[cleanup] removing global compiler configuration

This is a part of ongoing effort to remove any global state left over
from pybuda.

There should be no need of having global configuration of compiler. If
user wants to override some of the exposed settings, it should be done
locally, by creating a new (default) config and then passing it into the
compile function. All tests which need to modify the config should do
the same.

No feature should rely on reading from a "volatile global
configuration". Instead, each feature must either be designed to receive
the user-specified compiler configuration directly or defer
config-dependent decisions for later (until the compilation stage where
the appropriate configuration is available).

In majority of the cases I would assume that direct access to the user
passed config is easily available.

Closes #147

forge/forge/__init__.py

@@ -65,16 +65,6 @@ def set_home_paths():
 from .config import (
     CompilerConfig,
     CompileDepth,
-    set_configuration_options,
-    set_epoch_break,
-    set_chip_break,
-    override_op_size,
-    PerfTraceLevel,
-    insert_buffering_nop,
-    insert_nop,
-    _internal_insert_fj_buffering_nop,
-    override_dram_queue_placement,
-    configure_mixed_precision,
 )
 from .verify import DepricatedVerifyConfig
 from .forgeglobal import forge_reset, set_device_pipeline, is_silicon, get_tenstorrent_device

forge/forge/compile.py

@@ -14,7 +14,6 @@
 from forge.config import (
     CompilerConfig,
     CompileDepth,
-    _get_global_compiler_config,
 )
 from forge._C import (
     link_past_cache_ios,
@@ -183,6 +182,7 @@ def compile_main(
     optimizer: Optional[Union[torch.optim.Optimizer, forge.optimizers.Optimizer]] = None,
     training: bool = False,
     attach_to: Optional[CompiledModel] = None,
+    compiler_cfg: CompilerConfig = CompilerConfig(),
 ) -> CompiledModel:
     """
     Main entry point for compiling modules from different frameworks for Tenstorrent devices.
@@ -218,9 +218,6 @@ def compile_main(
 
     assert isinstance(module, AnyModule), "Only PyTorch, TensorFlow, and Forge modules are supported."
 
-    compiler_cfg = _get_global_compiler_config()
-    compiler_cfg.apply_env_config_overrides()
-
     if module_name is None:
         module_name = module.__class__.__name__
 
@@ -440,7 +437,7 @@ def forge_compile_torch(
 
     inputs = list(inputs)
 
-    compiler_cfg = _get_global_compiler_config()
+    compiler_cfg = CompilerConfig()
     compiler_cfg.apply_env_config_overrides()
 
     compile_context: CompileContext = CompileContext(
@@ -461,7 +458,7 @@ def forge_compile(
     graph_name: str,
     *inputs: Union[Tensor, List[Any], Dict[str, Any]],
     targets: List[Tensor] = [],
-    compiler_cfg: Optional[CompilerConfig] = None,
+    compiler_cfg: CompilerConfig = CompilerConfig(),
     verify_cfg: Optional[DepricatedVerifyConfig] = None,
     losses: Optional[List[Tensor]] = None,
     microbatch_size: int = 1,
@@ -505,9 +502,6 @@ def forge_compile(
     if verify_cfg is None:
         verify_cfg = DepricatedVerifyConfig.disabled()  # no verification config provided, disable by default
 
-    if compiler_cfg is None:
-        compiler_cfg = _get_global_compiler_config()
-
     compiler_cfg.apply_env_config_overrides()
 
     compile_context: CompileContext = CompileContext(
@@ -642,9 +636,6 @@ def init_compile(context: CompileContext) -> CompileDepth:
     if force_full:
         compiler_cfg.compile_depth = CompileDepth.FULL
 
-    context.backend_output_directory = compiler_cfg.backend_output_dir
-    ci.initialize_output_build_directory(context.backend_output_directory)
-
     # compiler_cfg is fully formed
     if "FORGE_LOAD_CONFIG" in os.environ:
         compiler_cfg = load_compiler_cfg(compiler_cfg)
@@ -1098,9 +1089,6 @@ def generate_graph(
     graph = Graph(graph_name)
     graph.set_microbatch(1)
 
-    if compiler_cfg is None:
-        compiler_cfg = _get_global_compiler_config()
-
     # Trace through the modules
     all_subgraph_outputs = []
     outputs = inputs