Skip to content

X
x393
Commit 23ea64ab authored by Andrey Filippov's avatar Andrey Filippov
Browse files
Options

Working on phase rotator for partial pixel shifts

parent 10056a60
Hide whitespace changes
Inline Side-by-side
Showing with 295 additions and 0 deletions
dsp/phase_rotator.v 0 → 100644
View file @ 23ea64ab
/*!
* <b>Module:</b> phase_rotator
* @file phase_rotator.v
* @date 2017-12-11
* @author eyesis
*
* @brief 2-d phase rotator in frequency domain (subpixel shift)
*
* @copyright Copyright (c) 2017 <set up in Preferences-Verilog/VHDL Editor-Templates> .
*
* <b>License </b>
*
* phase_rotator.v is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* phase_rotator.v is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/> .
*
* Additional permission under GNU GPL version 3 section 7:
* If you modify this Program, or any covered work, by linking or combining it
* with independent modules provided by the FPGA vendor only (this permission
* does not extend to any 3-rd party modules, "soft cores" or macros) under
* different license terms solely for the purpose of generating binary "bitstream"
* files and/or simulating the code, the copyright holders of this Program give
* you the right to distribute the covered work without those independent modules
* as long as the source code for them is available from the FPGA vendor free of
* charge, and there is no dependence on any encrypted modules for simulating of
* the combined code. This permission applies to you if the distributed code
* contains all the components and scripts required to completely simulate it
* with at least one of the Free Software programs.
*/
`timescale 1ns/1ps
module phase_rotator#(
parameter FD_WIDTH = 25, // input/output data width, signed
parameter SHIFT_WIDTH = 7, // x/y subpixel shift, signed -0.5<=shift<0.5
parameter DSP_B_WIDTH = 18, // signed, output from sin/cos ROM
parameter DSP_A_WIDTH = 25,
parameter DSP_P_WIDTH = 48
)(
input clk, //!< system clock, posedge
input rst, //!< sync reset
input start, //!< single-cycle start pulse that goes 1 cycle before first data
input signed [SHIFT_WIDTH-1:0] shift_h, //!< subpixel shift horizontal
input signed [SHIFT_WIDTH-1:0] shift_v, //!< subpixel shift vertical
// input data CC,CS,SC,SS in column scan order (matching DTT)
input signed [FD_WIDTH-1:0] fd_din, //!< frequency domain data in
output reg signed [FD_WIDTH-1:0] fd_out, //!< frequency domain data in
output reg pre_first_out, //!< 1 cycle before output data valid
output reg fd_dv //!< output data valid
);
// Generating cos_hor, sin_hor, cos_vert, sin_vert coefficients and cosine and sine signs for both horizontal and vertical
// passes. For continuing operation, the sequence of the coefficients should be as following (starting at the same time as CC0):
// CV62-SV62-CH00-SH00-CV63-SV63-CH01-SH01-CV0-SV0-..., where*V62, *V63 - data from the previous tile.
// Signed subpixel shifts in the range [-0.5,0.5) are converted to sign and [0,0.5] and both limits are fit to the same slot as
// the shift -0.5 : cos((i+0.5)/16*pi), -sin((i+0.5)/16*pi) for i=0..7 is symmetrical around the center (odd for sine, even - cosine)
// ROM input MSB - 0- cos, 1 - sin, 3 LSB s - index (0..7). Signs for cos and sin are passed to DSPs
// shift i ROM A[8:3] ROM A[2:0] sign cos sign sin
// 1000000 (-0.5) 000 0 000 0 1
// 001 0 001 0 1
// 010 0 010 0 1
// 011 0 011 0 1
// 100 0 011 1 1
// 101 0 010 1 1
// 110 0 001 1 1
// 111 0 000 1 1
//
// 1xxxxxx (<0) nnn -xxxxxx nnn 0 1
//
// 0000000 (==0) nnn 0 100 0 0
//
// 0xxxxxx (>0) nnn xxxxxx nnn 0 0
reg [5:0] start_d; // delayed versions of start (TODO: adjust length)
reg [7:0] cntr_h; // input sample counter
reg run_h;
wire [7:0] cntr_v; // delayed sample counter
wire run_v;
wire run_hv = run_h || run_v;
reg [2:0] hv_index; // horizontal/vertical index
reg [1:0] hv_phase; //
reg hv_sin; // 0 - cos, 1 - sin
reg [SHIFT_WIDTH-1:0] shift_hr;
reg [SHIFT_WIDTH-1:0] shift_v0;
reg [SHIFT_WIDTH-1:0] shift_vr;
reg [SHIFT_WIDTH-1:0] shift_hv; // combined horizonta and vertical shifts to match cntr_mux;
reg sign_cs; // sign for cos / sin, feed to DSP
reg [SHIFT_WIDTH-2:0] rom_a_shift; // ~shift absolute value
reg [2:0] rom_a_indx; // rom index (hor/vert)
reg rom_a_sin; // rom cos =0; sin = 1
wire [SHIFT_WIDTH+2:0] rom_a = {rom_a_sin,rom_a_shift,rom_a_indx};
wire shift_ends_0 = shift_hv[SHIFT_WIDTH-2:0] == 0;
reg [1:0] rom_re_regen;
wire signed [DSP_B_WIDTH-1:0] cos_sin_w;
always @ (posedge clk) begin
if (rst) start_d <= 0;
else start_d <= {start_d[4:0], start};
if (start) shift_hr <= shift_h;
if (start) shift_v0 <= shift_v;
if (start_d[5]) shift_vr <= shift_v0;
if (rst) run_h <= 0;
else if (start_d[0]) run_h <= 1;
else if (&cntr_h) run_h <= 0;
if (!run_h) cntr_h <= 0;
else cntr_h <= cntr_h + 1;
if (!run_hv) hv_phase <= 0;
else hv_phase <= hv_phase + 1;
// combine horizontal and vertical counters and shifts to feed to ROM
hv_index <= (run_v && cntr_v[1]) ? cntr_v[4:2] : cntr_h[7:5]; // input data "down first" (transposed)
hv_sin <= (run_v && cntr_v[1]) ? cntr_v[0] : cntr_h[0];
shift_hv <= (run_v && cntr_v[1]) ? shift_vr : shift_hr;
// convert index, shift to ROM address
rom_a_indx <= shift_ends_0 ? (shift_hv[SHIFT_WIDTH-1]?{1'b0,hv_index[2]?~hv_index[1:0]:hv_index[1:0]}:3'h4) : hv_index;
rom_a_shift <= shift_hv[SHIFT_WIDTH-1] ? -shift_hv[SHIFT_WIDTH-2:0] : shift_hv[SHIFT_WIDTH-2:0];
rom_a_sin <= hv_sin;
sign_cs <= shift_hv[SHIFT_WIDTH-1] & ( hv_sin | (shift_ends_0 & hv_index[2]));
rom_re_regen <= {rom_re_regen[0],run_hv};
end
dly_var #(
.WIDTH(9),
.DLY_WIDTH(4)
) dly_cntrv_i (
.clk (clk), // input
.rst (rst), // input
.dly (4'h3), // input[3:0]
.din ({run_h, cntr_h}), // input[0:0]
.dout ({run_v, cntr_v}) // output[0:0]
);
ram18tp_var_w_var_r #(
.REGISTERS_A(1),
.REGISTERS_B(1),
.LOG2WIDTH_A(4),
.LOG2WIDTH_B(4)
`ifdef PRELOAD_BRAMS
`include "mclt_fold_rom.vh" // TODO: put real!
`endif
) i_mclt_rot_rom (
.clk_a (clk), // input
.addr_a (rom_a), // input[9:0]
.en_a (rom_re_regen[0]), // input
.regen_a (rom_re_regen[1]), // input
.we_a (1'b0), // input
.data_out_a(cos_sin_w), // output[17:0]
.data_in_a (18'b0), // input[17:0]
.clk_b (1'b0), // input
.addr_b (10'b0), // input[9:0]
.en_b (1'b0), // input
.regen_b (1'b0), // input
.we_b (1'b0), // input
.data_out_b(), // output[17:0]
.data_in_b (18'b0) // input[17:0]
);
// horizontal shift stage
dsp_ma_preadd #(
.B_WIDTH(DSP_B_WIDTH),
.A_WIDTH(DSP_A_WIDTH),
.P_WIDTH(DSP_P_WIDTH),
.A_INPUT("DIRECT"),
.B_INPUT("DIRECT")
) dsp_1_i (
.clk(), // input
.rst(), // input
.bin(), // input[17:0] signed
.ceb1(), // input
.ceb2(), // input
.selb(), // input
.ain(), // input[24:0] signed
.cea1(), // input
.cea2(), // input
.din(), // input[24:0] signed
.ced(), // input
.cead(), // input
.sela(), // input
.en_a(), // input
.en_d(), // input
.sub_a(), // input
.neg_m(), // input
.accum(), // input
.pout() // output[47:0] signed
);
dsp_ma_preadd #(
.B_WIDTH(DSP_B_WIDTH),
.A_WIDTH(DSP_A_WIDTH),
.P_WIDTH(DSP_P_WIDTH),
.A_INPUT("DIRECT"),
.B_INPUT("CASCADE")
) dsp_2_i (
.clk(), // input
.rst(), // input
.bin(), // input[17:0] signed
.ceb1(), // input
.ceb2(), // input
.selb(), // input
.ain(), // input[24:0] signed
.cea1(), // input
.cea2(), // input
.din(), // input[24:0] signed
.ced(), // input
.cead(), // input
.sela(), // input
.en_a(), // input
.en_d(), // input
.sub_a(), // input
.neg_m(), // input
.accum(), // input
.pout() // output[47:0] signed
);
// vertical shift stage
dsp_ma_preadd #(
.B_WIDTH(DSP_B_WIDTH),
.A_WIDTH(DSP_A_WIDTH),
.P_WIDTH(DSP_P_WIDTH),
.A_INPUT("DIRECT"),
.B_INPUT("DIRECT")
) dsp_3_i (
.clk(), // input
.rst(), // input
.bin(), // input[17:0] signed
.ceb1(), // input
.ceb2(), // input
.selb(), // input
.ain(), // input[24:0] signed
.cea1(), // input
.cea2(), // input
.din(), // input[24:0] signed
.ced(), // input
.cead(), // input
.sela(), // input
.en_a(), // input
.en_d(), // input
.sub_a(), // input
.neg_m(), // input
.accum(), // input
.pout() // output[47:0] signed
);
dsp_ma_preadd #(
.B_WIDTH(DSP_B_WIDTH),
.A_WIDTH(DSP_A_WIDTH),
.P_WIDTH(DSP_P_WIDTH),
.A_INPUT("DIRECT"),
.B_INPUT("CASCADE")
) dsp_4_i (
.clk(), // input
.rst(), // input
.bin(), // input[17:0] signed
.ceb1(), // input
.ceb2(), // input
.selb(), // input
.ain(), // input[24:0] signed
.cea1(), // input
.cea2(), // input
.din(), // input[24:0] signed
.ced(), // input
.cead(), // input
.sela(), // input
.en_a(), // input
.en_d(), // input
.sub_a(), // input
.neg_m(), // input
.accum(), // input
.pout() // output[47:0] signed
);
endmodule
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment