I've developed an application that frequently utilizes pread and pwrite for reading and writing to a network devices. The operation typically takes around 7 minutes to complete.

我已经开发了一个应用程序，该应用程序经常使用扩展和写入来读取和写入网络设备。手术通常需要大约7分钟才能完成。

However, when I execute two separate instances of this application concurrently but to two different devices, the completion time extends to 15 minutes. I'm seeking to understand the underlying mechanism behind pread64 and pwrite64 and identify the potential bottleneck.

然而，当我同时在两个不同的设备上执行这个应用程序的两个单独的实例时，完成时间延长到15分钟。我正在努力了解pread64和pWrite64背后的底层机制，并确定潜在的瓶颈。

It's important to note that these two processes are not accessing the same device, and there are no shared resources between them.

重要的是要注意，这两个进程访问的不是同一设备，并且它们之间没有共享资源。

I'm using Linux.

我使用的是Linux。