Pseudo-code:
if SVi[3:4] = 0b11 then
- # store subvl, pack and unpack in SVSTATE
+ # store pack and unpack in SVSTATE
SVSTATE[53] <- SVi[5]
SVSTATE[54] <- SVi[6]
RT <- [0]*62 || SVSTATE[53:54]
# FRC (coefficients)
SVSHAPE2[28:29] <- 0b10 # k schedule
# set schedule up for (i)DCT Inner butterfly
- # SVrm Mode 2 (Mode 6 for iDCT) is for pre-calculated coefficients,
# SVrm Mode 4 (Mode 12 for iDCT) is for on-the-fly (Vertical-First Mode)
- if ((SVrm = 0b0010) | (SVrm = 0b0100) |
- (SVrm = 0b1010) | (SVrm = 0b1100)) then
+ if ((SVrm = 0b0100) |
+ (SVrm = 0b1100)) then
# calculate O(N log2 N)
n <- [0] * 3
do while n < 5
SVSHAPE0[0:5] <- (0b0 || SVxd) # xdim
SVSHAPE0[12:17] <- (0b0 || SVzd) # zdim - "striding" (2D DCT)
mscale <- (0b0 || SVzd) + 1
- if (SVrm = 0b1010) | (SVrm = 0b1100) then
+ if (SVrm = 0b1100) then
SVSHAPE0[30:31] <- 0b11 # iDCT mode
SVSHAPE0[18:20] <- 0b011 # iDCT Inner Butterfly sub-mode
else
SVSHAPE0[30:31] <- 0b01 # DCT mode
SVSHAPE0[18:20] <- 0b001 # DCT Inner Butterfly sub-mode
SVSHAPE0[21:23] <- 0b001 # "inverse" on outer loop
- if (SVrm = 0b1100) | (SVrm = 0b0100) then
- SVSHAPE0[6:11] <- 0b000011 # (i)DCT Inner Butterfly mode 4
- else
- SVSHAPE0[6:11] <- 0b000001 # (i)DCT Inner Butterfly mode 2
+ SVSHAPE0[6:11] <- 0b000011 # (i)DCT Inner Butterfly mode 4
# copy
SVSHAPE1[0:31] <- SVSHAPE0[0:31]
SVSHAPE2[0:31] <- SVSHAPE0[0:31]
else
SVSHAPE0[30:31] <- 0b11 # DCT mode
SVSHAPE0[6:11] <- 0b000101 # DCT "half-swap" mode
- # set schedule up for parallel reduction
+ # set schedule up for parallel reduction or prefix-sum
if (SVrm = 0b0111) then
+ # is scan/prefix-sum
+ is_scan <- SVyd = 2
# calculate the total number of operations (brute-force)
vlen[0:6] <- [0] * 7
itercount[0:6] <- (0b00 || SVxd) + 0b0000001
- step[0:6] <- 0b0000001
- i[0:6] <- 0b0000000
- do while step <u itercount
- newstep <- step[1:6] || 0b0
- j[0:6] <- 0b0000000
- do while (j+step <u itercount)
- j <- j + newstep
- i <- i + 1
- step <- newstep
- # VL in Parallel-Reduce is the number of operations
- vlen[0:6] <- i
+ if is_scan then
+ # prefix sum algorithm with operations replaced with
+ # incrementing vlen
+ dist <- 1
+ vlen[0:6] <- 0
+ do while dist <u itercount
+ start <- dist * 2 - 1
+ step <- dist * 2
+ i <- start
+ do while i <u itercount
+ vlen[0:6] <- vlen[0:6] + 1
+ i <- i + step
+ dist <- dist * 2
+ dist <- dist / 2
+ do while dist != 0
+ i <- dist * 3 - 1
+ do while i <u itercount
+ vlen[0:6] <- vlen[0:6] + 1
+ i <- i + dist * 2
+ dist <- dist / 2
+ else
+ step <- 0b0000001
+ i <- 0b0000000
+ do while step <u itercount
+ newstep <- step[1:6] || 0b0
+ j[0:6] <- 0b0000000
+ do while (j+step <u itercount)
+ j <- j + newstep
+ i <- i + 1
+ step <- newstep
+ # VL in Parallel-Reduce is the number of operations
+ vlen[0:6] <- i
# set up template in SVSHAPE0, then copy to 1. only 2 needed
SVSHAPE0[0:5] <- (0b0 || SVxd) # xdim
SVSHAPE0[12:17] <- (0b0 || SVzd) # zdim - "striding" (2D DCT)
mscale <- (0b0 || SVzd) + 1
- SVSHAPE0[30:31] <- 0b10 # parallel reduce submode
+ SVSHAPE0[30:31] <- 0b10 # parallel reduce/prefix submode
# copy
SVSHAPE1[0:31] <- SVSHAPE0[0:31]
- # set up right operand (left operand 28:29 is zero)
- SVSHAPE1[28:29] <- 0b01 # right operand
+ # set up submodes: parallel or prefix
+ SVSHAPE0[28:29] <- 0b00 # left operand
+ SVSHAPE1[28:29] <- 0b01 # right operand
+ if is_scan then
+ SVSHAPE0[28:29] <- 0b10 # left operand
+ SVSHAPE1[28:29] <- 0b11 # right operand
# set VL, MVL and Vertical-First
m[0:12] <- vlen * mscale
maxvl[0:6] <- m[6:12]