| ??? 02/14/01 10:11 Read: times |
#9300 - RE: use of multiply in large data arrays |
Peter,
Yes I agree with you that the MUL takes 4 cycles, my mistake was that I was thinking of 16 bits MUL this one takes more. But in my software I use an 8 bits mul. So this decreases the time consumed. Regarding the timing aspect, I want to have "As short as possible" software, at this moment I am writing a kernel which is based on routing and handling of a lot of messages. The messages are delivered from 3 sources on a maximum speed of 125 kbs. This means that have to (in worst case) handle(or save) a message in 0,3ms from each source. This results in 0,1ms per message avaiable. (remember worst case). This could happen in peek moments. Any decrement of the time consumed when I parse messages will eventually result in a faster (more reliable) system. Patrick |



