Comments on: Peterson’s Solution on Modern Multiprocessors http://www.alper.net/hardware/peterson%e2%80%99s-solution-on-modern-multiprocessors/ Eyup Alper Yoney Sun, 28 Aug 2011 09:14:25 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Alper http://www.alper.net/hardware/peterson%e2%80%99s-solution-on-modern-multiprocessors/comment-page-1/#comment-10 Alper Mon, 30 Aug 2010 09:42:46 +0000 http://www.alper.net/?p=498#comment-10 I tested with the volatile, the test results are similar. I disassembler the code before and after the volatile, the produced codes are identical. Actually, GCC doesn't perform any optimization like keeping the values in registers. The store instruction is executed. However, two threads enter the critical section at the same time because of the store buffer and/or instruction reordering. Regards, Alper I tested with the volatile, the test results are similar. I disassembler the code before and after the volatile, the produced codes are identical. Actually, GCC doesn’t perform any optimization like keeping the values in registers. The store instruction is executed. However, two threads enter the critical section at the same time because of the store buffer and/or instruction reordering.

Regards,
Alper

]]> By: Juergen Wuebbelmann http://www.alper.net/hardware/peterson%e2%80%99s-solution-on-modern-multiprocessors/comment-page-1/#comment-9 Juergen Wuebbelmann Thu, 26 Aug 2010 14:38:28 +0000 http://www.alper.net/?p=498#comment-9 Very interesting test and results. Just for my curiosity: have you ever tried to use the volatile attribute for the turn and interested variables? Regards Juergen Very interesting test and results. Just for my curiosity: have you ever tried to use the volatile attribute for the turn and interested variables?

Regards
Juergen

]]> By: Vitaliy Tsvayer http://www.alper.net/hardware/peterson%e2%80%99s-solution-on-modern-multiprocessors/comment-page-1/#comment-8 Vitaliy Tsvayer Sun, 11 Apr 2010 16:12:18 +0000 http://www.alper.net/?p=498#comment-8 Interesting topic. It reminded me of the "Double-Checked Locking" problem in Java related to Out-of-order writes in pre Java 5 memory model. http://www.ibm.com/developerworks/library/j-dcl.html Interesting topic. It reminded me of the “Double-Checked Locking” problem in Java related to Out-of-order writes in pre Java 5 memory model. http://www.ibm.com/developerworks/library/j-dcl.html

]]>