Refactor Logging for Improved Debugging (#115)

This pull request introduces several improvements and refactors across the codebase, focusing on enhanced logging, memory tracking, type checking, and developer usability. The most significant changes include replacing print statements with structured logging, adding detailed memory usage tracking for buffers, improving type and level checking for data types, and providing more informative string representations and debugging messages for core classes.

All runners and the generation scripts support different verbosity levels:
- `-v`: Show errors, warning and info messages
- `-vv`: Show errors, warnings, info and debug messages
- `-vvv`: Maximum verbosity with information about where the message was emitted. Also enable verbose compilation during the build step for the test runners.

- Logging and Error Handling Improvements: Replaced all `print` statements with structured logging using `DEFAULT_LOGGER` throughout the codebase, improving consistency and enabling better log management.
- Added debug-level logs to type checking and parsing routines to indicate success/failure and binding exhaustion, aiding troubleshooting.
- Implemented `__repr__` methods for core classes (`NodeTemplate`, `NodeBinding`, `NodeMapper`) to provide more informative string representations for debugging and logging
- Added `checkNumLevels` methods to relevant data type classes to validate that the number of levels assigned to buffers is supported by their type, and integrated warning logs if mismatches are detected during type inference.
- Introduced a `sizeInBytes` method to `VariableBuffer`, `TransientBuffer`, and related classes to standardize buffer size calculation.

- Report the maximum static and dynamic memory usage for all memory level:
  - Added dynamic and maximum memory usage tracking to `NetworkContext` via `_dynamicSize` and `_maxDynamicSize` dictionaries, and updated memory allocation/deallocation logic in `MemoryAllocation.py` to maintain these statistics per memory level. (This is only used for untitled platforms to track the maximum buffer size.)
  - Exposed new summary methods (`_printMemorySummary`, `_printInputOutputSummary`) in deployer wrappers and implemented input/output summary logging in `SignPropDeployer`.

- Deployer workflow now uses `prepare(...)` instead of `generateFunction(...)`.

Author: Xeratec

Deeploy/DeeployTypes.py

@@ -398,6 +398,17 @@ def sizeInBytes(self) -> int:
         """
         return (math.prod(self.shape) * (self._type.referencedType.typeWidth)) // 8
 
+    def sizeInBytes(self) -> int:
+        """Returns the size of this VariableBuffer in bytes
+
+        Returns
+        -------
+        int
+            Size of this VariableBuffer in bytes
+
+        """
+        return (math.prod(self.shape) * (self._type.referencedType.typeWidth)) // 8
+
 
 class TransientBuffer(VariableBuffer):
     """Class to represent memory space required by kernels that is not covered by input and output tensors, e.g. im2col buffers in convolutions
@@ -433,6 +444,9 @@ def __repr__(self) -> str:
     def sizeInBytes(self) -> int:
         return int(self.size)
 
+    def sizeInBytes(self) -> int:
+        return int(self.size)
+
 
 class ConstantBuffer(VariableBuffer):
     """Class to represent compile-time constant tensors (weights, biases, other parameters) within Deeploy.

Deeploy/MemoryLevelExtension/NetworkDeployers/MemoryLevelDeployer.py

@@ -128,6 +128,18 @@ def getTargetMemoryLevelMapping(self) -> TargetMemoryLevelMapping:
             f"Platform should be a MemoryPlatform or MemoryPlatformWrapper! Got {type(self.Platform).__name__}"
         return TargetMemoryLevelMapping(self.graph, self.Platform, self.ctxt)
 
+    def _parseNode(self, node: ONNXLayer, ctxt: NetworkContext,
+                   default_channels_first: bool) -> Tuple[NetworkContext, bool]:
+
+        newCtxt, parsePass = super()._parseNode(node, ctxt, default_channels_first)
+
+        if not parsePass:
+            return ctxt, False
+
+        newCtxt, self.graph = self.memoryLevelAnnotationOptimizer.optimize(newCtxt, self.graph)
+
+        return newCtxt, parsePass
+
     def bind(self):
         log.info("- Perform Memory Level Annotation")
         # LMACAN: Annotate before bind because during binding (specifically alignToContext) templates
@@ -138,6 +150,7 @@ def bind(self):
         if not ret:
             return False
 
+        log.info("- Perform Memory Level Annotation")
         # SCHEREMO: There might be hoisting; reassign memoryLevel preferences
         self.ctxt, self.graph = self.memoryLevelAnnotationOptimizer.optimize(self.ctxt, self.graph)
 
@@ -182,6 +195,7 @@ def bind(self):
         if not ret:
             return False
 
+        log.info("- Perform Memory Level Annotation")
         # SCHEREMO: There might be hoisting; reassign memoryLevel preferences
         self.ctxt, self.graph = self.memoryLevelAnnotationOptimizer.optimize(self.ctxt, self.graph)
 
@@ -217,6 +231,7 @@ def bind(self):
         if not ret:
             return False
 
+        log.info("- Perform Memory Level Annotation")
         # SCHEREMO: There might be hoisting; reassign memoryLevel preferences
         self.ctxt, self.graph = self.memoryLevelAnnotationOptimizer.optimize(self.ctxt, self.graph)
 

Deeploy/Targets/PULPOpen/Platform.py

@@ -233,7 +233,8 @@ class PULPStructBuffer(StructBuffer):
     PULPMatMulRequantMergePass(),
     PULPAddRequantMergePass(),
     RemoveEmptyConvBiasPass(),
-])
+],
+                                  name = "PULPOptimizer")
 
 # SCHEREMO: stdint is included before pulp_nn_kernels.h because it is supposed to be included in there, but isn't...
 _includeList = [

DeeployTest/generateNetwork.py

@@ -154,29 +154,6 @@ def generateNetwork(args):
 
     generateTestNetwork(deployer, test_inputs, test_outputs, args.dumpdir, verbosityCfg)
 
-    if args.verbose:
-        print()
-        print("=" * 80)
-        print("Output:")
-        for i in range(len(test_outputs)):
-            buffer = deployer.ctxt.lookup(f"output_{i}")
-            logLine = f" - '{buffer.name}': Type: {buffer._type.referencedType.typeName}"
-            if signProp:
-                logLine += f", nLevels: {buffer.nLevels}, Signed: {buffer._signed}"
-            print(logLine)
-        print('Input:')
-        for i in range(len(test_inputs)):
-            buffer = deployer.ctxt.lookup(f"input_{i}")
-            print(
-                f" - '{buffer.name}': Type: {buffer._type.referencedType.typeName}, Offset: {inputOffsets[buffer.name]}"
-            )
-        print("=" * 80)
-        num_ops = deployer.numberOfOps(args.verbose)
-        print("=" * 80)
-        print()
-        print(f"{'Number of Ops:' :<{_TEXT_ALIGN}} {num_ops}")
-        print(f"{'Model Parameters: ' :<{_TEXT_ALIGN}} {deployer.getParameterSize()}")
-
 
 if __name__ == '__main__':
 

DeeployTest/testMVP.py

@@ -291,31 +291,3 @@ def setupDeployer(graph: gs.Graph, memoryHierarchy: MemoryHierarchy, defaultTarg
                     values -= buffer.nLevels // 2
 
         generateTestNetwork(deployer, test_inputs, test_outputs, args.dumpdir, verbosityCfg)
-
-        if args.verbose:
-            print()
-            print("=" * 80)
-            print("Output:")
-            for i in range(len(test_outputs)):
-                buffer = deployer.ctxt.lookup(f"output_{i}")
-                logLine = f" - '{buffer.name}': Type: {buffer._type.referencedType.typeName}"
-                if signProp:
-                    logLine += f", nLevels: {buffer.nLevels}, Signed: {buffer._signed}"
-                print(logLine)
-            print('Input:')
-            for i in range(len(test_inputs)):
-                buffer = deployer.ctxt.lookup(f"input_{i}")
-                print(
-                    f" - '{buffer.name}': Type: {buffer._type.referencedType.typeName}, Offset: {inputOffsets[buffer.name]}"
-                )
-            print("=" * 80)
-            num_ops = deployer.numberOfOps(args.verbose)
-            print("=" * 80)
-            print()
-            print(f"{'Number of Ops:' :<{_TEXT_ALIGN}} {num_ops}")
-            print('Worst Case Buffer Size:')
-            for level in deployer.worstCaseBufferSize.keys():
-                print(f"{'  ' + str(level) + ':' :<{_TEXT_ALIGN}} {deployer.worstCaseBufferSize[level]}")
-            print(f"{'Model Parameters: ' :<{_TEXT_ALIGN}} {deployer.getParameterSize()}")
-
-        print("\033[92mCode Generation test ended, no memory violations!\033[0m")