You are viewing paulmck

Mar. 20th, 2012 (UTC)

I think it's worth pointing out that HTM can also be helpful in other scenarios, because it should give decent performance to other programming abstractions for synchronization.

For example, if the red-black tree is read-mostly, then even though RCU and hazard-pointers might give similar or somewhat better performance, it's likely still easier to wrap existing sequential code in a transaction than to restructure the code or the data to use a less transparent abstraction (e.g., RCU). (Of course, this only holds for programming-language transactions, not for a raw use of HTM.)

Therefore, one should consider all the aspects when thinking about the concurrent programming toolbox. (Paul, I know you are aware of this, but it can't hurt to make sure everyone is.)

Comment Form

No HTML allowed in subject

  
 
   
 

Notice! This user has turned on the option that logs your IP address when posting. 

(will be screened)