Hi All
I have just completed a rather large overhaul of the CASPER FFT family.
Rigorous testing has yet to be performed but it works at the moment
and has been pushed to the ska-sa repo on github for early adopters/testers.
The main aim (and reason this email continues this thread) was to
Wow! Thanks, Andrew, that sounds like a lot of work! Do you have any
utilization comparisons of old vs new?
Dave
On Jul 16, 2013, at 6:22 AM, Andrew Martens wrote:
Hi All
I have just completed a rather large overhaul of the CASPER FFT family.
Rigorous testing has yet to be performed
Hi all,
The memory footprint of the fft_direct block could be reduced if it is
split into two blocks, one for the operator to manipulate the phase of the
FFT of each input and another block to calculate the true direct-form FFT,
i.e. not mapping together a larger FFT (see p. 615 of
Hi Dave
Do you have any utilization comparisons of old vs new?
Not yet, I hope to do a basic one soon though. Savings will depend on
use case. Also, you can optimise for resources in a few ways (DSPs vs
BRAMs when using fft_direct, logic vs BRAMS in the biplex stages and in
reset point
Hey
I haven't seen the Goertzel algorithm before, but it looks like a great
idea for this: we might be able to produce a coefficient DDS in just two
DSPs!
The Goertzel algorithm predates the FFT even and is used to calculate
sparse FFTs. It basically just calculates each bin individually.
Hi Dan
we used to use CORDIC for generating coefficients.
not sure how cordic comares to goertzel.
there are a few open source VHDL cordics.
Goertzel uses the coefficient factor of the previous calculation stage
to calculate the coefficient factor currently required i.e rotate the
previous
You guys probably appreciate this already, but although the coefficients in
the PFB FIR are generally symmetric around the center tap, the upper and
lower taps use these coefficients in reverse order from one another. In
order to take advantage of the symmetry, you'll have to use dual-port ROMs
hi aaron,
if you use xilinx brams for coefficients, they can be configured as dual
port memories,
so you can get the PFB reverse and forward coefficients both at the same
time,
from the same memory, almost for free, without any memory size penalty
over single port,
dan
On Mon, Jan 21, 2013
agreed. anybody already have, or want to develop, a coefficient
interpolator?
dan
On Mon, Jan 21, 2013 at 3:44 PM, Aaron Parsons apars...@astron.berkeley.edu
wrote:
Agreed.
The coefficient interpolator, however, could get substantial savings
beyond that, even, and could be applicable to
PS3. You could also have done the 2^16 FFT's coefficients as a narrow
cmult... you'd need to use a BRAM to store the 2^13 reset points, but it
would still indicate a reduction in memory use by a factor of 4 -- not
trivial by any means.
On Mon, Jan 21, 2013 at 9:39 PM, Ryan Monroe
10 matches
Mail list logo