Comments (4)
诸如 mov v2.8h,v7.8h 改为 mov v2.16b,v7.16b,不要改成 mov v2.8b,v7.8b
from mnn.
把MNNAbsMaxFP16.S 文件中 15行到20行的内容删掉,再编译。即
from mnn.
谢大佬!但之后有另一个问题:
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S: Assembler messages:
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:66: Error: operand mismatch -- `mov v2.8h,v7.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:66: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:66: Info: mov v2.8b, v7.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:66: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:66: Info: mov v2.16b, v7.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:68: Error: operand mismatch -- `mov v1.8h,v6.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:68: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:68: Info: mov v1.8b, v6.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:68: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:68: Info: mov v1.16b, v6.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:104: Error: operand mismatch -- `mov v2.8h,v7.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:104: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:104: Info: mov v2.8b, v7.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:104: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:104: Info: mov v2.16b, v7.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:106: Error: operand mismatch -- `mov v1.8h,v6.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:106: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:106: Info: mov v1.8b, v6.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:106: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:106: Info: mov v1.16b, v6.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:158: Error: operand mismatch -- `mov v2.8h,v7.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:158: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:158: Info: mov v2.8b, v7.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:158: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:158: Info: mov v2.16b, v7.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:160: Error: operand mismatch -- `mov v1.8h,v6.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:160: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:160: Info: mov v1.8b, v6.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:160: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:160: Info: mov v1.16b, v6.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:196: Error: operand mismatch -- `mov v2.8h,v7.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:196: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:196: Info: mov v2.8b, v7.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:196: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:196: Info: mov v2.16b, v7.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:198: Error: operand mismatch -- `mov v1.8h,v6.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:198: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:198: Info: mov v1.8b, v6.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:198: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:198: Info: mov v1.16b, v6.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:366: Error: operand mismatch -- `mov v3.8h,v21.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:366: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:366: Info: mov v3.8b, v21.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:366: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:366: Info: mov v3.16b, v21.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:367: Error: operand mismatch -- `mov v4.8h,v21.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:367: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:367: Info: mov v4.8b, v21.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:367: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:367: Info: mov v4.16b, v21.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:417: Error: operand mismatch -- `mov v3.8h,v21.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:417: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:417: Info: mov v3.8b, v21.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:417: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:417: Info: mov v3.16b, v21.16b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:418: Error: operand mismatch -- `mov v4.8h,v21.8h'
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:418: Info: did you mean this?
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:418: Info: mov v4.8b, v21.8b
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:418: Info: other valid variant(s):
/home/ubuntu/mnn-llm/MNN/source/backend/arm82/asm/arm64/low_memory/MNNPackedMatMulFP16_int4.S:418: Info: mov v4.16b, v21.16b
from mnn.
跑得动了!太牛太快了🐄 好期待 opencl 的来临 (wangzhaode/mnn-llm#136 (comment)) !
from mnn.
Related Issues (20)
- Release Python 3.11 Wheels on PyPI
- mnn输入名字和onnx输入名字不一致 HOT 1
- MNN not working with MPS partitioned Nvidia GPU
- windows下编译cuda版本报错 HOT 1
- opencl,批量推理错误 HOT 1
- 请问,这个编译宏MNN_INTERNAL_ENABLED是弃用了吗??? HOT 2
- 请问,我想获取MNN推理时,每个算子的耗时,或者整个流程的耗时,就是做profililng,我该怎么做??? HOT 1
- qwen2-7b执行llm_export.py被killed,0.5b和1.5b可以正常导出 HOT 1
- Support for GatherV2 & While operators in CoreML Backend
- MNN 支持 INT16 量化吗? HOT 1
- 能否添加c_api? HOT 1
- 将onnx fp32的模型,转换为mnn fp16模型,部分层的weight为空 HOT 2
- rust wrapper over mnn / mnn 上的 rust 包装器 HOT 2
- "Undefined symbols for architecture arm64:" when link with `-framework MNN` HOT 4
- readmap崩溃 HOT 5
- grid_sample-5D 算子转mnn后精度不达预期 HOT 2
- yolov8seg的640x640输入的推理能正常跑了,但是onnx在服务器4个目标结果,在mnn只出来2个。什么原因? HOT 2
- [bug] MNN::Interpreter::createFromFile输入中文路径时乱码无法正常加载
- tflite模型转换时层 [ REVERSE_V2 ]不支持,请问有好的解决办法吗?
- llmexport 导出mnn失败 These Op Not Support: ONNX::FakeLinear HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from mnn.