multithreading - Reorderings in java memory model -


Read again, find out I do not understand a sentence:

We consider only the variables here that are readable and writeable as a nuclear unit - That is, there is no bit field, unrecognized access , or larger than the word sizes available on any platform. is.

Anyone can explain the above cases on which the words are bold. On some architectures, the processor can access multiple storage in one command:

  • BIT FIELD: Setting up some pieces of memory; It is usually implemented with different access to reading and writing, using read-write-write.
  • Unwritten Access: On some CPU architectures, the processor impedes the barriers to the addresses you can use, when you access multibeat values, especially if you exceed the word limitations. Most modern processors Does not have this problem because they can perform multiple readsets to cover the required address range; If you read from 0x01 to 0x05, then it is read 0x01-0x04 and a 0x05. If you have to write there, then it is complicated; The same areas will be read for the first time, then it will be written, 3 times as per the requirement of many memory operations (4 is a valuable value) compared to the coalition's address. (Keep in mind that the problem is more complex on the modern CPU, because you have a very deep memory hierarchy with the lack of different per-layer alignment).

    • In these cases, an operation will be required to exceed the many memory reach, thus Tehher is potential for running conditions, thus no nuclear guarantee.


    Comments

    Post a Comment

    Popular posts from this blog

    c# - Reactive Extensions ControlScheduler -

    Well, I'm using Reactive Extension Event Handler to handle my app events, I run the headlines on the Yuri thread. I also use ControlSubular for However, recently I am getting cross thread exception despite using control scheduler and I do not know what the problem is. Code: Observable.FromEventPattern & Lt; String & gt; (CC, "UiAlertMessage", New Control Session ()) Subscribe (_ = & gt; {Alert Control. (Show this language, title, _.EventArgs.UppercaseFirst ());}); The new control scheduler (this) is not to run the code on the UI control thread, so I do not get cross threading exceptions? you should do Observable.FromEventPattern & lt; String & gt; (CC, "UiAlertMessage"). AboveOption (this). Subscribe (_ = & gt; {Alert Control. (Show this language, Titles, _.EventArgs.UppercaseFirst ());}); This is the standard way of sending a scheduler related to specific controls. Passing control scheduler as you have got subsc...
    Read more

    Removing From ArrayList, In Loop Based On It's Size, But Breaking After Remove Still Gives OutOfBounds -

    OK, so I remove an orange from the array list, break it, but I still get the outbound I am a little confused, someone can help me, I have tried to separate the problem, but I still can not understand it. Here is an error: Exception threads "Thread-2" java.lang.IndexOutOfBoundsException: Index: 0, Size: 0 on java.util.ArrayList.rangeCheck ( GameFunctions.act at GameFunctions.act on (ArrayList.java:635) at java.util.ArrayList.get (ArrayList.java:411) GameFunctions.act (GameFunctions.java147) GamePanel $ 1.run (GamePanel.java:50) Here are two methods (if (Actor equivalents ("germ")) {for (int i = 0; i & lt; g.size (); i ++) {if (live .t.T. Examples of Spinumonia) {g.get (i) .move (); G.get (i) .testBounds (); If (Czechs (G, I)) {System.out.println ("broken"); break; }}}} And if (Actor equal ("whiteboard seals")) {for (int i = 0; i & lt; p.size (); i ++) {pgate (i) .setTarget ( Closest goal (G, P. Mill (i))); P.get (i) .move (); }...
    Read more