问题
I need to do math calculations for a text in a memo. [Size of the file: ~2mb]
A fitting example would be an encoded text that I need to decode.
I pass the memo text to a string in order to decode it. I guess it would be way faster to run my decode function using threads. But after some googling I didn't find a good example fitting my purpose.
Example function:
function entr_base_N(my_text:String):String;
var
ts_hamil64:Integer;
begin
For ts_hamil64 := 1 to Length(my_text) do
begin
Result:= Result + Chr(Ord(my_text[ts_hamil64])+10)
end;
end;
.....
.....
Memo1.Text:=entr_base_N(Memo1.Text)
I would like to break the work into small pieces, divide the job equally, lets say 3..8 threads and assign my decode function to those threads. Can you please guide me on this?
Current time to process the text file: ~35seconds. Thank you for your kind help.
回答1:
Threading is not the problem. Your function entr_base_N runs instantly. Try inside the debugger. You'll find it takes no time at all. Processing a 2MB string is trivial on a modern computer. That said, I would always recommend pre-allocating a return buffer when possible.
All the time is spent sending the resulting string back to the memo control. What is happening is that you are converting the #13 and #10 characters to #23 and #20. For whatever reason, the memo control does not like that. It seems to me that you are sending back a string with no line feeds at all and the memo's word wrap code performs badly.
A quick and dirty way to see that this is so is to set WordWrap to False on your memo.
The important lesson here is that you must correctly identify the bottleneck before attempting to optimise. It's an easy trap to fall into though, as my initial fumbled efforts to answer this question demonstrate.
回答2:
To make the assignment to Memo1.Text faster, you can use the following:
memo1.Perform(wm_setredraw, 0, 0);
try
memo1.Text:= entr_base_N(memo1.Text);
finally
memo1.Perform(wm_setredraw, 1, 0);
memo1.invalidate;
end;
来源：https://stackoverflow.com/questions/11352059/divide-work-assign-task-to-an-array-of-threads