AHA4013B View Datasheet(PDF) - Unspecified
Part Name
Description
Manufacturer
AHA4013B Datasheet PDF : 28 Pages
| |||
Advanced Hardware Architectures, Inc.
sec. Note: Other frequency operations may be
derived similarly.
Output Buffer may be used to hold data from
one block while the Input Buffer is being filled with
the following block. Two rules listed in the caveats
are required to accomplish this. These are illustrated
in Figure 4.
Caveats:
1. Output of block i must start coincident with or
before the input of block i + 1.
2. Output of block i must be complete:
Processing Latency − N − 8 clocks
after the start of block i + l on the input.
Table 2: Burst Operation Using 50 MHz Clock and 1 Clock/Byte, Forward Order Output
CHECK BYTES ‘R’ = 20
CHECK BYTES ‘R’ = 2
BLOCK
LENGTHS ‘N’
MAXIMUM
LATENCY
(# of clocks)
25
155
50
230
100
380
150
530
200
680
255
845
MAXIMUM
LATENCY
(µsecs)
3.10
4.60
7.60
10.64
13.60
16.88
AVERAGE
RATE
(MBytes/sec)
8.06
10.88
13.13
14.13
14.75
15.13
MAXIMUM
LATENCY
(# of clocks)
137
212
362
512
662
827
MAXIMUM
LATENCY
(µsecs)
2.74
4.24
7.24
10.24
13.28
16.56
AVERAGE
RATE
(MBytes/sec)
9.13
11.79
13.75
14.63
15.13
15.38
Average Rate = -M-----a---x---i--m----u----m------L--N-a---t--e--n---c---y-----(--µ----s--e---c---)
2.9.2 CONTINUOUS OPERATION
Multiple blocks of data may be processed
through the device continuously as shown in Figure
4. Consecutive blocks are input into the device at
the rate of Ci clocks/byte. The output data stream
may or may not be continuous depending on
whether parity is being output (controlled by
NOPAR) and the choice of Co. Continuous
operation is described by several equations. The
following terms are used in these equations:
Ci - Input clock rate per byte: Ci ≥ 4 for
continuous operation
Co -
Cm -
N-
Output clock rate per byte: Co ≥ 2
Minimum of Ci and Co: If Ci < Co then
Cm = Ci else Cm = Co
Reed-Solomon block length
K - Reed-Solomon message length
R - Reed-Solomon parity length (R = N − K)
L - Output data length: If parity is being output
from the chip (NOPAR = 0), L = N; else if
the parity is not being output (NOPAR = 1)
L=K
Equation 2:
(N
–
1
)
×
Ci
≥
R
+
48
+
-N-----×-----C----i
Ci – 1
+
N------×-----C----m--
Cm – 1
B. Processing Latency
Processing latency is the time from the beginning
of a block on the input to the block being ready for
output. Maximum processing latency, expressed in
number of clocks, for continuous operation is:
Equation 3:
Latency
=
(N
–
1)
×
Ci
+
60
+
R
+
-N-----×-----C----m--
Cm – 1
C. Start and End of Output
Similar to the burst operation, Output Buffer
may be used to temporarily “hold” data from one
block while the Input Buffer is being filled.
However, these conditions must be satisfied: the
output of a data block must start after the latency
equation (Equation 3) is satisfied, but before the
maximum delay is reached. The maximum delay is:
A. Conditions for Continuous Operation
The allowable input and output data rates are
related to the Reed-Solomon block length by the
following two inequalities. Ci, Co, N and K must be
chosen so that these equations are satisfied.
Equation 1: R + 60 + -N-----×-----C----m--
------------------C----i--C----m-----–----1--- + N ≤ 367
Equation 4:
maximum_delay=
3
×
N
×
Ci
–
L
×
Co
–
-N-----×-----C----i
Ci – 1
i--f----m----a---x---i--m----u---m-----_---d---e---l-a---y-
Ci
≥
367,
then
maximum_delay
=
367
×Ci
i--f----m----a---x---i--m---C-u---im-----_---d---e---l-a---y- > 2 × N, then maximum_delay = 2 × N ×Ci
PS4013B-0600
Page 9 of 24
Share Link: 