Timeline for THREADS: Kernel threads vs. Kernel-supported threads vs. User-level threads?
Current License: CC BY-SA 3.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 14, 2011 at 21:37 | vote | accept | Dark Templar | ||
| Sep 18, 2011 at 1:03 | comment | added | David Schwartz | By 1-to-1 I mean that every time you call 'pthread_create', a new context is created that is scheduled by the kernel. In a hybrid approach, the number of contexts scheduled by the kernel is less than the number of threads created -- the kernel schedules one of those contexts, and then threading library schedules one of the threads created by the application. | |
| Sep 17, 2011 at 23:18 | comment | added | Dark Templar | "or a hybrid approach (where the kernel schedules a kernel-level thread which then, in user-space, schedules a user-level thread)." Hey David,could you explain this again? I'm not sure what is going on... When you say 1-to-1 mapping are you referring to Thread Control Blocks when mentioning kernel-level theads? | |
| Sep 14, 2011 at 10:02 | comment | added | David Schwartz | On all modern (2.6 kernel or later) Linux systems that I know of, the system's default threading implementation is 1-to-1, each user-level thread has a corresponding KSE (kernel scheduling entity -- basically, a thread scheduled by the Linux kernel). The thread you create with 'pthread_create' will be scheduled directly by the kernel. Modern systems generally use NPTL, older systems used LinuxThreads -- they're both 1-to-1. | |
| Sep 14, 2011 at 9:57 | comment | added | yati sagade | what is the implementation like on Linux? | |
| Sep 14, 2011 at 9:53 | history | edited | David Schwartz | CC BY-SA 3.0 |
added 1 characters in body
|
| Sep 13, 2011 at 17:12 | history | answered | David Schwartz | CC BY-SA 3.0 |