About training time

[cskunqin]

I train very slowly on 2080ti, I want to know your training time and how long it takes to train 100 iter

[llmpass]

I train very slowly on 2080ti, I want to know your training time and how long it takes to train 100 iter

On my side, I trained the model with Quadro RTX 6000, spending 48s per 100 iteration.

[YongpeiZhu]

I also train very slowly on V100. But I do not know where is the problem?

[cskunqin]

I also train very slowly on V100. But I do not know where is the problem?

Maybe you need to put the dataset in SSD

[YongpeiZhu]

I also train very slowly on V100. But I do not know where is the problem?

Maybe you need to put the dataset in SSD

How to do it?

[YongpeiZhu]

We have trained it in SSD. But what is the problem?

[cskunqin]

We have trained it in SSD. But what is the problem?

我这边的训练速度大概是一小时一万步，显卡是2080ti，操作系统是Windows，patchsize128*128，batchsize为16

[sui1999]

我们已经在 SSD 中对其进行了训练。但问题是什么？

大佬，请问你代码下载后是怎么处理的，为什么我在3060TI上，两分钟只能处理八张图片，patchsize256*256，bitchsize为8

[sui1999]

我们已经在 SSD 中对其进行了训练。但问题是什么？

我这边的训练速度大概是一小时一万步，显卡是2080ti，操作系统是Windows，patchsize128*128，batchsize为16

大佬，请问你代码下载后是怎么处理的，为什么我在3060TI上，两分钟只能处理八张图片，patchsize256*256，bitchsize为8,是因为没有提前处理保存lq部分的图片吗