mirror of https://github.com/m-labs/artiq.git
fir: register multiplier output
This commit is contained in:
parent
ca636ef28a
commit
b7a308d33d
|
@ -47,7 +47,7 @@ class FIR(Module):
|
||||||
self.i = Signal((width, True))
|
self.i = Signal((width, True))
|
||||||
self.o = Signal((width, True))
|
self.o = Signal((width, True))
|
||||||
n = len(coefficients)
|
n = len(coefficients)
|
||||||
self.latency = (n + 1)//2 + 1
|
self.latency = (n + 1)//2 + 2
|
||||||
|
|
||||||
###
|
###
|
||||||
|
|
||||||
|
@ -55,18 +55,22 @@ class FIR(Module):
|
||||||
x = [Signal((width, True)) for _ in range(n)]
|
x = [Signal((width, True)) for _ in range(n)]
|
||||||
self.sync += [xi.eq(xj) for xi, xj in zip(x, [self.i] + x)]
|
self.sync += [xi.eq(xj) for xi, xj in zip(x, [self.i] + x)]
|
||||||
|
|
||||||
# Wire up output
|
if shift is None:
|
||||||
|
shift = width - 1
|
||||||
|
|
||||||
|
# Make products
|
||||||
o = []
|
o = []
|
||||||
for i, c in enumerate(coefficients):
|
for i, c in enumerate(coefficients):
|
||||||
# simplify for halfband and symmetric filters
|
# simplify for halfband and symmetric filters
|
||||||
if c == 0 or c in coefficients[i + 1:]:
|
if c == 0 or c in coefficients[i + 1:]:
|
||||||
continue
|
continue
|
||||||
o.append(c*reduce(add, [
|
m = Signal((width + shift, True))
|
||||||
|
self.sync += m.eq(c*reduce(add, [
|
||||||
xj for xj, cj in zip(x[::-1], coefficients) if cj == c
|
xj for xj, cj in zip(x[::-1], coefficients) if cj == c
|
||||||
]))
|
]))
|
||||||
|
o.append(m)
|
||||||
|
|
||||||
if shift is None:
|
# Make sum
|
||||||
shift = width - 1
|
|
||||||
self.sync += self.o.eq(reduce(add, o) >> shift)
|
self.sync += self.o.eq(reduce(add, o) >> shift)
|
||||||
|
|
||||||
|
|
||||||
|
@ -85,7 +89,7 @@ class ParallelFIR(Module):
|
||||||
# input and output: old to young, decreasing delay
|
# input and output: old to young, decreasing delay
|
||||||
self.i = [Signal((width, True)) for i in range(p)]
|
self.i = [Signal((width, True)) for i in range(p)]
|
||||||
self.o = [Signal((width, True)) for i in range(p)]
|
self.o = [Signal((width, True)) for i in range(p)]
|
||||||
self.latency = (n + 1)//2//parallelism + 2 # minus .5
|
self.latency = (n + 1)//2//parallelism + 3 # minus one sample
|
||||||
|
|
||||||
###
|
###
|
||||||
|
|
||||||
|
@ -96,16 +100,19 @@ class ParallelFIR(Module):
|
||||||
if shift is None:
|
if shift is None:
|
||||||
shift = width - 1
|
shift = width - 1
|
||||||
|
|
||||||
# wire up each output
|
|
||||||
for j in range(p):
|
for j in range(p):
|
||||||
|
# Make products
|
||||||
o = []
|
o = []
|
||||||
for i, c in enumerate(coefficients):
|
for i, c in enumerate(coefficients):
|
||||||
# simplify for halfband and symmetric filters
|
# simplify for halfband and symmetric filters
|
||||||
if c == 0 or c in coefficients[i + 1:]:
|
if c == 0 or c in coefficients[i + 1:]:
|
||||||
continue
|
continue
|
||||||
o.append(c*reduce(add, [
|
m = Signal((width + shift, True))
|
||||||
|
self.sync += m.eq(c*reduce(add, [
|
||||||
xj for xj, cj in zip(x[-1 - j::-1], coefficients) if cj == c
|
xj for xj, cj in zip(x[-1 - j::-1], coefficients) if cj == c
|
||||||
]))
|
]))
|
||||||
|
o.append(m)
|
||||||
|
# Make sum
|
||||||
self.sync += self.o[j].eq(reduce(add, o) >> shift)
|
self.sync += self.o[j].eq(reduce(add, o) >> shift)
|
||||||
|
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue