Two cycle of matrix mul work

src/matrix_mul.v

@@ -140,6 +140,7 @@ always @(posedge clk or posedge rst) begin
                 end
             end
             GET_BLOCKS: begin
+
                 get_block_a <= 1;
                 get_block_b <= 1;
                 if (block_get_a_done && block_get_b_done) begin
@@ -163,12 +164,21 @@ always @(posedge clk or posedge rst) begin
             ACCUMULATE: begin
                 add_block <= 1;
                 // Wait for the add operation to complete
+                //display matrix C and matrix res
+                for (int i = 0; i < `A_M*`B_N; i = i + 1) begin
+                    $display("matrix_C[%d] = %d", i, matrix_C[i]);
+                    $display("matrix_res[%d] = %d", i, matrix_res[i]);
+                end
                 if (block_add_done) begin
+                    //the reason this is not a blocking statement is because somehow the next for loop which makes matrix_C = matrix_res would make matrix_C= matrix_res before add_block becomes 0, 
+                    //which would make matrix_res = 2 * matrix_C and not add the actual two matrices. However this might be fixed during actual synthesis.
+                    add_block = 0;
                     //make matrix_C = matrix_res
                     for (int i = 0; i < `A_M*`B_N; i = i + 1) begin
                         matrix_C[i] <= matrix_res[i];
+                        $display("matrix_C[%d] = %d", i, matrix_C[i]);
                     end
-                    add_block <= 0;
+
                     // Update your loop counters here
                     r <= r + `K;
                     if (r >= `A_P) begin

tests/Makefile

@@ -14,9 +14,9 @@ VERILOG_SOURCES = ../src/define.v \
                   ../src/matrix_mul.v
 
 # You will have to create separate test rules for each module
-#TOPLEVEL_LANG = verilog
+TOPLEVEL_LANG = verilog
 #TOPLEVEL = block_add
-MODULE = test_block_add
+#MODULE = test_block_add
 TOPLEVEL = matrix_mul
 MODULE = test_matrix_mul
 

tests/test_block_add.py

@@ -21,12 +21,18 @@ async def test_get_block(dut):
     await RisingEdge(dut.clk)
     for i in range(J*K):
             dut.multiplied_block[i].value = BinaryValue(value=float_to_float16(i+1), n_bits=16)
-
+    dut.multiplied_block[0].value = BinaryValue(value=float_to_float16(113), n_bits=16)
+    dut.multiplied_block[1].value = BinaryValue(value=float_to_float16(62), n_bits=16)
+    dut.multiplied_block[2].value = BinaryValue(value=float_to_float16(35), n_bits=16)
+    dut.multiplied_block[3].value = BinaryValue(value=float_to_float16(20), n_bits=16)
 
     # Initialize the buffer with sequential values
     for i in range(buffer_size):
         dut.buffer_temp[i].value = BinaryValue(value=float_to_float16(i), n_bits=16)
-
+    dut.buffer_temp[0].value = BinaryValue(value=float_to_float16(5), n_bits=16)
+    dut.buffer_temp[1].value = BinaryValue(value=float_to_float16(6), n_bits=16)
+    dut.buffer_temp[2].value = BinaryValue(value=float_to_float16(15), n_bits=16)
+    dut.buffer_temp[3].value = BinaryValue(value=float_to_float16(20), n_bits=16)
     # Define the start row and column for the block to be read
     start_row = 0
     start_col = 0

tests/test_matrix_mul.py

@@ -50,8 +50,7 @@ async def test_get_block(dut):
     print("Start of add", dut.block_add.start.value)
     #print_matrix(dut.block_add.buffer_temp, J, K, "Input to block_add")
     #print_matrix(dut.block_add.multiplied_block, J, K, "Input to block_add")
-    print_matrix(dut.block_add.buffer_result, 2,2, "Matrix C")
-    print("i,l,r", dut.i.value, dut.l.value, dut.r.value) 
+    print_matrix(dut.block_add.buffer_result, 2,2, "Matrix C") 
     for i in range(8):
         await RisingEdge(dut.clk)
         print("Counter", dut.systolic_array.counter.value)
@@ -66,13 +65,16 @@ async def test_get_block(dut):
     await RisingEdge(dut.clk)
     print("Current state", dut.state.value)
     print_matrix(dut.block_add.multiplied_block, 2,2, "Multiplied Result")
-    print_matrix(dut.block_add.buffer_temp, 2,2, "Buffer Temp")
-    #print i and l
     print(dut.add_block.value)
     await RisingEdge(dut.clk)
     print("Current state", dut.state.value)
     print_matrix(dut.matrix_res, J, K, "Resultant Matrix")
-
+    await RisingEdge(dut.clk)
+    print("Current state", dut.state.value)
+    await RisingEdge(dut.clk)
+    print("Current state", dut.state.value)
+    await RisingEdge(dut.clk)
+    print("Current state", dut.state.value)
 
     #here we should be going back to get block
     print("-------------------------")
@@ -99,6 +101,18 @@ async def test_get_block(dut):
     await RisingEdge(dut.clk)
     print("Current state", dut.state.value) 
     print("i,l,r", dut.i.value, dut.l.value, dut.r.value) 
-    print("Next state", dut.next_state.value)  
+    print("Next state", dut.next_state.value) 
+    await RisingEdge(dut.clk)
+    print("Current state", dut.state.value)
+    for i in range(8):
+        await RisingEdge(dut.clk)
+        print("Counter", dut.systolic_array.counter.value) 
+    print_matrix(dut.mul_result, J, K, "Matrix a after mult")
+    print_matrix(dut.matrix_C, J, K, "matrix C")
+    await RisingEdge(dut.clk)
+    print("Current state", dut.state.value)
+    await RisingEdge(dut.clk)
+    print("Current state", dut.state.value)
+    print_matrix(dut.matrix_C, J, K, "Resultant Matrix")