read_dataflow_reuse_pattern

Author: shtaxxx

tests/extension/thread_/stream_reuse/test_thread_stream_reuse.py

@@ -25,25 +25,25 @@ def mkLed():
     ram_c = vthread.RAM(m, 'ram_c', clk, rst, datawidth, addrwidth)
     reuse_size = 8
 
-    def comp_stream(strm, size, offset):
+    def comp_stream(strm, size, roffset, woffset):
         a0, a1 = strm.read_reuse(
-            ram_a, offset, size, reuse_size=reuse_size, num_outputs=2)
+            ram_a, roffset, size, reuse_size=reuse_size, num_outputs=2)
         b0, b1 = strm.read_reuse(
-            ram_b, offset, size, reuse_size=reuse_size, num_outputs=2)
+            ram_b, roffset, size, reuse_size=reuse_size, num_outputs=2)
         sum = a0 + a1 + b0 + b1
-        strm.write(ram_c, offset, size * (reuse_size // 2), sum)
+        strm.write(ram_c, woffset, size * (reuse_size // 2), sum)
 
-    def comp_sequential(size, offset):
+    def comp_sequential(size, roffset, woffset):
         sum = 0
         w = 0
         for i in range(0, size, 2):
             for r in range(reuse_size):
-                a0 = ram_a.read(i + 0 + offset)
-                a1 = ram_a.read(i + 1 + offset)
-                b0 = ram_b.read(i + 0 + offset)
-                b1 = ram_b.read(i + 1 + offset)
+                a0 = ram_a.read(i + 0 + roffset)
+                a1 = ram_a.read(i + 1 + roffset)
+                b0 = ram_b.read(i + 0 + roffset)
+                b1 = ram_b.read(i + 1 + roffset)
                 sum = a0 + a1 + b0 + b1
-                ram_c.write(w + offset, sum)
+                ram_c.write(w + woffset, sum)
                 w += 1
 
     def check(size, offset_stream, offset_seq):
@@ -59,21 +59,23 @@ def check(size, offset_stream, offset_seq):
             print('NG')
 
     def comp(size):
-        offset = 0
-        myaxi.dma_read(ram_a, offset, 0, size)
-        myaxi.dma_read(ram_b, offset, 0, size)
-        stream.run(size, offset)
+        roffset = 0
+        woffset = 0
+        myaxi.dma_read(ram_a, roffset, 0, size)
+        myaxi.dma_read(ram_b, roffset, 0, size)
+        stream.run(size, roffset, woffset)
         stream.join()
-        myaxi.dma_write(ram_c, offset, 1024, size)
+        myaxi.dma_write(ram_c, woffset, 1024, size)
 
-        offset = size
-        myaxi.dma_read(ram_a, offset, 0, size)
-        myaxi.dma_read(ram_b, offset, 0, size)
-        sequential.run(size, offset)
+        roffset = size
+        woffset = size * (reuse_size // 2)
+        myaxi.dma_read(ram_a, roffset, 0, size)
+        myaxi.dma_read(ram_b, roffset, 0, size)
+        sequential.run(size, roffset, woffset)
         sequential.join()
-        myaxi.dma_write(ram_c, offset, 1024 * 2, size)
+        myaxi.dma_write(ram_c, woffset, 1024 * 2, size)
 
-        check(size, 0, offset)
+        check(size * (reuse_size // 2), 0, woffset)
 
     stream = vthread.Stream(m, 'mystream', clk, rst, comp_stream)
     sequential = vthread.Thread(m, 'th_sequential', clk, rst, comp_sequential)

tests/extension/thread_/stream_reuse/thread_stream_reuse.py

@@ -0,0 +1,29 @@
+TARGET=$(shell ls *.py | grep -v test | grep -v parsetab.py)
+ARGS=
+
+PYTHON=python3
+#PYTHON=python
+#OPT=-m pdb
+#OPT=-m cProfile -s time
+#OPT=-m cProfile -o profile.rslt
+
+.PHONY: all
+all: test
+
+.PHONY: run
+run:
+	$(PYTHON) $(OPT) $(TARGET) $(ARGS)
+
+.PHONY: test
+test:
+	$(PYTHON) -m pytest -vv
+
+.PHONY: check
+check:
+	$(PYTHON) $(OPT) $(TARGET) $(ARGS) > tmp.v
+	iverilog -tnull -Wall tmp.v
+	rm -f tmp.v
+
+.PHONY: clean
+clean:
+	rm -rf *.pyc __pycache__ parsetab.py .cache *.out *.png *.dot tmp.v uut.vcd