[GitHub] [tvm] manupa-arm commented on a change in pull request #9597: [microNPU] Move the compilation to use Target Hooks.

[GitBox <gi...@apache.org>]

manupa-arm commented on a change in pull request #9597:
URL: https://github.com/apache/tvm/pull/9597#discussion_r759000103



##########
File path: python/tvm/relay/backend/contrib/ethosu/codegen.py
##########
@@ -80,5 +63,50 @@ def _compile(ext_func):
     # that can perform scheduling based on user inputs such as
     # scratch memory size.
     tir_mod, params = lower_to_tir(mod["main"], copy_constants())
-    cmms, encoded_constants, scratch_size = tir_to_cs_translator.translate(tir_mod, params)
-    return cmms, encoded_constants, scratch_size
+
+    for idx in params.keys():
+        params[idx] = tvm.nd.array(params[idx])
+
+    primfunc = tir_mod["main"]
+    primfunc = primfunc.with_attr("global_symbol", ext_func.attrs["global_symbol"])
+    primfunc = primfunc.with_attr("ethos-u.constants", params)
+    primfunc = primfunc.with_attr("ethos-u.input_size", input_size)
+    primfunc = primfunc.with_attr("ethos-u.output_size", output_size)
+    return primfunc
+
+
+@tvm._ffi.register_func("relay.ext.ethos-u.primfunc_to_artifact")
+def primfunc_to_artifact(primfunc: tvm.tir.PrimFunc) -> util.CompilationArtifact:
+    """
+    This is hook for python-based lowering of TIR PrimFunc
+    that has undergone unified optimization to Compilation
+    Artifact destined for the microNPU.
+
+    Parameters
+    ----------
+    primfunc : tir.PrimFunc
+        TIR PrimFuncthat has undergone unified optimization

Review comment:
       Done in #9605 




-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: commits-unsubscribe@tvm.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org