default to rtlsim-based analytical model, reenable all tests in fifo sizing

Author: Lukas Stasytis

src/finn/builder/build_dataflow_config.py

@@ -298,7 +298,7 @@ class DataflowBuildConfig:
     #: to deduce the token access vectors empirically
     #: TREE_MODEL will use the tree mode of an operator if available, avoiding the generation
     #: of IP cores.
-    tav_generation_strategy: Optional[TAVGenerationMethod] = TAVGenerationMethod.TREE_MODEL
+    tav_generation_strategy: Optional[TAVGenerationMethod] = TAVGenerationMethod.RTLSIM
 
     #: Which strategy will be used for token access vector generation for FIFO sizing.
     #: RTLSIM will result in performing RTLSIM for each node

tests/fpgadataflow/test_fifosizing.py

@@ -65,15 +65,15 @@ def fetch_test_model(topology, wbits=2, abits=2):
     "method",
     [
         "analytic_model_based",
-       #  "analytic_rtlsim",
-       #  "largefifo_rtlsim",
+         "analytic_rtlsim",
+         "largefifo_rtlsim",
     ],
 )
 @pytest.mark.parametrize(
     "topology",
     [
         "tfc",
-      #  "cnv"
+        "cnv"
     ],
 )
 def test_fifosizing_linear(method, topology):