On Wednesday, 20 August, 2014, at 02:40, dd <durga.d...@gmail.com> asked:
> Executing like query in a transaction. Query works with multiple tables >and table has 1 million records on desktop clients. CPU goes high when >transaction in progress. > Is there any way to control the CPU without adding sleep statements? As suggested, the only way to artificially impose limits on execution is through the OS dispatcher thread priority mechanisms. Execution will proceed at the maximum rate permitted by the first fully consumed resource. This resource may be CPU, I/O, memory bandwidth, video bandwidth, etc. In your case it would appear that the limit is set by CPU resources. By lowering the thread priority of the executing thread you can request that the OS prioritize CPU allocation to other competing threads. This will not reduce CPU usage -- it will only allow you to make this particular task have less priority than competing threads. Total CPU usage will remain at 100%. > Is there any alternative solution for like queries? (for ex: delete * >from emp where empname like "%a") Alternative solution for what? Looking through a table for all rows where the last character of a column is an 'a' character requires visiting every row in turn and testing the value of the last character and there is nothing you can do about this fact. You could, for example, avoid the full table scan by adding a field to the table called emanpme and put in it the reversed value of the empname field when you populate the table (and on every update of the empname field), and create an index on that column. You would then rephrase you query as "delete * from emp where emanpme like 'a%';" which would limit the visitation via a partial index scan. This is, however, a solution that you implement at design time. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
