Issue Station
Sign In
Tools hereToolbox

ollama - 💡(How to fix) Fix 0.30.6 Regression with 7800XT

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…

Error Message

Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.054Z level=WARN source=server.go:114 msg="requested context size too large for model" num_ctx=64000 n_ctx_train=16384 Jun 08 13:08:30 ai ollama[2299560]: /build/llama-server-cpu/_deps/llama_cpp-src/ggml/src/ggml-impl.h:318: fatal error Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.345Z level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server error" Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.595Z level=INFO source=sched.go:641 msg="Load failed" model=/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e error="llama-server process has terminated: signal: aborted (core dumped)"

Code Example

[Service]
ExecStart=
ExecStart=/usr/local/bin/ollama serve
Environment="HSA_OVERRIDE_GFX_VERSION=11.0.1"
Environment="OLLAMA_HOST=0.0.0.0:11434"
Environment="OLLAMA_ORIGINS=chrome-extension://*,moz-extension://*,safari-web-extension://* ollama serve"
Environment="LD_LIBRARY_PATH=/opt/rocm/lib"
Environment="OLLAMA_LIBRARY_PATH=/opt/rocm/lib"
LimitMEMLOCK=infinity
Environment="HSA_ENABLE_SDMA=0"
Environment="OLLAMA_FLASH_ATTENTION=1"
# How many requests to process in parallel (default 1)
Environment="OLLAMA_NUM_PARALLEL=2"

# Keep models loaded between requests (in seconds, -1 = forever)
Environment="OLLAMA_KEEP_ALIVE=60m"
Environment="OLLAMA_CONTEXT_LENGTH=64000"
# THE PRO MOVE: Use Quantized KV Cache to save VRAM on your 16GB card
Environment="OLLAMA_KV_CACHE_TYPE=q8_0"

---

Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.054Z level=INFO source=server.go:109 msg="using llama-server for model" model=/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.054Z level=WARN source=server.go:114 msg="requested context size too large for model" num_ctx=64000 n_ctx_train=16384
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.054Z level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="/usr/local/lib/ollama/llama-server --model /usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e --port 36143 --host 127.0.0.1 --no-webui --offline -c 32768 -np 2 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-jinja --chat-template chatml --cache-type-k q8_0 --cache-type-v q8_0 --flash-attn on -b 512 -ub 512"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=sched.go:613 msg="system memory" total="62.7 GiB" free="58.5 GiB" free_swap="8.0 GiB"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=sched.go:620 msg="gpu memory" id=0 library=ROCm available="15.4 GiB" free="15.8 GiB" minimum="457.0 MiB" overhead="0 B"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server not responding"
Jun 08 13:08:30 ai ollama[2299560]: common_params_print_info: build 1 (6f3a9f3de) with GNU 11.2.1 for Linux x86_64
Jun 08 13:08:30 ai ollama[2299560]: log_info: verbosity = 4 (adjust with the `-lv N` CLI arg)
Jun 08 13:08:30 ai ollama[2299560]: device_info:
Jun 08 13:08:30 ai ollama[2299560]:   - CPU     : AMD Ryzen 7 5700X 8-Core Processor (64204 MiB, 64204 MiB free)
Jun 08 13:08:30 ai ollama[2299560]:   - ROCm0   : AMD Radeon RX 7800 XT (16368 MiB, 16184 MiB free)
Jun 08 13:08:30 ai ollama[2299560]: system_info: n_threads = 8 (n_threads_batch = 8) / 16 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | REPACK = 1 | ROCm : NO_VMM = 1 | PEER_MAX_BATCH_SIZE = 128 |
Jun 08 13:08:30 ai ollama[2299560]: srv          init: using 15 threads for HTTP server
Jun 08 13:08:30 ai ollama[2299560]: srv          init: The UI is disabled
Jun 08 13:08:30 ai ollama[2299560]: srv          init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable
Jun 08 13:08:30 ai ollama[2299560]: srv         start: binding port with default address family
Jun 08 13:08:30 ai ollama[2299560]: srv  llama_server: loading model
Jun 08 13:08:30 ai ollama[2299560]: srv    load_model: loading model '/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e'
Jun 08 13:08:30 ai ollama[2299560]: common_init_result: fitting params to device memory ...
Jun 08 13:08:30 ai ollama[2299560]: common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)
Jun 08 13:08:30 ai ollama[2299560]: common_params_fit_impl: getting device memory data for initial parameters:
Jun 08 13:08:30 ai ollama[2299560]: /build/llama-server-cpu/_deps/llama_cpp-src/ggml/src/ggml-impl.h:318: fatal error
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libggml-base.so.0(+0x1a1c8)[0x74a4d2f561c8]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libggml-base.so.0(ggml_print_backtrace+0x1e6)[0x74a4d2f565a6]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libggml-base.so.0(ggml_abort+0x11d)[0x74a4d2f5672d]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libggml-base.so.0(ggml_backend_sched_split_graph+0x17d8)[0x74a4d2f70e98]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama.so.0(_ZN13llama_context13graph_reserveEjjjPK22llama_memory_context_ibPm+0x5d1)[0x74a4d2cd9421]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama.so.0(_ZN13llama_context13sched_reserveEv+0xf42)[0x74a4d2cda462]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama.so.0(_ZN13llama_contextC1ERK11llama_model20llama_context_params+0xb29)[0x74a4d2cdda29]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama.so.0(llama_init_from_model+0x1c9)[0x74a4d2cde709]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(_Z29common_get_device_memory_dataPKcPK18llama_model_paramsPK20llama_context_paramsRSt6vectorIP19ggml_backend_deviceSaIS9_EERjSD_SD_14ggml_log_level+0x153)[0x74a4d3214203]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(+0x214e73)[0x74a4d3214e73]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(_Z17common_fit_paramsPKcP18llama_model_paramsP20llama_context_paramsPfP32llama_model_tensor_buft_overridePmj14ggml_log_level+0x4e)[0x74a4d321886e]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(_ZN18common_init_resultC2ER13common_paramsb+0xe2)[0x74a4d31ebcc2]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(_Z23common_init_from_paramsR13common_paramsb+0x39)[0x74a4d31ed2a9]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-server-impl.so(_ZN19server_context_impl10load_modelER13common_params+0x58b)[0x74a4d3759f3b]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-server-impl.so(_Z12llama_serveriPPc+0x21a6)[0x74a4d36b0e26]
Jun 08 13:08:30 ai ollama[2299560]: /lib/x86_64-linux-gnu/libc.so.6(+0x2a1ca)[0x74a4d242a1ca]
Jun 08 13:08:30 ai ollama[2299560]: /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0x8b)[0x74a4d242a28b]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/llama-server[0x40107e]
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.345Z level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server error"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.595Z level=INFO source=sched.go:641 msg="Load failed" model=/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e error="llama-server process has terminated: signal: aborted (core dumped)"
Jun 08 13:08:30 ai ollama[2299560]: [GIN] 2026/06/08 - 13:08:30 | 500 |  1.612389955s |    10.1.242.239 | POST     "/api/chat"
RAW_BUFFERClick to expand / collapse

What is the issue?

I'm starting to see crashes with 0.30.6. It consistently crashes with the new gemma4:26b-a4b-it-qat but it also crashes with several other models.

It seems to be struggling with the context size set, despite this working with 0.30.4.

This is running on Ubuntu server 24.04 with 7800XT rocm on a amd 5700x CPU with a b550 chipset. The display out is going through a nvidia 710 that's not used at all for AI workloads so the 7800XT is 100% dedicated to AI. No x11/wayland running.

I've also tested with phi 4 with a smaller context length (32000 instead of 64000) phi4:latest fails to respond correctly with all ??????? to hello.

models that work with smaller context mistral:7b does work.
deepseek-coder:6.7b seems to be able to run OK. mistral-nemo:latest also works with small queries. gpt-oss:20b also seems to work.

overrides / config

[Service]
ExecStart=
ExecStart=/usr/local/bin/ollama serve
Environment="HSA_OVERRIDE_GFX_VERSION=11.0.1"
Environment="OLLAMA_HOST=0.0.0.0:11434"
Environment="OLLAMA_ORIGINS=chrome-extension://*,moz-extension://*,safari-web-extension://* ollama serve"
Environment="LD_LIBRARY_PATH=/opt/rocm/lib"
Environment="OLLAMA_LIBRARY_PATH=/opt/rocm/lib"
LimitMEMLOCK=infinity
Environment="HSA_ENABLE_SDMA=0"
Environment="OLLAMA_FLASH_ATTENTION=1"
# How many requests to process in parallel (default 1)
Environment="OLLAMA_NUM_PARALLEL=2"

# Keep models loaded between requests (in seconds, -1 = forever)
Environment="OLLAMA_KEEP_ALIVE=60m"
Environment="OLLAMA_CONTEXT_LENGTH=64000"
# THE PRO MOVE: Use Quantized KV Cache to save VRAM on your 16GB card
Environment="OLLAMA_KV_CACHE_TYPE=q8_0"

Relevant log output

Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.054Z level=INFO source=server.go:109 msg="using llama-server for model" model=/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.054Z level=WARN source=server.go:114 msg="requested context size too large for model" num_ctx=64000 n_ctx_train=16384
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.054Z level=INFO source=llama_server.go:403 msg="starting llama-server" cmd="/usr/local/lib/ollama/llama-server --model /usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e --port 36143 --host 127.0.0.1 --no-webui --offline -c 32768 -np 2 --log-verbosity 4 --no-log-prefix --no-log-timestamps --no-jinja --chat-template chatml --cache-type-k q8_0 --cache-type-v q8_0 --flash-attn on -b 512 -ub 512"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=sched.go:613 msg="system memory" total="62.7 GiB" free="58.5 GiB" free_swap="8.0 GiB"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=sched.go:620 msg="gpu memory" id=0 library=ROCm available="15.4 GiB" free="15.8 GiB" minimum="457.0 MiB" overhead="0 B"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=llama_server.go:886 msg="loading model via llama-server" model=/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=llama_server.go:1131 msg="waiting for llama-server to start responding"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.055Z level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server not responding"
Jun 08 13:08:30 ai ollama[2299560]: common_params_print_info: build 1 (6f3a9f3de) with GNU 11.2.1 for Linux x86_64
Jun 08 13:08:30 ai ollama[2299560]: log_info: verbosity = 4 (adjust with the `-lv N` CLI arg)
Jun 08 13:08:30 ai ollama[2299560]: device_info:
Jun 08 13:08:30 ai ollama[2299560]:   - CPU     : AMD Ryzen 7 5700X 8-Core Processor (64204 MiB, 64204 MiB free)
Jun 08 13:08:30 ai ollama[2299560]:   - ROCm0   : AMD Radeon RX 7800 XT (16368 MiB, 16184 MiB free)
Jun 08 13:08:30 ai ollama[2299560]: system_info: n_threads = 8 (n_threads_batch = 8) / 16 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | REPACK = 1 | ROCm : NO_VMM = 1 | PEER_MAX_BATCH_SIZE = 128 |
Jun 08 13:08:30 ai ollama[2299560]: srv          init: using 15 threads for HTTP server
Jun 08 13:08:30 ai ollama[2299560]: srv          init: The UI is disabled
Jun 08 13:08:30 ai ollama[2299560]: srv          init: Use --ui/--no-ui (or deprecated --webui/--no-webui) to enable/disable
Jun 08 13:08:30 ai ollama[2299560]: srv         start: binding port with default address family
Jun 08 13:08:30 ai ollama[2299560]: srv  llama_server: loading model
Jun 08 13:08:30 ai ollama[2299560]: srv    load_model: loading model '/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e'
Jun 08 13:08:30 ai ollama[2299560]: common_init_result: fitting params to device memory ...
Jun 08 13:08:30 ai ollama[2299560]: common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)
Jun 08 13:08:30 ai ollama[2299560]: common_params_fit_impl: getting device memory data for initial parameters:
Jun 08 13:08:30 ai ollama[2299560]: /build/llama-server-cpu/_deps/llama_cpp-src/ggml/src/ggml-impl.h:318: fatal error
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libggml-base.so.0(+0x1a1c8)[0x74a4d2f561c8]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libggml-base.so.0(ggml_print_backtrace+0x1e6)[0x74a4d2f565a6]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libggml-base.so.0(ggml_abort+0x11d)[0x74a4d2f5672d]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libggml-base.so.0(ggml_backend_sched_split_graph+0x17d8)[0x74a4d2f70e98]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama.so.0(_ZN13llama_context13graph_reserveEjjjPK22llama_memory_context_ibPm+0x5d1)[0x74a4d2cd9421]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama.so.0(_ZN13llama_context13sched_reserveEv+0xf42)[0x74a4d2cda462]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama.so.0(_ZN13llama_contextC1ERK11llama_model20llama_context_params+0xb29)[0x74a4d2cdda29]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama.so.0(llama_init_from_model+0x1c9)[0x74a4d2cde709]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(_Z29common_get_device_memory_dataPKcPK18llama_model_paramsPK20llama_context_paramsRSt6vectorIP19ggml_backend_deviceSaIS9_EERjSD_SD_14ggml_log_level+0x153)[0x74a4d3214203]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(+0x214e73)[0x74a4d3214e73]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(_Z17common_fit_paramsPKcP18llama_model_paramsP20llama_context_paramsPfP32llama_model_tensor_buft_overridePmj14ggml_log_level+0x4e)[0x74a4d321886e]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(_ZN18common_init_resultC2ER13common_paramsb+0xe2)[0x74a4d31ebcc2]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-common.so.0(_Z23common_init_from_paramsR13common_paramsb+0x39)[0x74a4d31ed2a9]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-server-impl.so(_ZN19server_context_impl10load_modelER13common_params+0x58b)[0x74a4d3759f3b]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/libllama-server-impl.so(_Z12llama_serveriPPc+0x21a6)[0x74a4d36b0e26]
Jun 08 13:08:30 ai ollama[2299560]: /lib/x86_64-linux-gnu/libc.so.6(+0x2a1ca)[0x74a4d242a1ca]
Jun 08 13:08:30 ai ollama[2299560]: /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0x8b)[0x74a4d242a28b]
Jun 08 13:08:30 ai ollama[2299560]: /usr/local/lib/ollama/llama-server[0x40107e]
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.345Z level=INFO source=llama_server.go:1186 msg="waiting for llama-server to become available" status="llm server error"
Jun 08 13:08:30 ai ollama[2299560]: time=2026-06-08T13:08:30.595Z level=INFO source=sched.go:641 msg="Load failed" model=/usr/share/ollama/.ollama/models/blobs/sha256-8a8189132ef70cd737e136d38b706e8cb832f9cdf3c744547694769e9313550e error="llama-server process has terminated: signal: aborted (core dumped)"
Jun 08 13:08:30 ai ollama[2299560]: [GIN] 2026/06/08 - 13:08:30 | 500 |  1.612389955s |    10.1.242.239 | POST     "/api/chat"

OS

Linux

GPU

AMD

CPU

AMD

Ollama version

0.30.6

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING