Sign in

username:

password:



Not a member?

Search c54x



Search tips

Subscribe to c54x



c54x by Keywords

5409 | 5416 | AD5 | ADC | BIOS | Boot | Booting | Bootloader | C540 | C5402 | C5409 | C5416 | CCS | Codec | DMA | Dmad | DSK | DSKPlus | Dsplib | EVM | FFT | FIR | Flash | GPIO | HPI | Initialization | Interrupt | JTAG | LOG_printf | MCBSP | RFFT | RTDX | Sampling | STLM | UART | VC540

Ads

Discussion Groups

Technical discussions about the TI C54x DSPs (including the c5401, c5402, c5402a, c5404, c5407, c5409, c5409a, c5410, c5410a, c5416, c5420, c5421, c5441, c549, c5470 and c5471).

  

Post a new Thread

Re: Real FFT - Eris Ristemena - Oct 30 8:34:00 1999



anytime ... :)

-----Original Message-----
From: Andrew B <>
To: <>
Date: Saturday, October 30, 1999 3:29 PM
Subject: [c54x] Re: Real FFT >I have just found out the answer....
>
>Apparently,
>
>B = A - *AR2+<<16
>
>has the same meaning with
>
>B = *AR2+<<16 - A >thanks anyway................





(You need to be a member of c54x -- send a blank email to c54x-subscribe@yahoogroups.com )

Real FFT - Andrew B - Oct 30 23:29:00 1999

I have a problem with 'C542 DSKplus' Real FFT example especially
in phase 2 (N complex FFT) during stage 3 thru (LOGN-1) computation.

Here is the code:
-----------------
group: data(BRC) = MMR(AR6) ; # of butterflies in each
group
dblockrepeat(bend-1)
T = *AR4 ; T := WR
A = T * *AR3+ ; A := QR*WR || AR3->QI

********************************************************************************\
**************
*As what the comment says:
*A contains QR*WR
********************************************************************************\
**************

A = rnd(A + *AR5+0% * *AR3-), T = *AR5+0% ; A := QR*WR+QI*WI
|| AR3->QR
********************************************************************************\
**************
*A contains QR*WR+QI*WI
********************************************************************************\
**************

B = A + *AR2<<16 ; B :=
(QR*WR+QI*WI)+PR
********************************************************************************\
**************
*AR2 points to PR
*B contains (QR*WR+QI*WI) + PR
********************************************************************************\
**************

*AR2 = HI(B)<<ASM ;
PR':=((QR*WR+QI*WI)+PR)/2
********************************************************************************\
**************
*PR' = (QR*WR+QI*WI) + PR
********************************************************************************\
**************
|| B = A - *AR2+<<16 ; B :=
PR-(QR*WR+QI*WI) || AR2->PI
********************************************************************************\
**************
*Here is the problem:
*----------------------------------
*B contains (QR*WR+QI*WI)-PR
*where B should be PR-(QR*WR+QI*WI)
********************************************************************************\
**************

*AR3 = HI(B)<<ASM ; QR':=
(PR-(QR*WR+QI*WI))/2
|| A = T * *AR3+ ; A := QR*WI [T=WI]
|| AR3->QI
A = rnd(A - *AR3 * *AR4+0%), T = *AR3 ; A := QR*WI-QI*WR
B = A + *AR2<<16 ; B :=
(QR*WI-QI*WR)+PI

*AR3+ = HI(B)<<ASM ;
QI':=((QR*WI-QI*WR)+PI)/2 || AR3->QR
|| B = A - *AR2<<16 ; B :=
PI-(QR*WI-QI*WR)
********************************************************************************\
**************
*Here is another problem:
*----------------------------------
*B contains (QR*WI-QI*WR)-PI
*where B should be PI-(QR*WI-QI*WR)
********************************************************************************\
**************

T = *AR4
*AR2+ = HI(B)<<ASM ; PI':=
(PI-(QR*WI-QI*WR))/2 || AR2->PR
|| A = T * *AR3+ ; A := QR*WR
|| AR3->QI

bend: Can anyone help me?

sincerely,

Andrew

______________________________________________________






(You need to be a member of c54x -- send a blank email to c54x-subscribe@yahoogroups.com )

Re: Real FFT - Andrew B - Oct 31 1:26:00 1999

I have just found out the answer....

Apparently,

B = A - *AR2+<<16

has the same meaning with

B = *AR2+<<16 - A thanks anyway................

______________________________________________________





(You need to be a member of c54x -- send a blank email to c54x-subscribe@yahoogroups.com )