-
Hi @RadxaYuntian. As per discussions yesterday I have installed the rockchip RKNPU2 and the toolkit2 on the latest debian provided by Radxa, both are version 1.5.2 so the latest versions however I've …
-
使用3588的板子,信息如下:
I RKNN: [02:17:34.734] RKNN Runtime Information, librknnrt version: 2.0.0b14 (2d891ba573@2024-05-31T11:36:07)
I RKNN: [02:17:34.734] RKNN Driver Information, version: 0.8.2
W RKNN: …
-
I am trying to port the project to run on an RV1126 device, but I encountered the following issues:
1. **Slow Model Loading with Full RKNPU Driver:** When using the full RKNPU driver, face recognit…
-
**1. 环境(environment)**
- Build OS and Version: ubuntu
- RunTime OS Version: Linux
- RunTime DEVICE: ARM
-
按照教程【https://github.com/Tencent/TNN/blob/master/doc/cn/faq.md#如何创建rknpu编译环境】成功编译好R…
-
我的板子是RK3399Pro debian系统
1、更新https://github.com/rockchip-linux/rknpu 替换rk3399pro代码仓库下的npu/external/rknpu,然后编译替换之前的npu固件
2、rk3399pro开发板上安装rknn_toolkit-1.7.1-cp37-cp37m-linux_aarch64.whl
3、adb push …
-
我在1.7.1版本的rknpu和rktoolkit下,用rv1126跑网盘提供的yolov5s_relu_u8预编译的模型,最终花费时间是80+ms,与公开的30+ms相差很大。个人觉得仅仅两个小版本的rknpu应该不至于造成这么大的时间差距,不知对于排查问题出在哪里是否有建议。
-
we are trying to run the simple resnet18 model on aarch64 platform and get the following error.
```D target set by user is: None
D Starting ntp or adb, target soc is RK3588, device id is: None
E RK…
-
hello,
I run TNN's model by this rknpu_ddk on rv1126. but the debug message is RKNPU Setup(175): create tensor fail.
I don't know why, and what cause it.
-
https://github.com/unifreq/linux-5.10.y-rk35xx/tree/main/drivers/video/rockchip/mpp
https://github.com/unifreq/linux-5.10.y-rk35xx/tree/main/drivers/video/rockchip/rga2
https://github.com/unifreq/li…
-
**1. 环境(environment)**
- Build OS and Version: Debian10
- RunTime DEVICE: RK3399Pro
**2. Github版本**
- branch:master
**3. 编译方式(compile method)**
参考以下网站,下载rknpu_ddk,加入-DTNN_RK_NPU_ENA…