Hi all,
I need to implement this code (below) with a embedded processor FPGA. Do you have any information to help me in the choice of this FPGA (performance) Altera // niosII, Xilinx // microblaze Lattice // mico32 For information, I'm trying this exemple in a Nios II (STR10 Kit) at
----------------------------------------- my code :
double a ; double b ; double c ;
...
for (j = 1; j
MicroBlaze has h/w floating point, but I think it may only be single precision.
Cheers, Jon
You can improve the speed of this code by a factor of 1 million:
result=1000*1000*a+1000*b+c
but I assume you want to use the intermediate "result" results, too :-) Then you should think about converting the doubles to fixed point numbers, if possible, and implementing it in VHDL, maybe triggered from a Nios program with a custom instruction or simply a port.
Yes, you can, but your reduction of the sum of sum it is not correct ... Try again :-)
-- mmihai
My bad, I've read result+=...
-- mmihai
Actually, even if it was +=, as long as the boundaries are fixed, the two loops can be reduced by pre-calculating the sums of j and k^2...
1) sum of j over 1..1000 = 500500Rewrite after eliminating the 'j' loop:
2) sum of k^2 over 1..1000 = 333833500
Rewrite after eliminating the 'k' loop:
There you are, the result of even += through the two loops without any loops.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required