rknn-toolkit2

rknn-toolkit2 copied to clipboard

想测试一下ddr带宽利用率，请问rk-msch-probe-for-user这个工具要到哪里获取？

2

在一组写操作占用的时间内，有效时间长度为T1，无效时间长度是T2，DDR带宽利用率=T1/T2，看到官方平台有rk-msch-probe-for-user查看内存信息的工具，请问要到哪里下载呢？

SoulProficiency

CAN NOT install Toolkit2 1.4 on Ubuntu-20.4 with Python 3.8

5

There is no Python 3.6 available, therefore use 3.8. The numpy, bfloat16 are installed. ERROR message as following: (RKNN38) pj@B560M:~/rknn-toolkit2$ pip install bfloat16 Requirement already satisfied: bfloat16 in /home/pj/anaconda3/envs/RKNN38/lib/python3.8/site-packages (1.1)...

PJ5168

convert vit onnx model to rknn model, rknn_export .rknn model file size problem

3

we tried to convert vit onnx model to rknn model using rknn-toolkit2, vit: D RKNN: [10:14:17.424] Total Weight Memory Size: 19588032 (19 MB only ) D RKNN: [10:14:17.424] Total Internal...

ChandraDL

Custom model not working with RK3588S NPU

1

I am attempting to get a simple TFlite model working based off the examples and user guides that were provided in this repository and I am having a lot of...

venomboss985

rk3568推理报错: Exception: RKNN init failed. error code: RKNN_ERR_MALLOC_FAIL

2

smartwj

请问大神有没有人能出一个这个从.pt文件转换成rknn并部署到板端的教程呀？

hengaiw

Jayzhen2000

请问rv1106板子能支持batch_size>1输入吗

3

想在rv1106板子上进行batch_size=4推理，该怎么设置呢，谢谢！

blueskywwc

static int nms(int validCount, std::vector& outputLocations, std::vector classIds, std::vector& order, int filterId, float threshold) { for (int i = 0; i < validCount; ++i) { if (order[i] == -1 ||...

Eden417

Does RK3588 Support INT4 Model Inference on NPU?

1

Regarding the NPU on the RK3588, it is mentioned: "The built-in NPU supports INT4/INT8/INT16/FP16 hybrid operation and computing power is up to 6TOPs." However, the current demos are all INT8....

Leo5050xvjf