Microchip Technology MA330019-2 Data Sheet
dsPIC33FJ32GP302/304, dsPIC33FJ64GPX02/X04, AND dsPIC33FJ128GPX02/X04
DS70292G-page 64
© 2007-2012 Microchip Technology Inc.
4.6.3
MODULO ADDRESSING
APPLICABILITY
APPLICABILITY
Modulo Addressing can be applied to the Effective
Address (EA) calculation associated with any W
register. Address boundaries check for addresses
equal to:
• The upper boundary addresses for incrementing
Address (EA) calculation associated with any W
register. Address boundaries check for addresses
equal to:
• The upper boundary addresses for incrementing
buffers
• The lower boundary addresses for decrementing
buffers
It is important to realize that the address boundaries
check for addresses less than or greater than the upper
(for incrementing buffers) and lower (for decrementing
buffers) boundary addresses (not just equal to).
Address changes can, therefore, jump beyond
boundaries and still be adjusted correctly.
check for addresses less than or greater than the upper
(for incrementing buffers) and lower (for decrementing
buffers) boundary addresses (not just equal to).
Address changes can, therefore, jump beyond
boundaries and still be adjusted correctly.
4.7
Bit-Reversed Addressing
Bit-Reversed Addressing mode is intended to simplify
data reordering for radix-2 FFT algorithms. It is
supported by the X AGU for data writes only.
The modifier, which can be a constant value or register
contents, is regarded as having its bit order reversed.
The address source and destination are kept in normal
order. Thus, the only operand requiring reversal is the
modifier.
data reordering for radix-2 FFT algorithms. It is
supported by the X AGU for data writes only.
The modifier, which can be a constant value or register
contents, is regarded as having its bit order reversed.
The address source and destination are kept in normal
order. Thus, the only operand requiring reversal is the
modifier.
4.7.1
BIT-REVERSED ADDRESSING
IMPLEMENTATION
IMPLEMENTATION
Bit-Reversed Addressing mode is enabled in any of
these situations:
• BWM bits (W register selection) in the MODCON
these situations:
• BWM bits (W register selection) in the MODCON
register are any value other than ‘15’ (the stack
cannot be accessed using Bit-Reversed
Addressing)
cannot be accessed using Bit-Reversed
Addressing)
• The BREN bit is set in the XBREV register
• The addressing mode used is Register Indirect
• The addressing mode used is Register Indirect
with Pre-Increment or Post-Increment
If the length of a bit-reversed buffer is M = 2
N
bytes,
the last ‘N’ bits of the data buffer start address must
be zeros.
be zeros.
XB<14:0> is the Bit-Reversed Address modifier, or
‘pivot point,’ which is typically a constant. In the case of
an FFT computation, its value is equal to half of the FFT
data buffer size.
‘pivot point,’ which is typically a constant. In the case of
an FFT computation, its value is equal to half of the FFT
data buffer size.
When enabled, Bit-Reversed Addressing is executed
only for Register Indirect with Pre-Increment or Post-
Increment Addressing and word-sized data writes. It
does not function for any other addressing mode or for
byte-sized data, and normal addresses are generated
instead. When Bit-Reversed Addressing is active, the
W Address Pointer is always added to the address
modifier (XB), and the offset associated with the
Register Indirect Addressing mode is ignored. In
addition, as word-sized data is a requirement, the LSb
of the EA is ignored (and always clear).
only for Register Indirect with Pre-Increment or Post-
Increment Addressing and word-sized data writes. It
does not function for any other addressing mode or for
byte-sized data, and normal addresses are generated
instead. When Bit-Reversed Addressing is active, the
W Address Pointer is always added to the address
modifier (XB), and the offset associated with the
Register Indirect Addressing mode is ignored. In
addition, as word-sized data is a requirement, the LSb
of the EA is ignored (and always clear).
If Bit-Reversed Addressing has already been enabled
by setting the BREN bit (XBREV<15>), a write to the
XBREV register should not be immediately followed by
an indirect read operation using the W register that has
been designated as the bit-reversed pointer.
by setting the BREN bit (XBREV<15>), a write to the
XBREV register should not be immediately followed by
an indirect read operation using the W register that has
been designated as the bit-reversed pointer.
Note:
The modulo corrected effective address is
written back to the register only when Pre-
Modify or Post-Modify Addressing mode is
used to compute the effective address.
When an address offset (such as [W7 +
W2]) is used, Modulo Address correction
is performed but the contents of the
register remain unchanged.
written back to the register only when Pre-
Modify or Post-Modify Addressing mode is
used to compute the effective address.
When an address offset (such as [W7 +
W2]) is used, Modulo Address correction
is performed but the contents of the
register remain unchanged.
Note:
All bit-reversed EA calculations assume
word-sized data (LSb of every EA is
always clear). The XB value is scaled
accordingly to generate compatible (byte)
addresses.
word-sized data (LSb of every EA is
always clear). The XB value is scaled
accordingly to generate compatible (byte)
addresses.
Note:
Modulo Addressing and Bit-Reversed
Addressing should not be enabled
together. If an application attempts to do so,
Bit-Reversed Addressing assumes priority
when active for the X WAGU and X WAGU,
Modulo Addressing is disabled. However,
Modulo Addressing continues to function in
the X RAGU.
Addressing should not be enabled
together. If an application attempts to do so,
Bit-Reversed Addressing assumes priority
when active for the X WAGU and X WAGU,
Modulo Addressing is disabled. However,
Modulo Addressing continues to function in
the X RAGU.