Databases
StackOverflow2013
Postsdbo
PO
 

Note that there are some explanatory texts on larger screens.

plurals
  1. PO
    primarykey
    data
    text
    <p>There is no such thing as "lock-free threading" these days. It was an interesting playground for academia and the like, back in the end of the last century when computer hardware was slow and expensive. <a href="http://en.wikipedia.org/wiki/Dekker%27s_algorithm" rel="noreferrer">Dekker's algorithm</a> was always my favorite, modern hardware has put it out to pasture. It doesn't work anymore.</p> <p>Two developments have ended this: the growing disparity between the speed of RAM and the CPU. And the ability of chip manufacturers to put more than one CPU core on a chip.</p> <p>The RAM speed problem required the chip designers to put a buffer on the CPU chip. The buffer stores code and data, quickly accessible by the CPU core. And can be read and written from/to RAM at a much slower rate. This buffer is called the CPU cache, most CPUs have at least two of them. The 1st level cache is small and fast, the 2nd is big and slower. As long as the CPU can read data and instructions from the 1st level cache it will run fast. A cache miss is really expensive, it puts the CPU to sleep for as many as 10 cycles if the data is not in the 1st cache, as many as 200 cycles if it isn't in the 2nd cache and it needs to be read from RAM.</p> <p>Every CPU core has its own cache, they store their own "view" of RAM. When the CPU writes data the write is made to cache which is then, slowly, flushed to RAM. Inevitable, each core will now have a different view of the RAM contents. In other words, one CPU doesn't know what another CPU has written until that RAM write cycle completed <em>and</em> the CPU refreshes its own view.</p> <p>That is dramatically incompatible with threading. You always <em>really</em> care what the state of another thread is when you must read data that was written by another thread. To ensure this, you need to explicitly program a so-called memory barrier. It is a low-level CPU primitive that ensures that all CPU caches are in a consistent state and have an up to date view of RAM. All pending writes have to flushed to RAM, the caches then need to be refreshed.</p> <p>This is available in .NET, the Thread.MemoryBarrier() method implements one. Given that this is 90% of the job that the lock statement does (and 95+% of the execution time), you are simply not ahead by avoiding the tools that .NET gives you and trying to implement your own.</p>
    singulars
    1. This table or related slice is empty.
    1. POLock-free multi-threading is for real threading experts
      singulars
      1. PTQuestion
    1. PTAnswer
    1. This table or related slice is empty.
    1. USHans Passant
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    2. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    3. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTUpMod
    1. COI'm confused by your answer. Cache synchronization and contention of resources are only some of the penalties of OS provided locking. What about the cost of rescheduling? Not sure what your bottom line point is here
      singulars
      1. PO
      1. USSam Post
    2. CONot sure how rescheduling plays a role at all. The point of threading is to have a lock *not* block the thread 99.9% of the time. If it is a lot less then you don't get any value for the money. If it does block, there isn't anything you can do but wait. The lock statement already implements a spinwait.
      singulars
      1. PO
      1. USHans Passant
    3. CORight I get that, I guess my question is why does "cache synchronization is slow" === "lock free is no better than locking"? I am new to lock free and like the sound of it but you seem to think its not super useful.
      singulars
      1. PO
      1. USSam Post
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload