URL
https://opencores.org/ocsvn/rtf_sprite_controller/rtf_sprite_controller/trunk
Subversion Repositories rtf_sprite_controller
[/] [rtf_sprite_controller/] [trunk/] [rtl/] [verilog/] [rfSpriteController.sv] - Rev 9
Compare with Previous | Blame | View Log
`timescale 1ns / 1ps
// ============================================================================
// __
// \\__/ o\ (C) 2005-2022 Robert Finch, Waterloo
// \ __ / All rights reserved.
// \/_// robfinch<remove>@finitron.ca
// ||
//
// rfSpriteController.sv
// sprite / hardware cursor controller
//
// BSD 3-Clause License
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are met:
//
// 1. Redistributions of source code must retain the above copyright notice, this
// list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other materials provided with the distribution.
//
// 3. Neither the name of the copyright holder nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
// DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
// FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
// DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
// SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
// CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
// OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
//
// Sprite Controller
//
// FEATURES
// - parameterized number of sprites 1,2,4,6,8,14,16 or 32
// - sprite image cache buffers
// - each image cache is capable of holding multiple
// sprite images
// - an embedded DMA controller is used for sprite reload
// - programmable image offset within cache
// - programmable sprite width,height, and pixel size
// - sprite width and height may vary from 1 to 256 as long
// as the product doesn't exceed 4096.
// - pixels may be programmed to be 1,2,3 or 4 video clocks
// both height and width are programmable
// - programmable sprite position
// - programmable 8, 16 or 32 bits for color
// eg 32k color + 1 bit alpha blending indicator (1,5,5,5)
// - fixed display and DMA priority
// sprite 0 highest, sprite 31 lowest
// - graphics plane control
//
// This core requires an external timing generator to
// provide horizontal and vertical sync signals, but
// otherwise can be used as a display controller on it's
// own. However, normally this core would be embedded
// within another core such as a VGA controller. Sprite
// positions are referenced to the rising edge of the
// vertical and horizontal sync pulses.
// The core includes an embedded dual port RAM to hold the
// sprite images. The image RAM is updated using a built in DMA
// controller. The DMA controller uses 32 bit accesses to fill
// the sprite buffers. The circuit features an automatic bus
// transaction timeout; if the system bus hasn't responded
// within 20 clock cycles, the DMA controller moves onto the
// next address.
// The controller uses a ram underlay to cache the values
// of the registers. This is a lot cheaper resource wise than
// using a 32 to 1 multiplexor (well at least for an FPGA).
//
// All registers are 32 bits wide
//
// These registers repeat in incrementing block of four registers
// and pertain to each sprite
// 00: - position register
// HPOS [11: 0] horizontal position (hctr value)
// VPOS [27:16] vertical position (vctr value)
//
// 04: SZ - size register
// bits
// [ 7: 0] width of sprite in pixels - 1
// [15: 8] height of sprite in pixels -1
// [19:16] size of horizontal pixels - 1 in clock cycles
// [23:20] size of vertical pixels in scan-lines - 1
// * the product of width * height cannot exceed 2048 !
// if it does, the display will begin repeating
// [27:24] output plane
// [31:30] color depth 01=RGB332,10=RGB555+A,11=RGB888+A
//
// 08: ADR [31:12] 20 bits sprite image address bits
// This registers contain the high order address bits of the
// location of the sprite image in system memory.
// The DMA controller will assign the low order 12 bits
// during DMA.
// [11:0] image offset bits [11:0]
// offset of the sprite image within the sprite image cache
// typically zero
//
// 0C: TC [23:0] transparent color
// This register identifies which color of the sprite
// is transparent
//
//
//
// 0C-1FC: registers reserved for up to thirty-one other sprites
//
// 200: DMA burst reg sprite 0
// [8:0] burst start
// [24:16] burst end
// ...
// 27C: DMA burst reg sprite 31
//
// 280: [ 7: 0] Frame size, multiples of 16 pixels
// [15: 8] Number of frames of animation
// [25:16] Animation Rate
// [29:26] Frame size, LSBs 0 to 3
// [30] Auto repeat
// [31] Enable animation
// ...
// 2FC: Animation register sprite #31
//
// Global status and control
// 3C0: EN [31:0] sprite enable register
// 3C4: IE [31:0] sprite interrupt enable / status
// 3C8: SCOL [31:0] sprite-sprite collision register
// 3CC: BCOL [31:0] sprite-background collision register
// 3D0: DT [31:0] sprite DMA trigger on
// 3D4: DT [31:0] sprite DMA trigger off
// 3D8: VDT [31:0] sprite vertical sync DMA trigger
// 3EC: BC [29:0] background color
// 3FC: ADDR [31:0] sprite DMA address bits [63:32]
//
// 7200
//=============================================================================
import wishbone_pkg::*;
module rfSpriteController(
// Bus Slave interface
//------------------------------
// Slave signals
input rst_i, // reset
input s_clk_i, // clock
input s_cs_i,
input wb_write_request32_t wbs_req,
output wb_read_response32_t wbs_resp,
//------------------------------
// Bus Master Signals
input m_clk_i, // clock
output wb_write_request128_t wbm_req,
input wb_read_response128_t wbm_resp,
output [4:0] m_spriteno_o,
//--------------------------
input dot_clk_i, // video dot clock
input hsync_i, // horizontal sync pulse
input vsync_i, // vertical sync pulse
input blank_i, // blanking signal
input [39:0] zrgb_i, // input pixel stream
output [39:0] zrgb_o, // output pixel stream 12-12-12-4
output irq, // interrupt request
input test,
input xonoff_i
);
reg m_soc_o;
wire vclk = dot_clk_i;
wire hSync = hsync_i;
wire vSync = vsync_i;
reg [39:0] zrgb_o;
//--------------------------------------------------------------------
// Core Parameters
//--------------------------------------------------------------------
parameter pnSpr = 32; // number of sprites
parameter phBits = 12; // number of bits in horizontal timing counter
parameter pvBits = 12; // number of bits in vertical timing counter
localparam pnSprm = pnSpr-1;
//--------------------------------------------------------------------
// Variable Declarations
//--------------------------------------------------------------------
reg ce; // controller enable
wb_write_request32_t wb_reqs; // synchronized request
wire [4:0] sprN = wb_reqs.adr[8:4];
reg [phBits-1:0] hctr; // horizontal reference counter (counts dots since hSync)
reg [pvBits-1:0] vctr; // vertical reference counter (counts scanlines since vSync)
reg sprSprIRQ;
reg sprBkIRQ;
reg [31:0] out; // sprite output
reg outact; // sprite output is active
reg [3:0] outplane;
reg [pnSprm:0] bkCollision; // sprite-background collision
reg [29:0] bgTc; // background transparent color
reg [29:0] bkColor; // background color
reg [pnSprm:0] sprWe; // block ram write enable for image cache update
reg [pnSprm:0] sprRe; // block ram read enable for image cache update
// Global control registers
reg [31:0] sprEn; // enable sprite
reg [pnSprm:0] sprCollision; // sprite-sprite collision
reg sprSprIe; // sprite-sprite interrupt enable
reg sprBkIe; // sprite-background interrupt enable
reg sprSprIRQPending; // sprite-sprite collision interrupt pending
reg sprBkIRQPending; // sprite-background collision interrupt pending
reg sprSprIRQPending1; // sprite-sprite collision interrupt pending
reg sprBkIRQPending1; // sprite-background collision interrupt pending
reg sprSprIRQ1; // vclk domain regs
reg sprBkIRQ1;
// Sprite control registers
reg [31:0] sprSprCollision;
reg [pnSprm:0] sprSprCollision1;
reg [31:0] sprBkCollision;
reg [pnSprm:0] sprBkCollision1;
reg [23:0] sprTc [pnSprm:0]; // sprite transparent color code
// How big the pixels are:
// 1 to 16 video clocks
reg [3:0] hSprRes [pnSprm:0]; // sprite horizontal resolution
reg [3:0] vSprRes [pnSprm:0]; // sprite vertical resolution
reg [7:0] sprWidth [pnSprm:0]; // number of pixels in X direction
reg [7:0] sprHeight [pnSprm:0]; // number of vertical pixels
reg [3:0] sprPlane [pnSprm:0]; // output plane sprite is in
reg [1:0] sprColorDepth [pnSprm:0];
reg [1:0] colorBits;
// Sprite DMA control
reg [8:0] sprBurstStart [pnSprm:0];
reg [8:0] sprBurstEnd [pnSprm:0];
reg [31:0] vSyncT; // DMA on vSync
// display and timing signals
reg [31:0] hSprReset; // horizontal reset
reg [31:0] vSprReset; // vertical reset
reg [31:0] hSprDe; // sprite horizontal display enable
reg [31:0] vSprDe; // sprite vertical display enable
reg [31:0] sprDe; // display enable
reg [phBits-1:0] hSprPos [pnSprm:0]; // sprite horizontal position
reg [pvBits-1:0] vSprPos [pnSprm:0]; // sprite vertical position
reg [7:0] hSprCnt [pnSprm:0]; // sprite horizontal display counter
reg [7:0] vSprCnt [pnSprm:0]; // vertical display counter
reg [11:0] sprImageOffs [pnSprm:0]; // offset within sprite memory
reg [12:0] sprAddr [pnSprm:0]; // index into sprite memory (pixel number)
reg [9:0] sprAddr1 [pnSprm:0]; // index into sprite memory
reg [2:0] sprAddr2 [pnSprm:0]; // index into sprite memory
reg [2:0] sprAddr3 [pnSprm:0]; // index into sprite memory
reg [2:0] sprAddr4 [pnSprm:0]; // index into sprite memory
reg [2:0] sprAddr5 [pnSprm:0]; // index into sprite memory
reg [11:0] sprAddrB [pnSprm:0]; // backup address cache for rescan
wire [31:0] sprOut4 [pnSprm:0]; // sprite image data output
reg [31:0] sprOut [pnSprm:0]; // sprite image data output
reg [31:0] sprOut5 [pnSprm:0]; // sprite image data output
wire [5:0] actcnt; // count of sprites active at a given pixel location of the screen
// Animation
reg [11:0] sprFrameSize [pnSprm:0];
reg [7:0] sprFrames [pnSprm:0];
reg [7:0] sprCurFrame [pnSprm:0];
reg [9:0] sprRate [pnSprm:0];
reg [9:0] sprCurRateCount [pnSprm:0];
reg [pnSprm:0] sprEnableAnimation;
reg [pnSprm:0] sprAutoRepeat;
reg [11:0] sprFrameProd [pnSprm:0];
// DMA access
reg [31:12] sprSysAddr [pnSprm:0]; // system memory address of sprite image (low bits)
reg [4:0] dmaOwner; // which sprite has the DMA channel
reg [31:0] sprDt; // DMA trigger register
reg dmaActive; // this flag indicates that a block DMA transfer is active
genvar g;
//--------------------------------------------------------------------
// DMA control / bus interfacing
//--------------------------------------------------------------------
reg cs_regs;
always_ff @(posedge s_clk_i)
cs_regs <= wbs_req.cyc & wbs_req.stb & s_cs_i;
always_ff @(posedge s_clk_i)
wb_reqs <= wbs_req;
wire s_ack_o;
ack_gen #(
.READ_STAGES(3),
.WRITE_STAGES(1),
.REGISTER_OUTPUT(1)
)
uag1 (
.clk_i(s_clk_i),
.ce_i(1'b1),
.i(cs_regs),
.we_i(cs_regs & wb_reqs.we),
.o(s_ack_o),
.rid_i(0),
.wid_i(0),
.rid_o(),
.wid_o()
);
always_comb
begin
wbs_resp.ack = s_ack_o & wbs_req.cyc & wbs_req.stb;
wbs_resp.next = s_ack_o & wbs_req.cyc & wbs_req.stb;
end
assign irq = sprSprIRQ|sprBkIRQ;
//--------------------------------------------------------------------
// DMA control / bus interfacing
//--------------------------------------------------------------------
reg [5:0] dmaStart;
reg [8:0] cob; // count of burst cycles
assign wbm_req.bte = LINEAR;
assign wbm_req.cti = CLASSIC;
assign wbm_req.blen = 6'd63;
assign wbm_req.stb = wbm_req.cyc;
assign wbm_req.sel = 16'hFFFF;
assign wbm_req.cid = 4'd5;
assign m_spriteno_o = dmaOwner;
reg [2:0] mstate;
parameter IDLE = 3'd0;
parameter ACTIVE = 3'd1;
parameter ACK = 3'd2;
parameter NACK = 3'd3;
wire pe_m_ack_i;
edge_det ued2 (.rst(rst_i), .clk(m_clk_i), .ce(1'b1), .i(wbm_resp.ack), .pe(pe_m_ack_i), .ne(), .ee());
reg [11:0] tocnt;
always_ff @(posedge m_clk_i)
if (rst_i)
tocnt <= 'd0;
else begin
if (wbm_req.cyc)
tocnt <= tocnt + 2'd1;
else
tocnt <= 'd0;
end
always_ff @(posedge m_clk_i)
if (rst_i)
mstate <= IDLE;
else begin
case(mstate)
IDLE:
if (|sprDt & ce)
mstate <= ACTIVE;
ACTIVE:
mstate <= ACK;
ACK:
if (wbm_resp.ack | wbm_resp.err | tocnt[10])
mstate <= NACK;
NACK:
if (~(wbm_resp.ack|wbm_resp.err))
mstate <= cob==sprBurstEnd[dmaOwner] ? IDLE : ACTIVE;
default:
mstate <= IDLE;
endcase
end
integer n30;
always_ff @(posedge m_clk_i)
begin
case(mstate)
IDLE:
begin
dmaOwner <= 5'd0;
for (n30 = pnSprm; n30 >= 0; n30 = n30 - 1)
if (sprDt[n30])
dmaOwner <= n30;
end
default: ;
endcase
end
always_ff @(posedge m_clk_i)
if (rst_i)
dmaStart <= 6'b0;
else begin
dmaStart <= {dmaStart[4:0],1'b0};
case(mstate)
IDLE:
if (|sprDt & ce)
dmaStart <= 6'h3F;
default: ;
endcase
end
integer n32;
always_ff @(posedge m_clk_i)
begin
case(mstate)
IDLE:
for (n32 = pnSprm; n32 >= 0; n32 = n32 - 1)
if (sprDt[n32] & ce)
cob <= sprBurstStart[n32];
ACTIVE:
cob <= cob + 2'd2;
default: ;
endcase
end
always_ff @(posedge m_clk_i)
if (rst_i)
wb_m_nack();
else begin
case(mstate)
IDLE:
wb_m_nack();
ACTIVE:
begin
wbm_req.cyc <= 1'b1;
wbm_req.adr <= {sprSysAddr[dmaOwner],cob[8:1],4'h0};
end
ACK:
if (wbm_resp.ack|wbm_resp.err|tocnt[10])
wb_m_nack();
endcase
end
task wb_m_nack;
begin
wbm_req.cyc <= 1'b0;
end
endtask
// generate a write enable strobe for the sprite image memory
integer n1;
reg [8:0] m_adr_or; // 64-bit value address
reg [127:0] m_dat_ir;
reg ack1;
always_ff @(posedge m_clk_i)
for (n1 = 0; n1 < pnSpr; n1 = n1 + 1)
sprWe[n1] <= (dmaOwner==n1 && (pe_m_ack_i||ack1));
always_ff @(posedge m_clk_i)
ack1 <= pe_m_ack_i;
always_ff @(posedge m_clk_i)
if (pe_m_ack_i|ack1)
m_adr_or <= {wbm_req.adr[11:4],ack1};
always_ff @(posedge m_clk_i)
if (pe_m_ack_i) begin
if (test)
m_dat_ir <= {8{1'b0,dmaOwner,10'b0}};
else
m_dat_ir <= {wbm_resp.dat[63:0],wbm_resp.dat[127:64]};
end
else if (ack1)
m_dat_ir <= {64'd0,m_dat_ir[127:64]};
//--------------------------------------------------------------------
//--------------------------------------------------------------------
reg [31:0] reg_shadow [0:255];
reg [7:0] radr;
always_ff @(posedge s_clk_i)
begin
if (cs_regs & wb_reqs.we & wb_reqs.sel[0]) reg_shadow[wb_reqs.adr[9:2]][7:0] <= wb_reqs.dat[7:0];
if (cs_regs & wb_reqs.we & wb_reqs.sel[1]) reg_shadow[wb_reqs.adr[9:2]][15:8] <= wb_reqs.dat[15:8];
if (cs_regs & wb_reqs.we & wb_reqs.sel[2]) reg_shadow[wb_reqs.adr[9:2]][23:16] <= wb_reqs.dat[23:16];
if (cs_regs & wb_reqs.we & wb_reqs.sel[3]) reg_shadow[wb_reqs.adr[9:2]][31:24] <= wb_reqs.dat[31:24];
end
always @(posedge s_clk_i)
radr <= wb_reqs.adr[9:2];
wire [31:0] reg_shadow_o = reg_shadow[radr];
// register/sprite memory output mux
always_ff @(posedge s_clk_i)
if (cs_regs)
case (wb_reqs.adr[9:2]) // synopsys full_case parallel_case
8'b11110000: wbs_resp.dat <= sprEn;
8'b11110001: wbs_resp.dat <= {sprBkIRQPending|sprSprIRQPending,5'b0,sprBkIRQPending,sprSprIRQPending,6'b0,sprBkIe,sprSprIe};
8'b11110010: wbs_resp.dat <= sprSprCollision;
8'b11110011: wbs_resp.dat <= sprBkCollision;
8'b11110100: wbs_resp.dat <= sprDt;
default: wbs_resp.dat <= reg_shadow_o;
endcase
else
wbs_resp.dat <= 32'h0;
// vclk -> clk_i
always @(posedge s_clk_i)
begin
sprSprIRQ <= sprSprIRQ1;
sprBkIRQ <= sprBkIRQ1;
sprSprIRQPending <= sprSprIRQPending1;
sprBkIRQPending <= sprBkIRQPending1;
sprSprCollision <= sprSprCollision1;
sprBkCollision <= sprBkCollision1;
end
// register updates
// on the clk_i domain
reg vSync1;
integer n33;
always_ff @(posedge s_clk_i)
if (rst_i) begin
vSyncT <= 32'hFFFFFFFF;//FFFFFFFF;
sprEn <= 32'hFFFFFFFF;
sprDt <= 0;
for (n33 = 0; n33 < pnSpr; n33 = n33 + 1) begin
sprSysAddr[n33] <= 20'b0000_0000_0011_0000_0000 + n33; //0030_0000
end
sprSprIe <= 0;
sprBkIe <= 0;
// Set reasonable starting positions on the screen
// so that the sprites might be visible for testing
for (n33 = 0; n33 < pnSpr; n33 = n33 + 1) begin
hSprPos[n33] <= 200 + (n33 & 7) * 70;
vSprPos[n33] <= 100 + (n33 >> 3) * 100;
sprTc[n33] <= 24'h7FFF; // White 16 bpp
sprWidth[n33] <= 8'd24; // 16x16 sprites
sprHeight[n33] <= 8'd21;
hSprRes[n33] <= 2'd2; // our standard display
vSprRes[n33] <= 2'd2;
sprImageOffs[n33] <= 12'h0;
sprPlane[n33] <= 4'h7;//n[3:0];
sprBurstStart[n33] <= 9'h000;
sprBurstEnd[n33] <= 9'h1FE;
sprColorDepth[n33] <= 2'b10;
if (n33 >= 5'd29) begin
sprFrameSize[n33] <= 12'd1936;
sprFrames[n33] <= 8'd0;
end
else begin
sprFrameSize[n33] <= 12'd504;
sprFrames[n33] <= 8'd3;
end
sprRate[n33] <= 12'd10;
sprEnableAnimation[n33] <= n33 < 5'd29;
sprAutoRepeat[n33] <= 1'b1;
end
hSprPos[0] <= 210;
vSprPos[0] <= 72;
bgTc <= 24'h08_08_08;
bkColor <= 24'hFF_FF_60;
end
else begin
ce <= xonoff_i;
vSync1 <= vSync;
if (vSync & ~vSync1)
sprDt <= sprDt | vSyncT;
// clear DMA trigger bit once DMA is recognized
if (dmaStart[5])
sprDt[dmaOwner] <= 1'b0;
// Disable animation after frame count expired, if not auto-repeat.
for (n33 = 0; n33 < pnSpr; n33 = n33 + 1)
if (sprCurFrame[n33] >= sprFrames[n33] && !sprAutoRepeat[n33])
sprEnableAnimation[n33] <= 1'b0;
if (cs_regs & wb_reqs.we) begin
casez (wb_reqs.adr[9:2])
8'b100?????:
begin
if (&wb_reqs.sel[1:0]) sprBurstStart[wb_reqs.adr[6:2]] <= {wb_reqs.dat[8:1],1'b0};
if (&wb_reqs.sel[3:2]) sprBurstEnd[wb_reqs.adr[6:2]] <= {wb_reqs.dat[24:17],1'b0};
end
8'b101?????:
begin
if (wb_reqs.sel[0]) sprFrameSize[wb_reqs.adr[6:2]][11:4] <= wb_reqs.dat[7:0];
if (wb_reqs.sel[1]) sprFrames[wb_reqs.adr[6:2]] <= wb_reqs.dat[15:8];
if (wb_reqs.sel[2]) sprRate[wb_reqs.adr[6:2]][7:0] <= wb_reqs.dat[23:16];
if (wb_reqs.sel[3])
begin
sprRate[wb_reqs.adr[6:2]][9:8] <= wb_reqs.dat[25:24];
sprFrameSize[wb_reqs.adr[6:2]][3:0] <= wb_reqs.dat[29:26];
sprAutoRepeat[wb_reqs.adr[6:2]] <= wb_reqs.dat[30];
sprEnableAnimation[wb_reqs.adr[6:2]] <= wb_reqs.dat[31];
end
end
8'b11110000: // 3C0
begin
if (wb_reqs.sel[0]) sprEn[7:0] <= wb_reqs.dat[7:0];
if (wb_reqs.sel[1]) sprEn[15:8] <= wb_reqs.dat[15:8];
if (wb_reqs.sel[2]) sprEn[23:16] <= wb_reqs.dat[23:16];
if (wb_reqs.sel[3]) sprEn[31:24] <= wb_reqs.dat[31:24];
end
8'b11110001: // 3C4
if (wb_reqs.sel[0]) begin
sprSprIe <= wb_reqs.dat[0];
sprBkIe <= wb_reqs.dat[1];
end
// update DMA trigger
// s_wb_reqs.dat[7:0] indicates which triggers to set (1=set,0=ignore)
// s_wb_reqs.dat[7:0] indicates which triggers to clear (1=clear,0=ignore)
8'b11110100: // 3D0
begin
if (wb_reqs.sel[0]) sprDt[7:0] <= sprDt[7:0] | wb_reqs.dat[7:0];
if (wb_reqs.sel[1]) sprDt[15:8] <= sprDt[15:8] | wb_reqs.dat[15:8];
if (wb_reqs.sel[2]) sprDt[23:16] <= sprDt[23:16] | wb_reqs.dat[23:16];
if (wb_reqs.sel[3]) sprDt[31:24] <= sprDt[31:24] | wb_reqs.dat[31:24];
end
8'b11110101: // 3D4
begin
if (wb_reqs.sel[0]) sprDt[7:0] <= sprDt[7:0] & ~wb_reqs.dat[7:0];
if (wb_reqs.sel[1]) sprDt[15:8] <= sprDt[15:8] & ~wb_reqs.dat[15:8];
if (wb_reqs.sel[2]) sprDt[23:16] <= sprDt[23:16] & ~wb_reqs.dat[23:16];
if (wb_reqs.sel[3]) sprDt[31:24] <= sprDt[31:24] & ~wb_reqs.dat[31:24];
end
8'b11110110: // 3D8
begin
if (wb_reqs.sel[0]) vSyncT[7:0] <= wb_reqs.dat[7:0];
if (wb_reqs.sel[1]) vSyncT[15:8] <= wb_reqs.dat[15:8];
if (wb_reqs.sel[2]) vSyncT[23:16] <= wb_reqs.dat[23:16];
if (wb_reqs.sel[3]) vSyncT[31:24] <= wb_reqs.dat[31:24];
end
8'b11111010: // 3E8
begin
if (wb_reqs.sel[0]) bgTc[7:0] <= wb_reqs.dat[7:0];
if (wb_reqs.sel[1]) bgTc[15:8] <= wb_reqs.dat[15:8];
if (wb_reqs.sel[2]) bgTc[23:16] <= wb_reqs.dat[23:16];
if (wb_reqs.sel[3]) bgTc[29:24] <= wb_reqs.dat[29:24];
end
8'b11111011: // 3EC
begin
if (wb_reqs.sel[0]) bkColor[7:0] <= wb_reqs.dat[7:0];
if (wb_reqs.sel[1]) bkColor[15:8] <= wb_reqs.dat[15:8];
if (wb_reqs.sel[2]) bkColor[23:16] <= wb_reqs.dat[23:16];
if (wb_reqs.sel[3]) bkColor[29:24] <= wb_reqs.dat[29:24];
end
// 8'b11111100: // 3F0
// if (wb_reqs.sel[0]) ce <= wb_reqs[0];
8'b0?????00:
begin
if (wb_reqs.sel[0]) hSprPos[sprN][ 7:0] <= wb_reqs.dat[ 7: 0];
if (wb_reqs.sel[1]) hSprPos[sprN][11:8] <= wb_reqs.dat[11: 8];
if (wb_reqs.sel[2]) vSprPos[sprN][ 7:0] <= wb_reqs.dat[23:16];
if (wb_reqs.sel[3]) vSprPos[sprN][11:8] <= wb_reqs.dat[27:24];
end
8'b0?????01:
begin
if (wb_reqs.sel[0]) begin
sprWidth[sprN] <= wb_reqs.dat[7:0];
end
if (wb_reqs.sel[1]) begin
sprHeight[sprN] <= wb_reqs.dat[15:8];
end
if (wb_reqs.sel[2]) begin
hSprRes[sprN] <= wb_reqs.dat[19:16];
vSprRes[sprN] <= wb_reqs.dat[23:20];
end
if (wb_reqs.sel[3]) begin
sprPlane[sprN] <= wb_reqs.dat[27:24];
sprColorDepth[sprN] <= wb_reqs.dat[31:30];
end
end
8'b0?????10:
begin // DMA address set on clk_i domain
if (wb_reqs.sel[0]) sprImageOffs[sprN][ 7:0] <= wb_reqs.dat[7:0];
if (wb_reqs.sel[1]) sprImageOffs[sprN][11:8] <= wb_reqs.dat[11:8];
if (wb_reqs.sel[1]) sprSysAddr[sprN][15:12] <= wb_reqs.dat[15:12];
if (wb_reqs.sel[2]) sprSysAddr[sprN][23:16] <= wb_reqs.dat[23:16];
if (wb_reqs.sel[3]) sprSysAddr[sprN][31:24] <= wb_reqs.dat[31:24];
end
8'b0?????11:
begin
if (wb_reqs.sel[0]) sprTc[sprN][ 7:0] <= wb_reqs.dat[ 7:0];
if (wb_reqs.sel[1]) sprTc[sprN][15:8] <= wb_reqs.dat[15:8];
if (wb_reqs.sel[2]) sprTc[sprN][23:16] <= wb_reqs.dat[23:16];
end
default: ;
endcase
end
end
//-------------------------------------------------------------
// Sprite Image Cache RAM
// This RAM is dual ported with an SoC side and a display
// controller side.
//-------------------------------------------------------------
integer n2;
always_ff @(posedge vclk)
for (n2 = 0; n2 < pnSpr; n2 = n2 + 1)
case(sprColorDepth[n2])
2'd1: sprAddr1[n2] <= {sprAddr[n2][11:3],~sprAddr[n2][2]};
2'd2: sprAddr1[n2] <= {sprAddr[n2][10:2],~sprAddr[n2][1]};
2'd3: sprAddr1[n2] <= {sprAddr[n2][ 9:1],~sprAddr[n2][0]};
default: ;
endcase
// The three LSBs of the image index need to be pipelined so they may be used
// to select the pixel. Output from the SRAM is always 32-bits wide so an
// additional mux is needed.
integer n4, n5, n27, n29;
always_ff @(posedge vclk)
for (n4 = 0; n4 < pnSpr; n4 = n4 + 1)
sprAddr2[n4] <= ~sprAddr[n4][2:0];
always_ff @(posedge vclk)
for (n5 = 0; n5 < pnSpr; n5 = n5 + 1)
sprAddr3[n5] <= sprAddr2[n5];
always_ff @(posedge vclk)
for (n27 = 0; n27 < pnSpr; n27 = n27 + 1)
sprAddr4[n27] <= sprAddr3[n27];
always_ff @(posedge vclk)
for (n29 = 0; n29 < pnSpr; n29 = n29 + 1)
sprAddr5[n29] <= sprAddr4[n29];
// The pixels are displayed from most signicant to least signicant bits of the
// word. Display order is opposite to memory storage. So, the least significant
// address bits are flipped to get the correct display.
integer n3;
always_ff @(posedge vclk)
for (n3 = 0; n3 < pnSpr; n3 = n3 + 1)
case(sprColorDepth[n3])
2'd1:
case(sprAddr5[n3][1:0])
2'd3: sprOut5[n3] <= sprOut4[n3][31:24];
2'd2: sprOut5[n3] <= sprOut4[n3][23:16];
2'd1: sprOut5[n3] <= sprOut4[n3][15:8];
2'd0: sprOut5[n3] <= sprOut4[n3][7:0];
endcase
2'd2:
case(sprAddr5[n3][0])
1'd0: sprOut5[n3] <= {sprOut4[n3][15],16'h0000,sprOut4[n3][14:0]};
1'd1: sprOut5[n3] <= {sprOut4[n3][31],16'h0000,sprOut4[n3][30:16]};
endcase
2'd3:
sprOut5[n3] <= sprOut4[n3];
default: ;
endcase
generate
for (g = 0; g < pnSpr; g = g + 1) begin : sprRam
SpriteRam sprRam0
(
.clka(m_clk_i),
.addra(m_adr_or),
.dina(m_dat_ir[63:0]),
.ena(sprWe[g]),
.wea(sprWe[g]),
// Core reg and output reg 3 clocks from read address
.clkb(vclk),
.addrb(sprAddr1[g]),
.doutb(sprOut4[g]),
.enb(1'b1)
);
end
endgenerate
//-------------------------------------------------------------
// Timing counters and addressing
// Sprites are like miniature bitmapped displays, they need
// all the same timing controls.
//-------------------------------------------------------------
// Create a timing reference using horizontal and vertical
// sync
wire hSyncEdge, vSyncEdge;
edge_det ed0(.rst(rst_i), .clk(vclk), .ce(1'b1), .i(hSync), .pe(hSyncEdge), .ne(), .ee() );
edge_det ed1(.rst(rst_i), .clk(vclk), .ce(1'b1), .i(vSync), .pe(vSyncEdge), .ne(), .ee() );
always_ff @(posedge vclk)
if (hSyncEdge) hctr <= {phBits{1'b0}};
else hctr <= hctr + 2'd1;
always_ff @(posedge vclk)
if (vSyncEdge) vctr <= {pvBits{1'b0}};
else if (hSyncEdge) vctr <= vctr + 2'd1;
// track sprite horizontal reset
integer n19;
always_ff @(posedge vclk)
for (n19 = 0; n19 < pnSpr; n19 = n19 + 1)
hSprReset[n19] <= hctr==hSprPos[n19];
// track sprite vertical reset
integer n20;
always_ff @(posedge vclk)
for (n20 = 0; n20 < pnSpr; n20 = n20 + 1)
vSprReset[n20] <= vctr==vSprPos[n20];
integer n21;
always_comb
for (n21 = 0; n21 < pnSpr; n21 = n21 + 1)
sprDe[n21] <= hSprDe[n21] & vSprDe[n21];
// take care of sprite size scaling
// video clock division
reg [31:0] hSprNextPixel;
reg [31:0] vSprNextPixel;
reg [3:0] hSprPt [31:0]; // horizontal pixel toggle
reg [3:0] vSprPt [31:0]; // vertical pixel toggle
integer n17;
always_comb
for (n17 = 0; n17 < pnSpr; n17 = n17 + 1)
hSprNextPixel[n17] = hSprPt[n17]==hSprRes[n17];
integer n18;
always_comb
for (n18 = 0; n18 < pnSpr; n18 = n18 + 1)
vSprNextPixel[n18] = vSprPt[n18]==vSprRes[n18];
// horizontal pixel toggle counter
integer n6;
always_ff @(posedge vclk)
for (n6 = 0; n6 < pnSpr; n6 = n6 + 1)
if (hSprReset[n6])
hSprPt[n6] <= 4'd0;
else if (hSprNextPixel[n6])
hSprPt[n6] <= 4'd0;
else
hSprPt[n6] <= hSprPt[n6] + 2'd1;
// vertical pixel toggle counter
integer n7;
always_ff @(posedge vclk)
for (n7 = 0; n7 < pnSpr; n7 = n7 + 1)
if (hSprReset[n7]) begin
if (vSprReset[n7])
vSprPt[n7] <= 4'd0;
else if (vSprNextPixel[n7])
vSprPt[n7] <= 4'd0;
else
vSprPt[n7] <= vSprPt[n7] + 2'd1;
end
// Animation rate count and frame increment.
integer n28;
always_ff @(posedge vclk)
if (vSyncEdge) begin
for (n28 = 0; n28 < pnSpr; n28 = n28 + 1) begin
if (sprEnableAnimation[n28]) begin
sprCurRateCount[n28] <= sprCurRateCount[n28] + 2'd1;
if (sprCurRateCount[n28] >= sprRate[n28]) begin
sprCurRateCount[n28] <= 'd0;
sprCurFrame[n28] <= sprCurFrame[n28] + 2'd1;
sprFrameProd[n28] <= sprFrameProd[n28] + sprFrameSize[n28];
if (sprCurFrame[n28] >= sprFrames[n28]) begin
sprCurFrame[n28] <= 'd0;
sprFrameProd[n28] <= 'd0;
end
end
end
else begin
sprCurFrame[n28] <= 'd0;
sprFrameProd[n28] <= 'd0;
end
end
end
// clock sprite image address counters
integer n8;
always_ff @(posedge vclk)
for (n8 = 0; n8 < pnSpr; n8 = n8 + 1) begin
// hReset and vReset - top left of sprite,
// reset address to image offset
if (hSprReset[n8] & vSprReset[n8]) begin
sprAddr[n8] <= sprImageOffs[n8] + sprFrameProd[n8];
sprAddrB[n8] <= sprImageOffs[n8] + sprFrameProd[n8];
end
// hReset:
// If the next vertical pixel
// set backup address to current address
// else
// set current address to backup address
// in order to rescan the line
else if (hSprReset[n8]) begin
if (vSprNextPixel[n8])
sprAddrB[n8] <= sprAddr[n8];
else
sprAddr[n8] <= sprAddrB[n8];
end
// Not hReset or vReset - somewhere on the sprite scan line
// just advance the address when the next pixel should be
// fetched
else if (hSprDe[n8] & hSprNextPixel[n8])
sprAddr[n8] <= sprAddr[n8] + 2'd1;
end
// clock sprite column (X) counter
integer n9;
always_ff @(posedge vclk)
for (n9 = 0; n9 < pnSpr; n9 = n9 + 1)
if (hSprReset[n9])
hSprCnt[n9] <= 8'd1;
else if (hSprNextPixel[n9])
hSprCnt[n9] <= hSprCnt[n9] + 2'd1;
// clock sprite horizontal display enable
integer n10;
always_ff @(posedge vclk)
for (n10 = 0; n10 < pnSpr; n10 = n10 + 1) begin
if (hSprReset[n10])
hSprDe[n10] <= 1'b1;
else if (hSprNextPixel[n10]) begin
if (hSprCnt[n10] == sprWidth[n10])
hSprDe[n10] <= 1'b0;
end
end
// clock the sprite row (Y) counter
integer n11;
always_ff @(posedge vclk)
for (n11 = 0; n11 < pnSpr; n11 = n11 + 1)
if (hSprReset[n11]) begin
if (vSprReset[n11])
vSprCnt[n11] <= 8'd1;
else if (vSprNextPixel[n11])
vSprCnt[n11] <= vSprCnt[n11] + 2'd1;
end
// clock sprite vertical display enable
integer n12;
always_ff @(posedge vclk)
for (n12 = 0; n12 < pnSpr; n12 = n12 + 1) begin
if (hSprReset[n12]) begin
if (vSprReset[n12])
vSprDe[n12] <= 1'b1;
else if (vSprNextPixel[n12]) begin
if (vSprCnt[n12] == sprHeight[n12])
vSprDe[n12] <= 1'b0;
end
end
end
//-------------------------------------------------------------
// Output stage
//-------------------------------------------------------------
// function used for color blending
// given an alpha and a color component, determine the resulting color
// this blends towards black or white
// alpha is eight bits ranging between 0 and 1.999...
// 1 bit whole, 7 bits fraction
function [11:0] fnBlend;
input [7:0] alpha;
input [11:0] color1bits;
input [11:0] color2bits;
begin
fnBlend = (({8'b0,color1bits} * alpha) >> 7) + (({8'h00,color2bits} * (9'h100 - alpha)) >> 7);
end
endfunction
// pipeline delays for display enable
reg [31:0] sprDe1, sprDe2, sprDe3, sprDe4, sprDe5, sprDe6;
reg [31:0] sproact;
integer n13;
always_ff @(posedge vclk)
for (n13 = 0; n13 < pnSpr; n13 = n13 + 1)
sprDe1[n13] <= sprDe[n13];
integer n22;
always_ff @(posedge vclk)
for (n22 = 0; n22 < pnSpr; n22 = n22 + 1)
sprDe2[n22] <= sprDe1[n22];
integer n23;
always_ff @(posedge vclk)
for (n23 = 0; n23 < pnSpr; n23 = n23 + 1)
sprDe3[n23] <= sprDe2[n23];
integer n24;
always_ff @(posedge vclk)
for (n24 = 0; n24 < pnSpr; n24 = n24 + 1)
sprDe4[n24] <= sprDe3[n24];
integer n25;
always_ff @(posedge vclk)
for (n25 = 0; n25 < pnSpr; n25 = n25 + 1)
sprDe5[n25] <= sprDe4[n25];
integer n26;
always_ff @(posedge vclk)
for (n26 = 0; n26 < pnSpr; n26 = n26 + 1)
sprDe6[n26] <= sprDe5[n26];
// Detect which sprite outputs are active
// The sprite output is active if the current display pixel
// address is within the sprite's area, the sprite is enabled,
// and it's not a transparent pixel that's being displayed.
integer n14;
always_ff @(posedge vclk)
for (n14 = 0; n14 < pnSpr; n14 = n14 + 1)
sproact[n14] <= sprEn[n14] && sprDe5[n14] && sprTc[n14]!=sprOut5[n14];
integer n15;
always_ff @(posedge vclk)
for (n15 = 0; n15 < pnSpr; n15 = n15 + 1)
sprOut[n15] <= sprOut5[n15];
// register sprite activity flag
// The image combiner uses this flag to know what to do with
// the sprite output.
always_ff @(posedge vclk)
outact <= |sproact & ce;
// Display data comes from the active sprite with the
// highest display priority.
// Make sure that alpha blending is turned off when
// no sprite is active.
integer n16;
always_ff @(posedge vclk)
begin
out <= 32'h0080; // alpha blend max (and off)
outplane <= 4'h0;
colorBits <= 2'b00;
for (n16 = pnSprm; n16 >= 0; n16 = n16 - 1)
if (sproact[n16]) begin
out <= sprOut[n16];
outplane <= sprPlane[n16];
colorBits <= sprColorDepth[n16];
end
end
// combine the text / graphics color output with sprite color output
// blend color output
wire [35:0] blendedColor32 = {
fnBlend(out[31:24],{out[23:16],4'h0},zrgb_i[35:24]),
fnBlend(out[31:24],{out[15:8],4'h0},zrgb_i[23:12]),
fnBlend(out[31:24],{out[7:0],4'h0},zrgb_i[11: 0])}
;
wire [35:0] blendedColor16 = {
fnBlend(out[7:0],zrgb_i[35:24],12'h0),
fnBlend(out[7:0],zrgb_i[23:12],12'h0),
fnBlend(out[7:0],zrgb_i[11: 0],12'h0)}
;
always_ff @(posedge vclk)
if (blank_i)
zrgb_o <= 0;
else begin
if (outact) begin
if (zrgb_i[39:36] > outplane) begin // rgb input is in front of sprite
zrgb_o <= zrgb_i;
end
else
if (!out[31]) begin // a sprite is displayed without alpha blending
case(colorBits)
2'd0: zrgb_o <= {outplane,out[7:5],9'b0,out[4:2],9'b0,out[1:0],10'b0};
2'd1: zrgb_o <= {outplane,out[7:5],9'b0,out[4:2],9'b0,out[1:0],10'b0};
2'd2: zrgb_o <= {outplane,out[14:10],7'b0,out[9:5],7'b0,out[4:0],7'b0};
2'd3: zrgb_o <= zrgb_o <= {outplane,blendedColor32}; // combine colors {outplane,out[23:16],4'h0,out[15:8],4'h0,out[7:0],4'h0};
endcase
end
else
case(colorBits)
2'd2: zrgb_o <= {outplane,blendedColor16}; // towards black/white
2'd3: zrgb_o <= {outplane,blendedColor32}; // combine colors
default: zrgb_o <= {outplane,out[7:5],9'b0,out[4:2],9'b0,out[1:0],10'b0}; // no blending
endcase
end
else
zrgb_o <= zrgb_i;
end
//--------------------------------------------------------------------
// Collision logic
//--------------------------------------------------------------------
// Detect when a sprite-sprite collision has occurred. The criteria
// for this is that a pixel from the sprite is being displayed, while
// there is a pixel from another sprite that could be displayed at the
// same time.
//--------------------------------------------------------------------
// ToDo: make collision also depend on plane
//--------------------------------------------------------------------
cntpop32 ucntp1 (
.i(sproact),
.o(actcnt)
);
// Detect when a sprite-background collision has occurred
integer n31;
always_comb
for (n31 = 0; n31 < pnSpr; n31 = n31 + 1)
bkCollision[n31] <=
sproact[n31] && zrgb_i[39:36]==sprPlane[n31];
// Load the sprite collision register. This register continually
// accumulates collision bits until reset by reading the register.
// Set the collision IRQ on the first collision and don't set it
// again until after the collision register has been read.
always @(posedge vclk)
if (rst_i) begin
sprSprIRQPending1 <= 0;
sprSprCollision1 <= 0;
sprSprIRQ1 <= 0;
end
else if (actcnt > 6'd1) begin
// isFirstCollision
if ((sprSprCollision1==0)||(cs_regs && wb_reqs.sel[0] && wb_reqs.adr[9:2]==8'b11110010)) begin
sprSprIRQPending1 <= 1;
sprSprIRQ1 <= sprSprIe;
sprSprCollision1 <= sproact;
end
else
sprSprCollision1 <= sprSprCollision1|sproact;
end
else if (cs_regs && wb_reqs.sel[0] && wb_reqs.adr[9:2]==8'b11110010) begin
sprSprCollision1 <= 0;
sprSprIRQPending1 <= 0;
sprSprIRQ1 <= 0;
end
// Load the sprite background collision register. This register
// continually accumulates collision bits until reset by reading
// the register.
// Set the collision IRQ on the first collision and don't set it
// again until after the collision register has been read.
// Note the background collision indicator is externally supplied,
// it will come from the color processing logic.
always @(posedge vclk)
if (rst_i) begin
sprBkIRQPending1 <= 0;
sprBkCollision1 <= 0;
sprBkIRQ1 <= 0;
end
else if (|bkCollision) begin
// Is the register being cleared at the same time
// a collision occurss ?
// isFirstCollision
if ((sprBkCollision1==0) || (cs_regs && wb_reqs.sel[0] && wb_reqs.adr[9:2]==8'b11110011)) begin
sprBkIRQ1 <= sprBkIe;
sprBkCollision1 <= bkCollision;
sprBkIRQPending1 <= 1;
end
else
sprBkCollision1 <= sprBkCollision1|bkCollision;
end
else if (cs_regs && wb_reqs.sel[0] && wb_reqs.adr[9:2]==8'b11110011) begin
sprBkCollision1 <= 0;
sprBkIRQPending1 <= 0;
sprBkIRQ1 <= 0;
end
endmodule
/*
module SpriteRam32 (
clka, adra, dia, doa, cea, wea,
clkb, adrb, dib, dob, ceb, web
);
input clka;
input [9:0] adra;
input [31:0] dia;
output [31:0] doa;
input cea;
input wea;
input clkb;
input [9:0] adrb;
input [31:0] dib;
output [31:0] dob;
input ceb;
input web;
reg [31:0] mem [0:1023];
reg [9:0] radra;
reg [9:0] radrb;
always @(posedge clka) if (cea) radra <= adra;
always @(posedge clkb) if (ceb) radrb <= adrb;
assign doa = mem [radra];
assign dob = mem [radrb];
always @(posedge clka)
if (cea & wea) mem[adra] <= dia;
always @(posedge clkb)
if (ceb & web) mem[adrb] <= dib;
endmodule
*/