Utilizing the large number of thread on recent Ryzen 3 Thread Rippper processor ?

2 次查看（过去 30 天）

Tuong Nguyen Minh 2020-7-23

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/569553-utilizing-the-large-number-of-thread-on-recent-ryzen-3-thread-rippper-processor

评论： Tuong Nguyen Minh 2020-7-23

When ever I used the parfor loop, Matlab automatically assign the number of worker to be equal to the number of physical CPU core. Is there anyway to let matlab 2020 know that each CPU core has two thread so that Matlab could use them properly ?

I am planning to purchase AMD Ryzen 3 3960X for my new work station.

2 个评论
显示无隐藏无

Rik 2020-7-23

编辑：Rik 2020-7-23

I don't work with parfor often, but can't you specify the number of workers explicitly yourself? Note that due to the extra load in your memory it might slow down the calculations to double the number of workers.

Tuong Nguyen Minh 2020-7-23

I am not sure if setting the nummber of worker = to the number of available thread would be beneficial or detrimental on Matlab 2020a or not. Older version of Matlab usually set the number of work = number of physical core.

请先登录，再进行评论。

请先登录，再回答此问题。