kvcache-ai/ktransformers

Fork 0

mirror of https://github.com/kvcache-ai/ktransformers.git synced 2026-11-05 01:07:28 +00:00

[GH-ISSUE #422] 使用v0.3.0加载模型会出现Illegal instruction (core dumped) #287

New issue

Closed

opened 2026-10-29 14:37:19 +00:00 by bsakaguchi · 5 comments

bsakaguchi commented

2026-10-29 14:37:19 +00:00

Owner

Originally created by @P-D-E-E on GitHub (Feb 17, 2025).
Original GitHub issue: https://github.com/kvcache-ai/ktransformers/issues/422

使用的指令是：
(ktransformers_v0.3.0) fin@controller-node-1:~$ python -m ktransformers.local_chat --model_path ~/q_model/DeepSeek-R1 --gguf_path ~/.cache/huggingface/hub/models--bartowski--DeepSeek-R1-GGUF/snapshots/2b8aeab13b423f67a1a09f2b640811185a9a71ba/DeepSeek-R1-Q4_K_M/ --cpu_infer 193 --max_new_tokens 10000

在Injecting阶段会出现DeepSeekV2Model的字样，并且Illegal instruction (core dumped)，跑0.2.0的时候并没有这个问题，以下为运行后的输出内容：using custom modeling_xxx.py.
using default_optimize_rule for DeepseekV3ForCausalLM
AVX512F
Injecting model as ktransformers.operators.models . KDeepseekV2Model
Injecting model.embed_tokens as default
Injecting model.layers as default
Injecting model.layers.0 as default
Injecting model.layers.0.self_attn as ktransformers.operators.attention . KDeepseekV2Attention
Illegal instruction (core dumped)

Originally created by @P-D-E-E on GitHub (Feb 17, 2025). Original GitHub issue: https://github.com/kvcache-ai/ktransformers/issues/422 使用的指令是： (ktransformers_v0.3.0) fin@controller-node-1:~$ python -m ktransformers.local_chat --model_path ~/q_model/DeepSeek-R1 --gguf_path ~/.cache/huggingface/hub/models--bartowski--DeepSeek-R1-GGUF/snapshots/2b8aeab13b423f67a1a09f2b640811185a9a71ba/DeepSeek-R1-Q4_K_M/ --cpu_infer 193 --max_new_tokens 10000 在Injecting阶段会出现DeepSeekV2Model的字样，并且Illegal instruction (core dumped)，跑0.2.0的时候并没有这个问题，以下为运行后的输出内容：using custom modeling_xxx.py. using default_optimize_rule for DeepseekV3ForCausalLM __AVX512F__ Injecting model as ktransformers.operators.models . KDeepseekV2Model Injecting model.embed_tokens as default Injecting model.layers as default Injecting model.layers.0 as default Injecting model.layers.0.self_attn as ktransformers.operators.attention . KDeepseekV2Attention Illegal instruction (core dumped)

bsakaguchi

2026-10-29 14:37:20 +00:00

closed this issue
added the
stale
label

bsakaguchi commented

2026-10-29 14:37:20 +00:00

Author

Owner

@P-D-E-E commented on GitHub (Feb 17, 2025):

硬件信息：双路9004 + 1.5TB内存 + h100-80GB

@P-D-E-E commented on GitHub (Feb 17, 2025): 硬件信息：双路9004 + 1.5TB内存 + h100-80GB

bsakaguchi commented

2026-10-29 14:37:20 +00:00

Author

Owner

@cskkx1 commented on GitHub (Feb 18, 2025):

应该是CPU不支持AMX指令集加速

@cskkx1 commented on GitHub (Feb 18, 2025): 应该是CPU不支持AMX指令集加速

bsakaguchi commented

2026-10-29 14:37:21 +00:00

Author

Owner

@yileld commented on GitHub (Feb 18, 2025):

应该是CPU不支持AMX指令集加速

必须要amx的机器才能跑吗？

@yileld commented on GitHub (Feb 18, 2025): > 应该是CPU不支持AMX指令集加速必须要amx的机器才能跑吗？

bsakaguchi commented

2026-10-29 14:37:21 +00:00

Author

Owner

@P-D-E-E commented on GitHub (Feb 18, 2025):

应该是CPU不支持AMX指令集加速

是的，后面查了下，确实是不支持。

@P-D-E-E commented on GitHub (Feb 18, 2025): > 应该是CPU不支持AMX指令集加速是的，后面查了下，确实是不支持。

bsakaguchi commented

2026-10-29 14:37:21 +00:00

Author

Owner

@ErvinXie commented on GitHub (Dec 11, 2025):

This issue has been automatically closed due to inactivity for 60 days.

If you believe this issue is still relevant, please feel free to reopen it with additional information or context.

Thank you for your contribution to KTransformers!

此 issue 因超过 60 天无活动而被自动关闭。

如果您认为此问题仍然存在，请随时重新打开并提供更多信息或上下文。

感谢您对 KTransformers 的贡献！

@ErvinXie commented on GitHub (Dec 11, 2025): This issue has been automatically closed due to inactivity for 60 days. If you believe this issue is still relevant, please feel free to reopen it with additional information or context. Thank you for your contribution to KTransformers! --- 此 issue 因超过 60 天无活动而被自动关闭。如果您认为此问题仍然存在，请随时重新打开并提供更多信息或上下文。感谢您对 KTransformers 的贡献！