// ================================================================ // NVDLA Open Source Project // // Copyright(c) 2016 - 2017 NVIDIA Corporation. Licensed under the // NVDLA Open Hardware License; Check "LICENSE" which comes with // this distribution for more information. // ================================================================ // File Name: NV_NVDLA_SDP_RDMA_EG_ro.v // ================================================================ // NVDLA Open Source Project // // Copyright(c) 2016 - 2017 NVIDIA Corporation. Licensed under the // NVDLA Open Hardware License; Check "LICENSE" which comes with // this distribution for more information. // ================================================================ // File Name: NV_NVDLA_SDP_define.h `include "simulate_x_tick.vh" module NV_NVDLA_SDP_RDMA_EG_ro ( nvdla_core_clk //|< i ,nvdla_core_rstn //|< i ,pwrbus_ram_pd //|< i ,cfg_dp_8 //|< i ,cfg_dp_size_1byte //|< i ,cfg_mode_per_element //|< i ,reg2dp_channel //|< i ,reg2dp_height //|< i ,reg2dp_width //|< i ,rod0_wr_pd //|< i ,rod1_wr_pd //|< i ,rod2_wr_pd //|< i ,rod3_wr_pd //|< i ,rod_wr_mask //|< i ,rod_wr_vld //|< i ,rod_wr_rdy //|> o ,roc_wr_pd //|< i ,roc_wr_vld //|< i ,roc_wr_rdy //|> o ,layer_end //|> o ,sdp_rdma2dp_pd //|> o ,sdp_rdma2dp_valid //|> o ,sdp_rdma2dp_ready //|< i ); input nvdla_core_clk; input nvdla_core_rstn; input [31:0] pwrbus_ram_pd; output sdp_rdma2dp_valid; input sdp_rdma2dp_ready; output [8*16:0] sdp_rdma2dp_pd; input [8*8 -1:0] rod0_wr_pd; input [8*8 -1:0] rod1_wr_pd; input [8*8 -1:0] rod2_wr_pd; input [8*8 -1:0] rod3_wr_pd; input [3:0] rod_wr_mask; input rod_wr_vld; output rod_wr_rdy; input [1:0] roc_wr_pd; input roc_wr_vld; output roc_wr_rdy; input cfg_dp_8; input cfg_dp_size_1byte; input cfg_mode_per_element; input [12:0] reg2dp_channel; input [12:0] reg2dp_height; input [12:0] reg2dp_width; output layer_end; wire [2:0] size_of_beat; wire [12:0] size_of_width; wire [13-3:0] size_of_surf; reg [1:0] beat_cnt; wire [2:0] beat_cnt_nxt; reg mon_beat_cnt; reg [12:0] count_h; reg [12:0] count_w; reg [13-3:0] count_c; wire is_last_beat; wire is_cube_end; wire is_last_c; wire is_last_h; wire is_last_w; wire is_line_end; wire is_surf_end; reg roc_rd_en; wire [1:0] roc_rd_pd; wire roc_rd_prdy; wire roc_rd_pvld; reg rodx_rd_en; wire [8*8 -1:0] rod0_rd_pd; wire rod0_rd_prdy; wire rod0_rd_pvld; wire rod0_wr_prdy; wire rod0_wr_pvld; wire [8*8 -1:0] rod1_rd_pd; wire rod1_rd_prdy; wire rod1_rd_pvld; wire rod1_wr_prdy; wire rod1_wr_pvld; wire [8*8 -1:0] rod2_rd_pd; wire rod2_rd_prdy; wire rod2_rd_pvld; wire rod2_wr_prdy; wire rod2_wr_pvld; wire [8*8 -1:0] rod3_rd_pd; wire rod3_rd_prdy; wire rod3_rd_pvld; wire rod3_wr_prdy; wire rod3_wr_pvld; wire [1:0] rod_sel; wire rod0_sel; wire rod1_sel; wire rod2_sel; wire rod3_sel; reg [8*8 -1:0] out_data_1bpe; wire [8*16 -1:0] out_data_1bpe_ext; reg [8*16 -1:0] out_data_2bpe; reg out_vld_1bpe; reg out_vld_2bpe; wire out_accept; wire out_rdy; wire out_vld; wire [8*16:0] out_pd; //======================================================= // DATA FIFO: WRITE SIDE //======================================================= assign rod_wr_rdy = ~(rod_wr_mask[0] & ~rod0_wr_prdy |rod_wr_mask[1] & ~rod1_wr_prdy | rod_wr_mask[2] & ~rod2_wr_prdy | rod_wr_mask[3] & ~rod3_wr_prdy ); assign rod0_wr_pvld = rod_wr_vld & rod_wr_mask[0] & ~(rod_wr_mask[1] & ~rod1_wr_prdy | rod_wr_mask[2] & ~rod2_wr_prdy | rod_wr_mask[3] & ~rod3_wr_prdy ); assign rod1_wr_pvld = rod_wr_vld & rod_wr_mask[1] & ~(rod_wr_mask[0] & ~rod0_wr_prdy | rod_wr_mask[2] & ~rod2_wr_prdy | rod_wr_mask[3] & ~rod3_wr_prdy ); assign rod2_wr_pvld = rod_wr_vld & rod_wr_mask[2] & ~(rod_wr_mask[0] & ~rod0_wr_prdy | rod_wr_mask[1] & ~rod1_wr_prdy | rod_wr_mask[3] & ~rod3_wr_prdy ); assign rod3_wr_pvld = rod_wr_vld & rod_wr_mask[3] & ~(rod_wr_mask[0] & ~rod0_wr_prdy | rod_wr_mask[1] & ~rod1_wr_prdy | rod_wr_mask[2] & ~rod2_wr_prdy ); NV_NVDLA_SDP_RDMA_EG_RO_dfifo u_rod0 ( .nvdla_core_clk (nvdla_core_clk) ,.nvdla_core_rstn (nvdla_core_rstn) ,.rod_wr_prdy (rod0_wr_prdy) ,.rod_wr_pvld (rod0_wr_pvld) ,.rod_wr_pd (rod0_wr_pd[8*8 -1:0]) ,.rod_rd_prdy (rod0_rd_prdy) ,.rod_rd_pvld (rod0_rd_pvld) ,.rod_rd_pd (rod0_rd_pd[8*8 -1:0]) ); NV_NVDLA_SDP_RDMA_EG_RO_dfifo u_rod1 ( .nvdla_core_clk (nvdla_core_clk) ,.nvdla_core_rstn (nvdla_core_rstn) ,.rod_wr_prdy (rod1_wr_prdy) ,.rod_wr_pvld (rod1_wr_pvld) ,.rod_wr_pd (rod1_wr_pd[8*8 -1:0]) ,.rod_rd_prdy (rod1_rd_prdy) ,.rod_rd_pvld (rod1_rd_pvld) ,.rod_rd_pd (rod1_rd_pd[8*8 -1:0]) ); NV_NVDLA_SDP_RDMA_EG_RO_dfifo u_rod2 ( .nvdla_core_clk (nvdla_core_clk) ,.nvdla_core_rstn (nvdla_core_rstn) ,.rod_wr_prdy (rod2_wr_prdy) ,.rod_wr_pvld (rod2_wr_pvld) ,.rod_wr_pd (rod2_wr_pd[8*8 -1:0]) ,.rod_rd_prdy (rod2_rd_prdy) ,.rod_rd_pvld (rod2_rd_pvld) ,.rod_rd_pd (rod2_rd_pd[8*8 -1:0]) ); NV_NVDLA_SDP_RDMA_EG_RO_dfifo u_rod3 ( .nvdla_core_clk (nvdla_core_clk) ,.nvdla_core_rstn (nvdla_core_rstn) ,.rod_wr_prdy (rod3_wr_prdy) ,.rod_wr_pvld (rod3_wr_pvld) ,.rod_wr_pd (rod3_wr_pd[8*8 -1:0]) ,.rod_rd_prdy (rod3_rd_prdy) ,.rod_rd_pvld (rod3_rd_pvld) ,.rod_rd_pd (rod3_rd_pd[8*8 -1:0]) ); //======================================================= // DATA FIFO: READ SIDE //======================================================= always @( cfg_mode_per_element or is_last_h or is_last_w ) begin begin if (cfg_mode_per_element) begin rodx_rd_en = 1'b1; end else begin rodx_rd_en = is_last_h & is_last_w; end end end assign rod0_rd_prdy = out_rdy & rodx_rd_en & rod0_sel & ~(rod1_sel & ~rod1_rd_pvld); assign rod1_rd_prdy = out_rdy & rodx_rd_en & rod1_sel & ~(rod0_sel & ~rod0_rd_pvld); assign rod2_rd_prdy = out_rdy & rodx_rd_en & rod2_sel & ~(rod3_sel & ~rod3_rd_pvld); assign rod3_rd_prdy = out_rdy & rodx_rd_en & rod3_sel & ~(rod2_sel & ~rod2_rd_pvld); //============== // CMD FIFO //============== // One entry in ROC indicates one entry from latency fifo NV_NVDLA_SDP_RDMA_EG_RO_cfifo u_roc ( .nvdla_core_clk (nvdla_core_clk) ,.nvdla_core_rstn (nvdla_core_rstn) ,.pwrbus_ram_pd (pwrbus_ram_pd) ,.roc_wr_prdy (roc_wr_rdy) ,.roc_wr_pvld (roc_wr_vld) ,.roc_wr_pd (roc_wr_pd[1:0]) ,.roc_rd_prdy (roc_rd_prdy) ,.roc_rd_pvld (roc_rd_pvld) ,.roc_rd_pd (roc_rd_pd[1:0]) ); always @( cfg_mode_per_element or is_surf_end or is_last_beat ) begin begin roc_rd_en = is_last_beat & (is_surf_end | cfg_mode_per_element); end end assign roc_rd_prdy = roc_rd_en & out_accept; assign size_of_beat = roc_rd_pvld ? (roc_rd_pd[1:0] + 1) : 3'b0; //============== // END //============== assign is_line_end = is_last_w; assign is_surf_end = is_line_end & is_last_h; assign is_cube_end = is_surf_end & is_last_c; //============== // Width Count //============== assign size_of_width = reg2dp_width; always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin count_w <= {13{1'b0}}; end else begin if (out_accept) begin begin if (is_line_end) begin count_w <= 0; end else begin count_w <= count_w + 1; end end end end end assign is_last_w = (count_w==size_of_width); //============== // HEIGHT Count //============== always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin count_h <= {13{1'b0}}; end else begin if (out_accept) begin if (is_surf_end) begin count_h <= 0; end else if (is_line_end) begin count_h <= count_h + 1; end end end end assign is_last_h = (count_h==reg2dp_height); //============== // SURF Count //============== assign size_of_surf = cfg_dp_8 ? {1'b0,reg2dp_channel[12:3]} : reg2dp_channel[12:3 -1]; always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin count_c <= {(14-3){1'b0}}; end else begin if (out_accept) begin if (is_cube_end) begin count_c <= 0; end else if (is_surf_end) begin count_c <= count_c + 1; end end end end assign is_last_c = (count_c==size_of_surf); //============== // BEAT CNT: used to foreach 1~4 16E rod FIFOs //============== wire [1:0] size_of_elem = (cfg_dp_size_1byte | !cfg_dp_8) ? 2'h1 : 2'h2; assign beat_cnt_nxt = beat_cnt + size_of_elem; always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin {mon_beat_cnt,beat_cnt} <= 3'h0; end else begin if (out_accept) begin if (cfg_mode_per_element) begin if (is_last_beat) begin {mon_beat_cnt,beat_cnt} <= 3'h0; end else begin {mon_beat_cnt,beat_cnt} <= beat_cnt_nxt; end end else if (is_surf_end) begin if (is_last_beat) begin {mon_beat_cnt,beat_cnt} <= 3'h0; end else begin {mon_beat_cnt,beat_cnt} <= beat_cnt_nxt; end end end end end assign is_last_beat = beat_cnt_nxt == size_of_beat; assign rod_sel = beat_cnt; assign rod0_sel = beat_cnt == 2'h0; assign rod2_sel = beat_cnt == 2'h2; assign rod1_sel = (cfg_dp_size_1byte | !cfg_dp_8)? beat_cnt == 2'h1 : beat_cnt == 2'h0; assign rod3_sel = (cfg_dp_size_1byte | !cfg_dp_8)? beat_cnt == 2'h3 : beat_cnt == 2'h2; ////dp int8 one byte per element or int16 two bytes per elment/////////// always @( rod_sel or rod0_rd_pd or rod1_rd_pd or rod2_rd_pd or rod3_rd_pd ) begin case (rod_sel) 2'd0: out_data_1bpe = rod0_rd_pd; 2'd1: out_data_1bpe = rod1_rd_pd; 2'd2: out_data_1bpe = rod2_rd_pd; 2'd3: out_data_1bpe = rod3_rd_pd; default : out_data_1bpe[8*8 -1:0] = {8*8{`x_or_0}}; endcase end always @( rod_sel or rod0_rd_pvld or rod1_rd_pvld or rod2_rd_pvld or rod3_rd_pvld ) begin case (rod_sel) 2'd0: out_vld_1bpe = rod0_rd_pvld; 2'd1: out_vld_1bpe = rod1_rd_pvld; 2'd2: out_vld_1bpe = rod2_rd_pvld; 2'd3: out_vld_1bpe = rod3_rd_pvld; default : out_vld_1bpe = {1{`x_or_0}}; endcase end //: my $m = 8; //: foreach my $i (0..${m}-1) { //: print "assign out_data_1bpe_ext[16*${i}+15:16*${i}] = {{8{out_data_1bpe[8*${i}+7]}}, out_data_1bpe[8*${i}+7:8*${i}]}; \n"; //: } //| eperl: generated_beg (DO NOT EDIT BELOW) assign out_data_1bpe_ext[16*0+15:16*0] = {{8{out_data_1bpe[8*0+7]}}, out_data_1bpe[8*0+7:8*0]}; assign out_data_1bpe_ext[16*1+15:16*1] = {{8{out_data_1bpe[8*1+7]}}, out_data_1bpe[8*1+7:8*1]}; assign out_data_1bpe_ext[16*2+15:16*2] = {{8{out_data_1bpe[8*2+7]}}, out_data_1bpe[8*2+7:8*2]}; assign out_data_1bpe_ext[16*3+15:16*3] = {{8{out_data_1bpe[8*3+7]}}, out_data_1bpe[8*3+7:8*3]}; assign out_data_1bpe_ext[16*4+15:16*4] = {{8{out_data_1bpe[8*4+7]}}, out_data_1bpe[8*4+7:8*4]}; assign out_data_1bpe_ext[16*5+15:16*5] = {{8{out_data_1bpe[8*5+7]}}, out_data_1bpe[8*5+7:8*5]}; assign out_data_1bpe_ext[16*6+15:16*6] = {{8{out_data_1bpe[8*6+7]}}, out_data_1bpe[8*6+7:8*6]}; assign out_data_1bpe_ext[16*7+15:16*7] = {{8{out_data_1bpe[8*7+7]}}, out_data_1bpe[8*7+7:8*7]}; //| eperl: generated_end (DO NOT EDIT ABOVE) ////dp int8 two byte per element/////////// always @( rod_sel or rod0_rd_pd or rod1_rd_pd or rod2_rd_pd or rod3_rd_pd ) begin case (rod_sel) 2'd0: out_data_2bpe = {rod1_rd_pd,rod0_rd_pd}; 2'd2: out_data_2bpe = {rod3_rd_pd,rod2_rd_pd}; default : out_data_2bpe[8*16 -1:0] = {8*16{`x_or_0}}; endcase end always @( rod_sel or rod0_rd_pvld or rod1_rd_pvld or rod2_rd_pvld or rod3_rd_pvld ) begin case (rod_sel) 2'd0: out_vld_2bpe = rod1_rd_pvld & rod0_rd_pvld; 2'd2: out_vld_2bpe = rod3_rd_pvld & rod2_rd_pvld; default : out_vld_2bpe = {1{`x_or_0}}; endcase end ////mux out data //// assign out_vld = (cfg_dp_size_1byte | !cfg_dp_8) ? out_vld_1bpe : out_vld_2bpe; assign out_pd[8*16 -1:0] = !cfg_dp_8 ? {{8*8{1'b0}},out_data_1bpe[8*8 -1:0]} : cfg_dp_size_1byte ? out_data_1bpe_ext[8*16 -1:0] : out_data_2bpe[8*16 -1:0]; assign out_pd[8*16] = is_cube_end; assign out_accept = out_vld & out_rdy; NV_NVDLA_SDP_RDMA_EG_RO_pipe_p1 pipe_p1 ( .nvdla_core_clk (nvdla_core_clk) //|< i ,.nvdla_core_rstn (nvdla_core_rstn) //|< i ,.out_pd (out_pd[8*16:0]) //|< w ,.out_vld (out_vld) //|< r ,.out_rdy (out_rdy) //|> w ,.sdp_rdma2dp_ready (sdp_rdma2dp_ready) //|< i ,.sdp_rdma2dp_pd (sdp_rdma2dp_pd[8*16:0]) //|> o ,.sdp_rdma2dp_valid (sdp_rdma2dp_valid) //|> o ); assign layer_end = sdp_rdma2dp_valid & sdp_rdma2dp_ready & sdp_rdma2dp_pd[8*16]; endmodule // NV_NVDLA_SDP_RDMA_EG_ro // ************************************************************************************************************** // Generated by ::pipe -m -bc sdp_rdma2dp_pd (sdp_rdma2dp_valid,sdp_rdma2dp_ready) <= out_pd[8*8:0] (out_vld, out_rdy) // ************************************************************************************************************** module NV_NVDLA_SDP_RDMA_EG_RO_pipe_p1 ( nvdla_core_clk ,nvdla_core_rstn ,out_pd ,out_vld ,out_rdy ,sdp_rdma2dp_ready ,sdp_rdma2dp_pd ,sdp_rdma2dp_valid ); input nvdla_core_clk; input nvdla_core_rstn; input out_vld; output out_rdy; input [8*16:0] out_pd; output [8*16:0] sdp_rdma2dp_pd; output sdp_rdma2dp_valid; input sdp_rdma2dp_ready; //: my $dw = 1 + 8*16; //: &eperl::pipe("-is -wid $dw -do sdp_rdma2dp_pd -vo sdp_rdma2dp_valid -ri sdp_rdma2dp_ready -di out_pd -vi out_vld -ro out_rdy"); //| eperl: generated_beg (DO NOT EDIT BELOW) // Reg reg out_rdy; reg skid_flop_out_rdy; reg skid_flop_out_vld; reg [129-1:0] skid_flop_out_pd; reg pipe_skid_out_vld; reg [129-1:0] pipe_skid_out_pd; // Wire wire skid_out_vld; wire [129-1:0] skid_out_pd; wire skid_out_rdy; wire pipe_skid_out_rdy; wire sdp_rdma2dp_valid; wire [129-1:0] sdp_rdma2dp_pd; // Code // SKID READY always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin out_rdy <= 1'b1; skid_flop_out_rdy <= 1'b1; end else begin out_rdy <= skid_out_rdy; skid_flop_out_rdy <= skid_out_rdy; end end // SKID VALID always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin skid_flop_out_vld <= 1'b0; end else begin if (skid_flop_out_rdy) begin skid_flop_out_vld <= out_vld; end end end assign skid_out_vld = (skid_flop_out_rdy) ? out_vld : skid_flop_out_vld; // SKID DATA always @(posedge nvdla_core_clk) begin if (skid_flop_out_rdy & out_vld) begin skid_flop_out_pd[129-1:0] <= out_pd[129-1:0]; end end assign skid_out_pd[129-1:0] = (skid_flop_out_rdy) ? out_pd[129-1:0] : skid_flop_out_pd[129-1:0]; // PIPE READY assign skid_out_rdy = pipe_skid_out_rdy || !pipe_skid_out_vld; // PIPE VALID always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin pipe_skid_out_vld <= 1'b0; end else begin if (skid_out_rdy) begin pipe_skid_out_vld <= skid_out_vld; end end end // PIPE DATA always @(posedge nvdla_core_clk) begin if (skid_out_rdy && skid_out_vld) begin pipe_skid_out_pd[129-1:0] <= skid_out_pd[129-1:0]; end end // PIPE OUTPUT assign pipe_skid_out_rdy = sdp_rdma2dp_ready; assign sdp_rdma2dp_valid = pipe_skid_out_vld; assign sdp_rdma2dp_pd = pipe_skid_out_pd; //| eperl: generated_end (DO NOT EDIT ABOVE) endmodule // NV_NVDLA_SDP_RDMA_EG_RO_pipe_p1 module NV_NVDLA_SDP_RDMA_EG_RO_dfifo ( nvdla_core_clk , nvdla_core_rstn , rod_wr_prdy , rod_wr_pvld , rod_wr_pd , rod_rd_prdy , rod_rd_pvld , rod_rd_pd ); input nvdla_core_clk; input nvdla_core_rstn; output rod_wr_prdy; input rod_wr_pvld; input [8*8 -1:0] rod_wr_pd; input rod_rd_prdy; output rod_rd_pvld; output [8*8 -1:0] rod_rd_pd; //: my $dw = 8*8; //: &eperl::pipe("-is -wid $dw -do rod_rd_pd -vo rod_rd_pvld -ri rod_rd_prdy -di rod_wr_pd -vi rod_wr_pvld -ro rod_wr_prdy"); //| eperl: generated_beg (DO NOT EDIT BELOW) // Reg reg rod_wr_prdy; reg skid_flop_rod_wr_prdy; reg skid_flop_rod_wr_pvld; reg [64-1:0] skid_flop_rod_wr_pd; reg pipe_skid_rod_wr_pvld; reg [64-1:0] pipe_skid_rod_wr_pd; // Wire wire skid_rod_wr_pvld; wire [64-1:0] skid_rod_wr_pd; wire skid_rod_wr_prdy; wire pipe_skid_rod_wr_prdy; wire rod_rd_pvld; wire [64-1:0] rod_rd_pd; // Code // SKID READY always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin rod_wr_prdy <= 1'b1; skid_flop_rod_wr_prdy <= 1'b1; end else begin rod_wr_prdy <= skid_rod_wr_prdy; skid_flop_rod_wr_prdy <= skid_rod_wr_prdy; end end // SKID VALID always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin skid_flop_rod_wr_pvld <= 1'b0; end else begin if (skid_flop_rod_wr_prdy) begin skid_flop_rod_wr_pvld <= rod_wr_pvld; end end end assign skid_rod_wr_pvld = (skid_flop_rod_wr_prdy) ? rod_wr_pvld : skid_flop_rod_wr_pvld; // SKID DATA always @(posedge nvdla_core_clk) begin if (skid_flop_rod_wr_prdy & rod_wr_pvld) begin skid_flop_rod_wr_pd[64-1:0] <= rod_wr_pd[64-1:0]; end end assign skid_rod_wr_pd[64-1:0] = (skid_flop_rod_wr_prdy) ? rod_wr_pd[64-1:0] : skid_flop_rod_wr_pd[64-1:0]; // PIPE READY assign skid_rod_wr_prdy = pipe_skid_rod_wr_prdy || !pipe_skid_rod_wr_pvld; // PIPE VALID always @(posedge nvdla_core_clk or negedge nvdla_core_rstn) begin if (!nvdla_core_rstn) begin pipe_skid_rod_wr_pvld <= 1'b0; end else begin if (skid_rod_wr_prdy) begin pipe_skid_rod_wr_pvld <= skid_rod_wr_pvld; end end end // PIPE DATA always @(posedge nvdla_core_clk) begin if (skid_rod_wr_prdy && skid_rod_wr_pvld) begin pipe_skid_rod_wr_pd[64-1:0] <= skid_rod_wr_pd[64-1:0]; end end // PIPE OUTPUT assign pipe_skid_rod_wr_prdy = rod_rd_prdy; assign rod_rd_pvld = pipe_skid_rod_wr_pvld; assign rod_rd_pd = pipe_skid_rod_wr_pd; //| eperl: generated_end (DO NOT EDIT ABOVE) endmodule `define FORCE_CONTENTION_ASSERTION_RESET_ACTIVE 1'b1 `include "simulate_x_tick.vh" module NV_NVDLA_SDP_RDMA_EG_RO_cfifo ( nvdla_core_clk , nvdla_core_rstn , roc_wr_prdy , roc_wr_pvld , roc_wr_pd , roc_rd_prdy , roc_rd_pvld , roc_rd_pd , pwrbus_ram_pd ); // spyglass disable_block W401 -- clock is not input to module input nvdla_core_clk; input nvdla_core_rstn; output roc_wr_prdy; input roc_wr_pvld; input [1:0] roc_wr_pd; input roc_rd_prdy; output roc_rd_pvld; output [1:0] roc_rd_pd; input [31:0] pwrbus_ram_pd; // Master Clock Gating (SLCG) // // We gate the clock(s) when idle or stalled. // This allows us to turn off numerous miscellaneous flops // that don't get gated during synthesis for one reason or another. // // We gate write side and read side separately. // If the fifo is synchronous, we also gate the ram separately, but if // -master_clk_gated_unified or -status_reg/-status_logic_reg is specified, // then we use one clk gate for write, ram, and read. // wire nvdla_core_clk_mgated_enable; // assigned by code at end of this module wire nvdla_core_clk_mgated; // used only in synchronous fifos NV_CLK_gate_power nvdla_core_clk_mgate( .clk(nvdla_core_clk), .reset_(nvdla_core_rstn), .clk_en(nvdla_core_clk_mgated_enable), .clk_gated(nvdla_core_clk_mgated) ); // // WRITE SIDE // wire wr_reserving; reg roc_wr_busy_int; // copy for internal use assign roc_wr_prdy = !roc_wr_busy_int; assign wr_reserving = roc_wr_pvld && !roc_wr_busy_int; // reserving write space? wire wr_popping; // fwd: write side sees pop? reg [2:0] roc_wr_count; // write-side count wire [2:0] wr_count_next_wr_popping = wr_reserving ? roc_wr_count : (roc_wr_count - 1'd1); // spyglass disable W164a W484 wire [2:0] wr_count_next_no_wr_popping = wr_reserving ? (roc_wr_count + 1'd1) : roc_wr_count; // spyglass disable W164a W484 wire [2:0] wr_count_next = wr_popping ? wr_count_next_wr_popping : wr_count_next_no_wr_popping; wire wr_count_next_no_wr_popping_is_4 = ( wr_count_next_no_wr_popping == 3'd4 ); wire wr_count_next_is_4 = wr_popping ? 1'b0 : wr_count_next_no_wr_popping_is_4; wire [2:0] wr_limit_muxed; // muxed with simulation/emulation overrides wire [2:0] wr_limit_reg = wr_limit_muxed; // VCS coverage off wire roc_wr_busy_next = wr_count_next_is_4 || // busy next cycle? (wr_limit_reg != 3'd0 && // check roc_wr_limit if != 0 wr_count_next >= wr_limit_reg) ; // VCS coverage on always @( posedge nvdla_core_clk_mgated or negedge nvdla_core_rstn ) begin if ( !nvdla_core_rstn ) begin roc_wr_busy_int <= 1'b0; roc_wr_count <= 3'd0; end else begin roc_wr_busy_int <= roc_wr_busy_next; if ( wr_reserving ^ wr_popping ) begin roc_wr_count <= wr_count_next; end //synopsys translate_off else if ( !(wr_reserving ^ wr_popping) ) begin end else begin roc_wr_count <= {3{`x_or_0}}; end //synopsys translate_on end end wire wr_pushing = wr_reserving; // data pushed same cycle as roc_wr_pvld // // RAM // reg [1:0] roc_wr_adr; // current write address // spyglass disable_block W484 always @( posedge nvdla_core_clk_mgated or negedge nvdla_core_rstn ) begin if ( !nvdla_core_rstn ) begin roc_wr_adr <= 2'd0; end else begin if ( wr_pushing ) begin roc_wr_adr <= roc_wr_adr + 1'd1; end end end // spyglass enable_block W484 wire rd_popping; reg [1:0] roc_rd_adr; // read address this cycle wire ram_we = wr_pushing && (roc_wr_count > 3'd0 || !rd_popping); // note: write occurs next cycle wire [1:0] roc_rd_pd_p; // read data out of ram wire [31 : 0] pwrbus_ram_pd; // Adding parameter for fifogen to disable wr/rd contention assertion in ramgen. // Fifogen handles this by ignoring the data on the ram data out for that cycle. NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2 ram ( .clk( nvdla_core_clk_mgated ) , .pwrbus_ram_pd ( pwrbus_ram_pd ) , .di ( roc_wr_pd ) , .we ( ram_we ) , .wa ( roc_wr_adr ) , .ra ( (roc_wr_count == 0) ? 3'd4 : {1'b0,roc_rd_adr} ) , .dout ( roc_rd_pd_p ) ); wire [1:0] rd_adr_next_popping = roc_rd_adr + 1'd1; // spyglass disable W484 always @( posedge nvdla_core_clk_mgated or negedge nvdla_core_rstn ) begin if ( !nvdla_core_rstn ) begin roc_rd_adr <= 2'd0; end else begin if ( rd_popping ) begin roc_rd_adr <= rd_adr_next_popping; end //synopsys translate_off else if ( !rd_popping ) begin end else begin roc_rd_adr <= {2{`x_or_0}}; end //synopsys translate_on end end // // SYNCHRONOUS BOUNDARY // assign wr_popping = rd_popping; // let it be seen immediately wire rd_pushing = wr_pushing; // let it be seen immediately // // READ SIDE // reg roc_rd_prdy_d; // roc_rd_prdy registered in cleanly always @( posedge nvdla_core_clk or negedge nvdla_core_rstn ) begin if ( !nvdla_core_rstn ) begin roc_rd_prdy_d <= 1'b1; end else begin roc_rd_prdy_d <= roc_rd_prdy; end end wire roc_rd_prdy_d_o; // combinatorial rd_busy wire roc_rd_pvld_p; // data out of fifo is valid reg roc_rd_pvld_int_o; // internal copy of roc_rd_pvld_o wire roc_rd_pvld_o = roc_rd_pvld_int_o; assign rd_popping = roc_rd_pvld_p && !(roc_rd_pvld_int_o && !roc_rd_prdy_d_o); reg [2:0] roc_rd_count_p; // read-side fifo count // spyglass disable_block W164a W484 wire [2:0] rd_count_p_next_rd_popping = rd_pushing ? roc_rd_count_p : (roc_rd_count_p - 1'd1); wire [2:0] rd_count_p_next_no_rd_popping = rd_pushing ? (roc_rd_count_p + 1'd1) : roc_rd_count_p; // spyglass enable_block W164a W484 wire [2:0] rd_count_p_next = rd_popping ? rd_count_p_next_rd_popping : rd_count_p_next_no_rd_popping; assign roc_rd_pvld_p = roc_rd_count_p != 0 || rd_pushing; always @( posedge nvdla_core_clk_mgated or negedge nvdla_core_rstn ) begin if ( !nvdla_core_rstn ) begin roc_rd_count_p <= 3'd0; end else begin if ( rd_pushing || rd_popping ) begin roc_rd_count_p <= rd_count_p_next; end //synopsys translate_off else if ( !(rd_pushing || rd_popping ) ) begin end else begin roc_rd_count_p <= {3{`x_or_0}}; end //synopsys translate_on end end // // SKID for -rd_busy_reg // reg [1:0] roc_rd_pd_o; // output data register wire rd_req_next_o = (roc_rd_pvld_p || (roc_rd_pvld_int_o && !roc_rd_prdy_d_o)) ; always @( posedge nvdla_core_clk_mgated or negedge nvdla_core_rstn ) begin if ( !nvdla_core_rstn ) begin roc_rd_pvld_int_o <= 1'b0; end else begin roc_rd_pvld_int_o <= rd_req_next_o; end end always @( posedge nvdla_core_clk_mgated ) begin if ( (rd_popping) ) begin roc_rd_pd_o <= roc_rd_pd_p; end //synopsys translate_off else if ( !((rd_popping)) ) begin end else begin roc_rd_pd_o <= {2{`x_or_0}}; end //synopsys translate_on end // // FINAL OUTPUT // assign roc_rd_pd = !roc_rd_prdy_d_o ? roc_rd_pd_o : roc_rd_pd_p; // skid reg assign reg roc_rd_pvld_d; // previous roc_rd_pvld assign roc_rd_prdy_d_o = !(roc_rd_pvld_d && !roc_rd_prdy_d ); assign roc_rd_pvld = !roc_rd_prdy_d_o ? roc_rd_pvld_o : roc_rd_pvld_p; always @( posedge nvdla_core_clk_mgated or negedge nvdla_core_rstn ) begin if ( !nvdla_core_rstn ) begin roc_rd_pvld_d <= 1'b0; end else begin roc_rd_pvld_d <= roc_rd_pvld; end end // Master Clock Gating (SLCG) Enables // // plusarg for disabling this stuff: // synopsys translate_off `ifndef SYNTH_LEVEL1_COMPILE `ifndef SYNTHESIS reg master_clk_gating_disabled; initial master_clk_gating_disabled = $test$plusargs( "fifogen_disable_master_clk_gating" ) != 0; `endif `endif // synopsys translate_on assign nvdla_core_clk_mgated_enable = ((wr_reserving || wr_pushing || wr_popping || (roc_wr_pvld && !roc_wr_busy_int) || (roc_wr_busy_int != roc_wr_busy_next)) || (rd_pushing || rd_popping || (roc_rd_pvld && roc_rd_prdy_d) || (roc_rd_pvld_int_o && roc_rd_prdy_d_o)) || (wr_pushing)) `ifdef FIFOGEN_MASTER_CLK_GATING_DISABLED || 1'b1 `endif // synopsys translate_off `ifndef SYNTH_LEVEL1_COMPILE `ifndef SYNTHESIS || master_clk_gating_disabled `endif `endif // synopsys translate_on ; // Simulation and Emulation Overrides of wr_limit(s) // `ifdef EMU `ifdef EMU_FIFO_CFG // Emulation Global Config Override // assign wr_limit_muxed = `EMU_FIFO_CFG.NV_NVDLA_SDP_RDMA_EG_RO_cfifo_wr_limit_override ? `EMU_FIFO_CFG.NV_NVDLA_SDP_RDMA_EG_RO_cfifo_wr_limit : 3'd0; `else // No Global Override for Emulation // assign wr_limit_muxed = 3'd0; `endif // EMU_FIFO_CFG `else // !EMU `ifdef SYNTH_LEVEL1_COMPILE // No Override for GCS Compiles // assign wr_limit_muxed = 3'd0; `else `ifdef SYNTHESIS // No Override for RTL Synthesis // assign wr_limit_muxed = 3'd0; `else // RTL Simulation Plusarg Override // VCS coverage off reg wr_limit_override; reg [2:0] wr_limit_override_value; assign wr_limit_muxed = wr_limit_override ? wr_limit_override_value : 3'd0; `ifdef NV_ARCHPRO event reinit; initial begin $display("fifogen reinit initial block %m"); -> reinit; end `endif `ifdef NV_ARCHPRO always @( reinit ) begin `else initial begin `endif wr_limit_override = 0; wr_limit_override_value = 0; // to keep viva happy with dangles if ( $test$plusargs( "NV_NVDLA_SDP_RDMA_EG_RO_cfifo_wr_limit" ) ) begin wr_limit_override = 1; $value$plusargs( "NV_NVDLA_SDP_RDMA_EG_RO_cfifo_wr_limit=%d", wr_limit_override_value); end end // VCS coverage on `endif `endif `endif // // Histogram of fifo depth (from write side's perspective) // // NOTE: it will reference `SIMTOP.perfmon_enabled, so that // has to at least be defined, though not initialized. // tbgen testbenches have it already and various // ways to turn it on and off. // `ifdef PERFMON_HISTOGRAM // synopsys translate_off `ifndef SYNTH_LEVEL1_COMPILE `ifndef SYNTHESIS perfmon_histogram perfmon ( .clk ( nvdla_core_clk ) , .max ( {29'd0, (wr_limit_reg == 3'd0) ? 3'd4 : wr_limit_reg} ) , .curr ( {29'd0, roc_wr_count} ) ); `endif `endif // synopsys translate_on `endif // spyglass disable_block W164a W164b W116 W484 W504 `ifdef SPYGLASS `else `ifdef FV_ASSERT_ON `else // synopsys translate_off `endif `ifdef ASSERT_ON `ifdef SPYGLASS wire disable_assert_plusarg = 1'b0; `else `ifdef FV_ASSERT_ON wire disable_assert_plusarg = 1'b0; `else wire disable_assert_plusarg = $test$plusargs("DISABLE_NESS_FLOW_ASSERTIONS"); `endif `endif wire assert_enabled = 1'b1 && !disable_assert_plusarg; `endif `ifdef FV_ASSERT_ON `else // synopsys translate_on `endif `ifdef ASSERT_ON //synopsys translate_off `ifndef SYNTH_LEVEL1_COMPILE `ifndef SYNTHESIS always @(assert_enabled) begin if ( assert_enabled === 1'b0 ) begin $display("Asserts are disabled for %m"); end end `endif `endif //synopsys translate_on `endif `endif // spyglass enable_block W164a W164b W116 W484 W504 //The NV_BLKBOX_SRC0 module is only present when the FIFOGEN_MODULE_SEARCH // define is set. This is to aid fifogen team search for fifogen fifo // instance and module names in a given design. `ifdef FIFOGEN_MODULE_SEARCH NV_BLKBOX_SRC0 dummy_breadcrumb_fifogen_blkbox (.Y()); `endif // spyglass enable_block W401 -- clock is not input to module // synopsys dc_script_begin // set_boundary_optimization find(design, "NV_NVDLA_SDP_RDMA_EG_RO_cfifo") true // synopsys dc_script_end endmodule // NV_NVDLA_SDP_RDMA_EG_RO_cfifo // // Flop-Based RAM // module NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2 ( clk , pwrbus_ram_pd , di , we , wa , ra , dout ); input clk; // write clock input [31 : 0] pwrbus_ram_pd; input [1:0] di; input we; input [1:0] wa; input [2:0] ra; output [1:0] dout; `ifndef FPGA NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_0 (.A(pwrbus_ram_pd[0])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_1 (.A(pwrbus_ram_pd[1])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_2 (.A(pwrbus_ram_pd[2])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_3 (.A(pwrbus_ram_pd[3])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_4 (.A(pwrbus_ram_pd[4])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_5 (.A(pwrbus_ram_pd[5])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_6 (.A(pwrbus_ram_pd[6])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_7 (.A(pwrbus_ram_pd[7])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_8 (.A(pwrbus_ram_pd[8])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_9 (.A(pwrbus_ram_pd[9])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_10 (.A(pwrbus_ram_pd[10])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_11 (.A(pwrbus_ram_pd[11])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_12 (.A(pwrbus_ram_pd[12])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_13 (.A(pwrbus_ram_pd[13])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_14 (.A(pwrbus_ram_pd[14])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_15 (.A(pwrbus_ram_pd[15])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_16 (.A(pwrbus_ram_pd[16])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_17 (.A(pwrbus_ram_pd[17])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_18 (.A(pwrbus_ram_pd[18])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_19 (.A(pwrbus_ram_pd[19])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_20 (.A(pwrbus_ram_pd[20])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_21 (.A(pwrbus_ram_pd[21])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_22 (.A(pwrbus_ram_pd[22])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_23 (.A(pwrbus_ram_pd[23])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_24 (.A(pwrbus_ram_pd[24])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_25 (.A(pwrbus_ram_pd[25])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_26 (.A(pwrbus_ram_pd[26])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_27 (.A(pwrbus_ram_pd[27])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_28 (.A(pwrbus_ram_pd[28])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_29 (.A(pwrbus_ram_pd[29])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_30 (.A(pwrbus_ram_pd[30])); NV_BLKBOX_SINK UJ_BBOX2UNIT_UNUSED_pwrbus_31 (.A(pwrbus_ram_pd[31])); `endif `ifdef EMU wire [1:0] dout_p; // we use an emulation ram here to save flops on the emulation board // so that the monstrous chip can fit :-) // reg [1:0] Wa0_vmw; reg we0_vmw; reg [1:0] Di0_vmw; always @( posedge clk ) begin Wa0_vmw <= wa; we0_vmw <= we; Di0_vmw <= di; end vmw_NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2 emu_ram ( .Wa0( Wa0_vmw ) , .we0( we0_vmw ) , .Di0( Di0_vmw ) , .Ra0( ra[1:0] ) , .Do0( dout_p ) ); assign dout = (ra == 4) ? di : dout_p; `else reg [1:0] ram_ff0; reg [1:0] ram_ff1; reg [1:0] ram_ff2; reg [1:0] ram_ff3; always @( posedge clk ) begin if ( we && wa == 2'd0 ) begin ram_ff0 <= di; end if ( we && wa == 2'd1 ) begin ram_ff1 <= di; end if ( we && wa == 2'd2 ) begin ram_ff2 <= di; end if ( we && wa == 2'd3 ) begin ram_ff3 <= di; end end reg [1:0] dout; always @(*) begin case( ra ) 3'd0: dout = ram_ff0; 3'd1: dout = ram_ff1; 3'd2: dout = ram_ff2; 3'd3: dout = ram_ff3; 3'd4: dout = di; //VCS coverage off default: dout = {2{`x_or_0}}; //VCS coverage on endcase end `endif // EMU endmodule // NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2 // emulation model of flopram guts // `ifdef EMU module vmw_NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2 ( Wa0, we0, Di0, Ra0, Do0 ); input [1:0] Wa0; input we0; input [1:0] Di0; input [1:0] Ra0; output [1:0] Do0; // Only visible during Spyglass to avoid blackboxes. `ifdef SPYGLASS_FLOPRAM assign Do0 = 2'd0; wire dummy = 1'b0 | (|Wa0) | (|we0) | (|Di0) | (|Ra0); `endif // synopsys translate_off `ifndef SYNTH_LEVEL1_COMPILE `ifndef SYNTHESIS reg [1:0] mem[3:0]; // expand mem for debug ease `ifdef EMU_EXPAND_FLOPRAM_MEM wire [1:0] Q0 = mem[0]; wire [1:0] Q1 = mem[1]; wire [1:0] Q2 = mem[2]; wire [1:0] Q3 = mem[3]; `endif // asynchronous ram writes always @(*) begin if ( we0 == 1'b1 ) begin #0.1; mem[Wa0] = Di0; end end assign Do0 = mem[Ra0]; `endif `endif // synopsys translate_on // synopsys dc_script_begin // synopsys dc_script_end // g2c if { [find / -null_ok -subdesign vmw_NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2] != {} } { set_attr preserve 1 [find / -subdesign vmw_NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2] } endmodule // vmw_NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2 //vmw: Memory vmw_NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2 //vmw: Address-size 2 //vmw: Data-size 2 //vmw: Sensitivity level 1 //vmw: Ports W R //vmw: terminal we0 WriteEnable0 //vmw: terminal Wa0 address0 //vmw: terminal Di0[1:0] data0[1:0] //vmw: //vmw: terminal Ra0 address1 //vmw: terminal Do0[1:0] data1[1:0] //vmw: //qt: CELL vmw_NV_NVDLA_SDP_RDMA_EG_RO_cfifo_flopram_rwsa_4x2 //qt: TERMINAL we0 TYPE=WE POLARITY=H PORT=1 //qt: TERMINAL Wa0[%d] TYPE=ADDRESS DIR=W BIT=%1 PORT=1 //qt: TERMINAL Di0[%d] TYPE=DATA DIR=I BIT=%1 PORT=1 //qt: //qt: TERMINAL Ra0[%d] TYPE=ADDRESS DIR=R BIT=%1 PORT=1 //qt: TERMINAL Do0[%d] TYPE=DATA DIR=O BIT=%1 PORT=1 //qt: `endif // EMU