiteaMirror commented 2026-04-12 17:04:53 -05:00
Author
Owner
Copy Link

@ProjectMoon commented on GitHub (Feb 6, 2025):

Well that would cause problems, I think. 😄

<!-- gh-comment-id:2639922701 --> @ProjectMoon commented on GitHub (Feb 6, 2025): Well that would cause problems, I think. 😄
GiteaMirror commented 2026-04-12 17:04:54 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Feb 6, 2025):

The libraries are not in the docker image either. There's been significant changes in the build procedure for 0.5.8. I don't use ROCm so I don't know if the missing libraries are made up by the addition of libggml-hip.so, but that's quite a few libs and several gigabytes that have been removed.

ollama/ollama   0.5.8-rc7-rocm     ac861eae2ac6   19 hours ago    5.38GB
ollama/ollama   0.5.7-rocm         4b1f3ce64bfb   2 weeks ago     8.28GB
<!-- gh-comment-id:2639949435 --> @rick-github commented on GitHub (Feb 6, 2025): The libraries are not in the docker image either. There's been significant changes in the build procedure for 0.5.8. I don't use ROCm so I don't know if the missing libraries are made up by the addition of `libggml-hip.so`, but that's quite a few libs and several gigabytes that have been removed. ``` ollama/ollama 0.5.8-rc7-rocm ac861eae2ac6 19 hours ago 5.38GB ollama/ollama 0.5.7-rocm 4b1f3ce64bfb 2 weeks ago 8.28GB ```
GiteaMirror commented 2026-04-12 17:04:56 -05:00
Author
Owner
Copy Link

@mattcaron commented on GitHub (Feb 6, 2025):

I can confirm this in 0.5.8-rc7 as well, on a Radeon 6900XT.

My install process is:

cd ~/ai/ollama
mkdir 0.5.8
cd 0.5.8
tar -xf ~/download/ollama-linux-amd64.tgz
tar -xf ~/download/ollama-linux-amd64-rocm.tgz
cd ..
rm current && ln -s 0.5.8 current

And then my scripts know where to find everything because they look in current.

There's a minor issue here in that the directory structure changed and now everything lives under dist/linux/amd64 whereas before it did not. If this is intended, then fine, I can adjust my symlink trickery so things can find it.

However, when I run ollama manually, jobs still gets scheduled on the CPU and run painfully slowly.

Thanks in advance.

<!-- gh-comment-id:2640853658 --> @mattcaron commented on GitHub (Feb 6, 2025): I can confirm this in `0.5.8-rc7` as well, on a Radeon 6900XT. My install process is: ``` cd ~/ai/ollama mkdir 0.5.8 cd 0.5.8 tar -xf ~/download/ollama-linux-amd64.tgz tar -xf ~/download/ollama-linux-amd64-rocm.tgz cd .. rm current && ln -s 0.5.8 current ``` And then my scripts know where to find everything because they look in `current`. There's a minor issue here in that the directory structure changed and now everything lives under `dist/linux/amd64` whereas before it did not. If this is intended, then fine, I can adjust my symlink trickery so things can find it. However, when I run `ollama` manually, jobs still gets scheduled on the CPU and run painfully slowly. Thanks in advance.
GiteaMirror commented 2026-04-12 17:04:57 -05:00
Author
Owner
Copy Link

@mxyng commented on GitHub (Feb 6, 2025):

this should be fixed by #8896 and #8899

<!-- gh-comment-id:2641326533 --> @mxyng commented on GitHub (Feb 6, 2025): this should be fixed by #8896 and #8899
GiteaMirror commented 2026-04-12 17:04:58 -05:00
Author
Owner
Copy Link

@ProjectMoon commented on GitHub (Feb 7, 2025):

Looks like it works now. Thanks for the quick fix.

<!-- gh-comment-id:2642412647 --> @ProjectMoon commented on GitHub (Feb 7, 2025): Looks like it works now. Thanks for the quick fix.
GiteaMirror commented 2026-04-12 17:04:59 -05:00
Author
Owner
Copy Link

@mattcaron commented on GitHub (Feb 10, 2025):

Can confirm fixed for me in 0.5.8-rc12 as well.

Also, the tarball hierarchy is back to matching 0.5.7, which makes me happy.

Thanks for the quick fix.

<!-- gh-comment-id:2648310875 --> @mattcaron commented on GitHub (Feb 10, 2025): Can confirm fixed for me in `0.5.8-rc12` as well. Also, the tarball hierarchy is back to matching `0.5.7`, which makes me happy. Thanks for the quick fix.
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 build
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#5759
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?