nullname

hmmm, may I ask which Android version and device model you're using? looks like `libQnnSystem.so` relies on `libstdc++.so.6` and has failed to load

> android 15 xiaomi 14 pro hyperos2.0.104.0 interesting.... my phone is also android 15, but there's no such error come out.... and this is symbol table dump of my `libQnnSystem.so`...

please make sure copy those so from the correct folder first as i said before, can have a look on my compile script: https://github.com/chraac/llama-cpp-qnn-builder/blob/dd7ba303a8e3213c8cafe330c0938b25c6bd788f/docker/build_in_container.sh#L35 

hmmm, i mean the required shared object (.so) files should be copied from the QNN SDK path at: `/qairt/2.31.0.250130/lib/aarch64-android` where `` refers to the base directory of your QNN SDK...

yeah, sure, feel free to use chinese here

问下你怎么跑的程序？看到你这里的几个lib和我本地的版本大小都不一样，应该是复制错了 

> 好像我的qnn ndk太老了，更新完，现在出新问题了( llama_init_from_model: n_ctx_per_seq (4096) < n_ctx_train (131072) -- the full capacity of the model will not be utilized extend_lib_search_path is nullptr, will use /data/local/tmp as defaultfailed to load...

性能问题的话可以关注下这个issue: https://github.com/chraac/llama.cpp/issues/34

> 我按照这个编译命令完成编译并在骁龙8gem3上成功运行，显示有qnn-cpu和qnn-gpu参与，但速度奇慢，decode只有0.02token/s，如下图（这是有qnn-cpu和qnn-gpu参与的意思吗？）  看截图这里确实用到了npu/gpu/cpu QNN这个backend用到NPU的话确实是比较慢的，现在的QNN框架下，除了自己写个Op加进去可能也没有啥其他好办法优化

> 尝试使能GGML_QNN_ENABLE_HEXAGON_BACKEND，提示需要指定HEXAGON_SDK，但我按照https://docs.qualcomm.com/bundle/publicresource/topics/80-77512-1/hexagon-dsp-sdk-getting-started.html?product=1601111740010422 中描述的步骤进行下载，安装了高通软件包下载器qpm3，但始终无法login（我确信账号密码是正确的），导致无法完成hexagon sdk的下载。 1. 这里你需要有qualcomm的账号，然后在cn的话还需要处理一些网络连接的问题，才能成功登录 2. 成功登录，并下载完成qualcomm的hexagon-sdk后，可以参考这个文件设置环境然后编译：[build_in_container.sh](https://github.com/chraac/llama-cpp-qnn-builder/blob/main/docker/build_in_container.sh)