URL
https://opencores.org/ocsvn/forwardcom/forwardcom/trunk
Subversion Repositories forwardcom
[/] [forwardcom/] [trunk/] [data_memory.sv] - Rev 18
Compare with Previous | Blame | View Log
//////////////////////////////////////////////////////////////////////////////////
// Engineer: Agner Fog
//
// Create Date: 2020-06-03
// Last modified: 2021-08-02
// Module Name: data_cache
// Project Name: ForwardCom soft core
// Target Devices: Artix 7
// Tool Versions: Vivado v. 2020.1
// License: CERN-OHL-W v. 2 or later
// Description: data memory or data cache for read/write data
//
//////////////////////////////////////////////////////////////////////////////////
`include "defines.vh"
// read/write data memory or cache, (2**`DATA_ADDR_WIDTH) bytes = 2**16 = 64kB
module data_memory (
input clock, // clock
input clock_enable, // clock enable. Used when single-stepping
input [`COMMON_ADDR_WIDTH-1:0] read_write_addr, // Address for reading and writing from/to ram
// The lower 3 bits of read_write_addr indicate a byte within an 8 bytes line
input read_enable, // read enable
input [1:0] read_data_size, // 8, 16, 32, or 64 bits read
input [7:0] write_enable, // write enable for each byte separately
input [63:0] write_data_in, // Data in. Always 64 bits. Any part of the write bus can be used when the data size is less than 64 bits
`ifdef DISTRIBUTED_RAM // Distributed RAM takes a lot of FPGA resources
output reg [`RB1:0] read_data_out // Data out
`else // Block RAM
output logic [`RB1:0] read_data_out // Data out
`endif
);
// read/write data ram
reg [63:0] dataram [0:(2**(`DATA_ADDR_WIDTH-3))-1]; // 64kB RAM
// split read/write address into double-word index, and byte index
logic [`DATA_ADDR_WIDTH-4:0] address_hi;
logic [2:0] address_lo;
logic address_valid;
always_comb begin
address_hi = read_write_addr[`DATA_ADDR_WIDTH-1:3]; // index to 64-bit lines
address_lo = read_write_addr[2:0]; // index to byte within line
address_valid = read_write_addr[`COMMON_ADDR_WIDTH-1:`DATA_ADDR_WIDTH] == 0; // exclude code addresses
end
// Data write:
always_ff @(posedge clock) if (clock_enable & address_valid) begin
// write data to RAM. Each byte enabled separately
if (write_enable[0]) dataram[address_hi][ 7: 0] <= write_data_in[ 7: 0];
if (write_enable[1]) dataram[address_hi][15: 8] <= write_data_in[15: 8];
if (write_enable[2]) dataram[address_hi][23:16] <= write_data_in[23:16];
if (write_enable[3]) dataram[address_hi][31:24] <= write_data_in[31:24];
if (write_enable[4]) dataram[address_hi][39:32] <= write_data_in[39:32];
if (write_enable[5]) dataram[address_hi][47:40] <= write_data_in[47:40];
if (write_enable[6]) dataram[address_hi][55:48] <= write_data_in[55:48];
if (write_enable[7]) dataram[address_hi][63:56] <= write_data_in[63:56];
end
// data read. Must have natural alignment
`ifdef DISTRIBUTED_RAM
// The multiplexer comes before the register. This is only possible with distributed RAM.
// Distributed RAM takes a lot of FPGA resources but may allow a slightly higher clock frequency.
always_ff @(posedge clock) if (clock_enable & address_valid & read_enable) begin
// Each 64-bit RAM line may be divided into
// eight bytes, four 16-bit halfwords, two 32-bit words, or one 64-bit double word:
case (address_lo)
0: read_data_out[7:0] <= dataram[address_hi][ 7: 0];
1: read_data_out[7:0] <= dataram[address_hi][15: 8];
2: read_data_out[7:0] <= dataram[address_hi][23:16];
3: read_data_out[7:0] <= dataram[address_hi][31:24];
4: read_data_out[7:0] <= dataram[address_hi][39:32];
5: read_data_out[7:0] <= dataram[address_hi][47:40];
6: read_data_out[7:0] <= dataram[address_hi][55:48];
7: read_data_out[7:0] <= dataram[address_hi][63:56];
endcase
case (address_lo[2:1])
0: read_data_out[15:8] <= dataram[address_hi][15: 8];
1: read_data_out[15:8] <= dataram[address_hi][31:24];
2: read_data_out[15:8] <= dataram[address_hi][47:40];
3: read_data_out[15:8] <= dataram[address_hi][63:56];
endcase
case (address_lo[2])
0: read_data_out[31:16] <= dataram[address_hi][31:16];
1: read_data_out[31:16] <= dataram[address_hi][63:48];
endcase
`ifdef SUPPORT_64BIT
read_data_out[63:32] <= dataram[address_hi][63:32];
`endif
end
`else
// block RAM. The multiplexer must come after the register
reg [63:0] read_data; // a whole line read from the RAM
reg [2:0] address_lo2; // address_lo saved
always_ff @(posedge clock) if (clock_enable & address_valid & read_enable) begin
read_data <= dataram[address_hi]; // read a 64 bits line from ram
address_lo2 <= address_lo; // save low part of address
end
always_comb begin
// Each 64-bit RAM line may be divided into eight bytes, four 16-bit halfwords,
// two 32-bit words, or one 64-bit double word.
// The speed of this multiplexer is very critical because it adds to the delay
// in the execution unit. We are saving time by not setting unused parts of
// read_data_out to zero.
case (address_lo2)
0: read_data_out[7:0] = read_data[ 7: 0];
1: read_data_out[7:0] = read_data[15: 8];
2: read_data_out[7:0] = read_data[23:16];
3: read_data_out[7:0] = read_data[31:24];
4: read_data_out[7:0] = read_data[39:32];
5: read_data_out[7:0] = read_data[47:40];
6: read_data_out[7:0] = read_data[55:48];
7: read_data_out[7:0] = read_data[63:56];
endcase
case (address_lo2[2:1])
0: read_data_out[15:8] = read_data[15: 8];
1: read_data_out[15:8] = read_data[31:24];
2: read_data_out[15:8] = read_data[47:40];
3: read_data_out[15:8] = read_data[63:56];
endcase
case (address_lo2[2])
0: read_data_out[31:16] = read_data[31:16];
1: read_data_out[31:16] = read_data[63:48];
endcase
`ifdef SUPPORT_64BIT
read_data_out[63:32] = read_data[63:32];
`endif
end
`endif
endmodule