OR4E10 데이터 시트보기 (PDF) - Agere -> LSI Corporation

부품명

상세내역

제조사

OR4E10

ORCA® Series 4 Field-Programmable Gate Arrays

Agere -> LSI Corporation 

ORCA Series 4 FPGAs

Preliminary Data Sheet

December 2000

Programmable Logic Cells (continued)

Half-Logic Mode

Series 4 FPGAs are based upon a twin-quad architec-

ture in the PFUs. The bytewide nature (eight LUTs,

eight latches/FFs) may just as easily be viewed as two

nibbles (two sets of four LUTs, four latches/FFs). The

two nibbles of the PFU are organized so that any nib-

blewide feature (excluding some softwired LUT topolo-

gies) can be swapped with any other nibblewide feature

in another PFU. This provides for very flexible use of

logic and for extremely flexible routing. The half-logic

mode of the PFU takes advantage of the twin-quad

architecture and allows half of a PFU, K[7:4] and associ-

ated latches/FFs, to be used in logic mode while the

other half of the PFU, K[3:0] and associated

latches/FFs, is used in ripple mode. In half-logic mode,

the ninth FF may be used as a general-purpose FF or

as a register in the ripple mode carry chain.

operation (K3, F[3:0]), respectively. The ripple mode

diagram (Figure 9) shows full PFU ripple operation,

with half-logic ripple connections shown as dashed

lines.

The result output and ripple output are calculated by

using generate/propagate circuitry. In ripple mode, the

two operands are input into KZ[1] and KZ[0] of each

LUT. The result bits, one per LUT, are F[7:0]/F[3:0] (see

Figure 9). The ripple output from LUT K7/K3 can be

routed on dedicated carry circuitry into any of four adja-

cent PLCs, and it can be placed on the PFU COUT/

FCOUT outputs. This allows the PLCs to be cascaded

in the ripple mode so that nibblewide ripple functions

can be expanded easily to any length.

Result outputs and the carry-out may optionally be reg-

istered within the PFU. The capability to register the

ripple results, including the carry output, provides for

improved counter performance and simplified pipelin-

ing in arithmetic functions.

Ripple Mode

The PFU LUTs can be combined to do bytewide ripple

functions with high-speed carry logic. Each LUT has a

dedicated carry-out net to route the carry to/from any

adjacent LUT. Using the internal carry circuits, fast

arithmetic, counter, and comparison functions can be

implemented in one PFU. Similarly, each PFU has

carry-in (CIN, FCIN) and carry-out (COUT, FCOUT)

ports for fast-carry routing between adjacent PFUs.

The ripple mode is generally used in operations on two

data buses. A single PFU can support an 8-bit ripple

function. Data buses of 4 bits and less can use the

nibblewide ripple chain that is available in half-logic

mode. This nibblewide ripple chain is also useful for

longer ripple chains where the length modulo 8 is four

or less. For example, a 12-bit adder (12 modulo 8 = 4)

can be implemented in one PFU in ripple mode (8 bits)

and one PFU in half-logic mode (4 bits), freeing half of

a PFU for general logic mode functions.

Each LUT has two operands and a ripple (generally

carry) input, and provides a result and ripple (generally

carry) output. A single bit is rippled from the previous

LUT and is used as input into the current LUT. For LUT

K0, the ripple input is from the PFU CIN or FCIN port.

The CIN/FCIN data can come from either the fast-carry

routing (FCIN) or the PFU input (CIN), or it can be tied

to logic 1 or logic 0.

In the following discussions, the notations LUT K7/K3

and F[7:0]/F[3:0] are used to denote the LUT that pro-

vides the carry-out and the data outputs for full PFU

ripple operation (K7, F[7:0]) and half-logic ripple

K7[1]

K7[0]

K6[1]

K6[0]

K5[1]

K5[0]

K4[1]

K4[0]

K3[1]

K3[0]

K2[1]

K2[0]

K1[1]

K1[0]

K0[1]

K0[0]

IN/FCIN

C

DQ

C

K7

DQ

K6

DQ

K5

DQ

K4

DQ

K3

DQ

K2

DQ

K1

DQ

K0

DQ

Figure 9. Ripple Mode

REGOUT

FCOUT

COUT

F7

Q7

F6

Q6

F5

Q5

F4

Q4

F3

Q3

F2

Q2

F1

Q1

F0

Q0

5-5755(F).

16

Lucent Technologies Inc.

Share Link: