https://github.com/tpoisonooo/chgemm

symmetric int8 gemm
https://github.com/tpoisonooo/chgemm

Last synced: 13 days ago
JSON representation

symmetric int8 gemm

• Host: GitHub
• URL: https://github.com/tpoisonooo/chgemm
• Owner: tpoisonooo
• Created: 2019-07-15T05:07:58.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2020-06-07T07:09:04.000Z (over 4 years ago)
• Last Synced: 2024-10-11T18:10:03.493Z (29 days ago)
• Language: Assembly
• Homepage:
• Size: 426 KB
• Stars: 66
• Watchers: 3
• Forks: 12
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• awesome-gemm - chgemm

README

        
# chgemm

---

![License](https://img.shields.io/badge/license-BSD--3--Clause-blue.svg) ![build](https://travis-ci.org/tpoisonooo/chgemm.svg?branch=master)

chgemm is an symmetric int8 project, which is slightly different from BLAS sgemm:

1. when you input an int8_t type of matrix [-127,+127], you will get an int32_t one. PS: pay attention to the overflow;

2. considering the application scene of the deeep learning, the packAB interface is open and can be adjusted;

3. the common design plan is `alpha*A*B+beta*C=C`, but mine is `C=A*B`, because they have no utility in deep learning inference;

4. row major;

5. the speed of this project is quicker than any other projects'.

chgemm 是一个 int8 gemm 工程，与 BLAS gemm 不完全相同：

1. 输入为 [-127, +127] 范围内的 int8_t 类型矩阵，输出为 int32_t 矩阵。需注意溢出；

2. 更多地为深度学习应用场景考虑，packAB 接口暴露出来可以调整；

3. 实现为 C = A * B。alpha 和 beta 在深度学习推理中无实用意义；

4. 行主序实现，放弃远古 fortran 时代的列主序；

5. 不低于其他项目的 symmint8 gemm 速度。

---

### test result

Compiled on RK3399 with `-O3` flag. The current peek can be 18.6 gflops, and the orange line is the single-core fp32 limit(14.3 gflops). 

### 速度

-O3 编译，目前在 rk3399 单核结果。目前极限可以到 18.6 gflops，橙线是 rk3399 单核 fp32 极限。 在 aws A72 单核测试约 23 gflops，是此实现方法的极限（发挥 100% 性能）。

![尺寸和gflops结果](0.png)

---

### 使用方式

1. 修改`makefile`中的`OLD`和`NEW`挑选不同实现方式。首次运行需要`OLD`和`NEW`是同一个

2. `make run` 即输出速度结果

3. `parameters.h`可修改测试参数

### 集成方式

参照 MMult_4x8_21.c 调用矩阵乘法，将代码嵌入到自己的项目中。可根据推理库的实现做相应修改。

### application with chgemm inside

chgemm is pleased to support [ncnn](https://github.com/tencent/ncnn) available, check [gemm_symm_int8.h](https://github.com/Tencent/ncnn/blob/master/src/layer/arm/gemm_symm_int8.h).