| ??? 04/15/05 19:30 Read: times Msg Score: +1 +2 Informative -1 Answer is Wrong |
#91695 - Just? Responding to: ???'s previous message |
Prahlad J. Purohit said:
- Max time= 3uS Max. approx. 15 Instruction cycles [ @ 12MHz 2 clock core LPC932A1]
- Internal RAM approx. 400 bytes free. Flash approx. 2k free. - Resolution accuracy down to 1 bit. You might just be able to do it with a look up table within these parameters. A 10 bit number squared needs 20 bits for the answer and a 1024 entry table of 20 bit values obviously will not fit into 2K. However, if you look at a hex table of squares you will find the LS hex digit repeats every 8 values i.e is totaly dependent on the LS 3 bits i.e. 0000H ^2 = 0000H 0001H ^2 = 0001H 0002H ^2 = 0004H 0003H ^2 = 0009H 0004H ^2 = 0010H 0005H ^2 = 0019H 0006H ^2 = 0024H 0007H ^2 = 0021H ................. 0008H ^2 = 0040H 0009H ^2 = 0051H 000AH ^2 = 0064H etc The LS hex digit sequence 0,1,4,9,0,9,4,1 repeats for the entire table. So, the 16 MS bits of the answer could be held in a 1024 entry (2K byte) table) and the LS hex digit in an 8 byte table. I have not coded it but it looks like it should be doable in 15 cycles. Ian |
| Topic | Author | Date |
| Fast Square. | 01/01/70 00:00 | |
| Square dancing | 01/01/70 00:00 | |
| table lookup??? | 01/01/70 00:00 | |
| code & algorithm | 01/01/70 00:00 | |
| 16*16 bit is slower than what I want. | 01/01/70 00:00 | |
| How fast? | 01/01/70 00:00 | |
| Re: How Fast | 01/01/70 00:00 | |
| ... probably impossible in 15 cycles | 01/01/70 00:00 | |
| why cycles ? | 01/01/70 00:00 | |
| Re: Microseconds | 01/01/70 00:00 | |
| table lookup | 01/01/70 00:00 | |
| Natsemi appnote or CORDIC | 01/01/70 00:00 | |
| Natsemi link to appnote | 01/01/70 00:00 | |
| (a+b)^2=a^2+2*a*b+b^2 | 01/01/70 00:00 | |
| Thats Slow. | 01/01/70 00:00 | |
| faster need hardware | 01/01/70 00:00 | |
| How fast do you need? | 01/01/70 00:00 | |
| Re: How Fast. | 01/01/70 00:00 | |
| Just? | 01/01/70 00:00 | |
| Incorrect | 01/01/70 00:00 | |
| Correct? | 01/01/70 00:00 | |
| Whooooopa... Sorry. | 01/01/70 00:00 | |
| Thanks | 01/01/70 00:00 | |
| I tried... | 01/01/70 00:00 | |
| optimum? table driven | 01/01/70 00:00 | |
| Jan metod | 01/01/70 00:00 | |
| Hardware? | 01/01/70 00:00 | |
| CPLD? | 01/01/70 00:00 | |
| SILabs f12x does it in hardware | 01/01/70 00:00 | |
| Re: SiLabs F12x | 01/01/70 00:00 | |
| Price | 01/01/70 00:00 | |
| F12x price | 01/01/70 00:00 | |
| F12x MAC | 01/01/70 00:00 | |
| provided in the datasheet | 01/01/70 00:00 | |
| Just out of interest | 01/01/70 00:00 | |
| clarification | 01/01/70 00:00 | |
| CPLD? | 01/01/70 00:00 | |
| too expensive | 01/01/70 00:00 | |
| Absolute rubbish Oleg | 01/01/70 00:00 | |
| explain | 01/01/70 00:00 | |
| your right | 01/01/70 00:00 | |
| especially for those... | 01/01/70 00:00 | |
| I need to say this.... | 01/01/70 00:00 | |
| By the way..... | 01/01/70 00:00 | |
| just a demo | 01/01/70 00:00 | |
| Hang on. | 01/01/70 00:00 | |
| Oh bollocks | 01/01/70 00:00 | |
| Well oleg | 01/01/70 00:00 | |
| Please check my answer. | 01/01/70 00:00 | |
| Here you go | 01/01/70 00:00 | |
| You're having me on. | 01/01/70 00:00 | |
| Pascal? | 01/01/70 00:00 | |
| Pascal? | 01/01/70 00:00 | |
| Why ? | 01/01/70 00:00 | |
| It was changed because,,, | 01/01/70 00:00 | |
| Its because | 01/01/70 00:00 | |
| For Jez | 01/01/70 00:00 | |
| For Michael | 01/01/70 00:00 | |
| simulation | 01/01/70 00:00 | |
| Re: Fast Square | 01/01/70 00:00 | |
| Prahlad, waithing for a conclusion | 01/01/70 00:00 | |
| just an exercise... | 01/01/70 00:00 | |
| Tricky | 01/01/70 00:00 | |
| Jez asked his cat, I asked my sheep | 01/01/70 00:00 | |
| Conclusion. | 01/01/70 00:00 | |
SPI EEPROM | 01/01/70 00:00 |