bigdl-core-npu 2.6.0b20241224__cp310-cp310-win_amd64.whl → 2.6.0b20241226__cp310-cp310-win_amd64.whl

{bigdl_core_npu-2.6.0b20241224.dist-info → bigdl_core_npu-2.6.0b20241226.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: bigdl-core-npu
-Version: 2.6.0b20241224
+Version: 2.6.0b20241226
 Summary: Intel® NPU Acceleration Library
 Home-page: https://github.com/intel/intel-npu-acceleration-library
 Author: Alessandro Palla

{bigdl_core_npu-2.6.0b20241224.dist-info → bigdl_core_npu-2.6.0b20241226.dist-info}/RECORD

@@ -1,10 +1,10 @@
 bigdl-core-npu/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-bigdl-core-npu/npu_llm.dll,sha256=VoXeir0s2n9sDtcW5UYwtvadk0fiHQLApF50qSTxC40,3923456
+bigdl-core-npu/npu_llm.dll,sha256=E-XoECLDiOffE7l6Od5dC9CWVLCS4vOrO1UY7nigPAQ,3923456
 bigdl-core-npu/npu_llm.lib,sha256=uG0enTHg0GNdEYMN36DPEKQXuYoioueKYpKXCKR3wyk,44328
 bigdl-core-npu/include/common.h,sha256=AWpPmCEAz6-oGUtq_ONHlELtOzSMqy4CBTMIppJeOII,2819
 bigdl-core-npu/include/npu_llm.h,sha256=Ar35OKx1qVbT3Q8TGTx_TQa6IlwIYBpn6oSo9q-t9YM,2509
 intel_npu_acceleration_library/__init__.py,sha256=ZKTIhGMDjF7P6pF-yX8KWcSXbeHWRk24AO_orsa18f8,536
-intel_npu_acceleration_library/_version.py,sha256=bGpy_0X6J5mqHLPOzXr3KwLUxDrUXJKcraaKC_anPPs,112
+intel_npu_acceleration_library/_version.py,sha256=uSR3Fqqy5_09oZgqJZWo4fR996ACvzrJvcijDaF_o1Q,112
 intel_npu_acceleration_library/compiler.py,sha256=3IdgqjamSC8MLexDBJypIeZRiWIcTFnvQSU1LPXUr7Y,6225
 intel_npu_acceleration_library/device.py,sha256=9bn8eVXJa5cXIqgfLsQAdkMVtVUQABb8z0-mQik5jRg,7424
 intel_npu_acceleration_library/dtypes.py,sha256=gdd06Wsc9zIZFHlauUEx4xcK9WGTn1Mu6GkuYDJeA-E,4683
@@ -12,8 +12,8 @@ intel_npu_acceleration_library/modelling.py,sha256=vSiQOWGJ0l6wGV7zWQtZEkHpnMQIM
 intel_npu_acceleration_library/optimizations.py,sha256=9NY8QoDFbs2LY12jbx6As8g2v0oInX4YzvkjnqViA70,5469
 intel_npu_acceleration_library/quantization.py,sha256=6N_04h1KX6TNbw-ceANV0Pmk4_lQ2Y9C7Pwn5x-zQzo,5566
 intel_npu_acceleration_library/backend/__init__.py,sha256=2NP6Ypr1dGUNXmLGW5GD9xrh0U9KJgqxTd_c7su1RUY,857
-intel_npu_acceleration_library/backend/base.py,sha256=0EXHZTMrelebJ6HOSe74zE1mhy9tghXrkYnRQDLzwk4,9492
-intel_npu_acceleration_library/backend/bindings.py,sha256=mu7EJ60X1cAFa1y17yA4r3n5lwFBSzsq5u-6Nj-OaJ0,10352
+intel_npu_acceleration_library/backend/base.py,sha256=KrL4odlFqzodOsRUwQIeq99Fg-R9OxAOltPrv9Jg7w8,10005
+intel_npu_acceleration_library/backend/bindings.py,sha256=9nL_eoJgWAgnJzDzFsUK-ALn6YSLBR1chTrKV-vQQNg,10446
 intel_npu_acceleration_library/backend/compression.py,sha256=Avz_zm2s_ELy5peVQ8zFGn8njBfh9nEGR16mflotBic,630
 intel_npu_acceleration_library/backend/convolution.py,sha256=cN3k78X3Y4Cbf7er-MFq0sJ4OwIvquj8PajpdEDmCo4,2018
 intel_npu_acceleration_library/backend/factory.py,sha256=RuVxbG-jHRhkkUU2q2kITVov3r-gBb6SGslWS-eL-Sk,41035
@@ -198,7 +198,7 @@ intel_npu_acceleration_library/external/openvino/torch/__init__.py,sha256=RXLzsf
 intel_npu_acceleration_library/functional/__init__.py,sha256=WWKwKOh6Sgovv7mKctA872TbLP98Pg5m5-MREvUmlAA,204
 intel_npu_acceleration_library/functional/scaled_dot_product_attention.py,sha256=yGUcg4tDQOLuUnP1g74cl-ec8TRr2SuAMcNLlN6qLvE,1620
 intel_npu_acceleration_library/lib/Release/cache.json,sha256=CyrSqZUWo0Ec4_7ydOiuKIC0Gm8AybrGdozUqUuHxBw,8840377
-intel_npu_acceleration_library/lib/Release/intel_npu_acceleration_library.dll,sha256=VolbV7pF3hFRrLi5MzbLaM00LBSxhtMn2Nte0kh4oi4,339968
+intel_npu_acceleration_library/lib/Release/intel_npu_acceleration_library.dll,sha256=WBTGRKJVvmX7Qwyzl1P8PwW3IrtCEgzWcQT3mmgbQJw,341504
 intel_npu_acceleration_library/lib/Release/openvino.dll,sha256=m7M119p3JBq2YYJJ2zzCaBDz6XivKK3nNykb8L1cvDU,13244768
 intel_npu_acceleration_library/lib/Release/openvino_auto_batch_plugin.dll,sha256=2v_I9P3Qo0St1bQZMEZscnFOUVvgZQQ0HvQlG3HtTd0,203104
 intel_npu_acceleration_library/lib/Release/openvino_auto_plugin.dll,sha256=e3Aj9CDRHN30dBEdPSk7OCWe52tWfhI4xeXgyFjuDHg,475488
@@ -228,7 +228,7 @@ intel_npu_acceleration_library/nn/functional.py,sha256=UfAKBc0u6RtyaMo14ldH2GpEn
 intel_npu_acceleration_library/nn/linear.py,sha256=Q06SoGQeLaI86nA_ky2GnFC6H2Fw1zyMDILKnpYC2eo,5739
 intel_npu_acceleration_library/nn/llm.py,sha256=P6dz36Yf6BHtzWcftaghC6QaMI_WeRfQwrCbO7fD6hk,15002
 intel_npu_acceleration_library/nn/module.py,sha256=EYxoTq6I_YgBDgTF76GPDxHrT8SupOTDGMzQaomBeq8,12667
-bigdl_core_npu-2.6.0b20241224.dist-info/METADATA,sha256=4YKnLD37v4ZTfcpH9TE1CEZ68xa4DRqOlCAOeU5Qcrs,1541
-bigdl_core_npu-2.6.0b20241224.dist-info/WHEEL,sha256=tcd-HDpskugT8GYYKyyid0lOlzoZtZdWwcrj5ormtfo,101
-bigdl_core_npu-2.6.0b20241224.dist-info/top_level.txt,sha256=iMQZlTsFPJjlD-Y0MqZEP_9ifI0LlbNCJIOTaMoGMjk,46
-bigdl_core_npu-2.6.0b20241224.dist-info/RECORD,,
+bigdl_core_npu-2.6.0b20241226.dist-info/METADATA,sha256=s4raV1hQE24vgYfahkhLmod_YL2uCBORa-JOeCe-VV8,1541
+bigdl_core_npu-2.6.0b20241226.dist-info/WHEEL,sha256=tcd-HDpskugT8GYYKyyid0lOlzoZtZdWwcrj5ormtfo,101
+bigdl_core_npu-2.6.0b20241226.dist-info/top_level.txt,sha256=iMQZlTsFPJjlD-Y0MqZEP_9ifI0LlbNCJIOTaMoGMjk,46
+bigdl_core_npu-2.6.0b20241226.dist-info/RECORD,,

intel_npu_acceleration_library/_version.py

@@ -3,4 +3,4 @@
 # SPDX-License-Identifier: Apache 2.0
 #
 
-__version__ = "2.6.0b20241224"
+__version__ = "2.6.0b20241226"

intel_npu_acceleration_library/backend/base.py

@@ -53,13 +53,14 @@ class BaseNPUBackend:
         ):
             backend_lib.destroyNNFactory(self._mm)
 
-    def save(self, path: str):
+    def save(self, path: str, compress_to_fp16: bool = True):
         """Save the Openvino model.
 
         Args:
             path (str): the model save path
+            compress_to_fp16 (bool): whether to compress floating point weights to FP16 (default: True).
         """
-        backend_lib.saveModel(self._mm, ctypes.c_char_p(path.encode()))
+        backend_lib.saveModel(self._mm, ctypes.c_char_p(path.encode()), compress_to_fp16)
 
     def saveCompiledModel(self, path: str):
         """Save the compiled model.
@@ -68,6 +69,15 @@ class BaseNPUBackend:
             path (str): the compiled model save path
         """
         backend_lib.saveCompiledModel(self._mm, ctypes.c_char_p(path.encode()))
+    
+    def serialize(self, xml_path: str, bin_path: str):
+        """Serialize the Openvino model.
+
+        Args:
+            xml_path (str): the model save xml path
+            bin_path (str): the model save bin path
+        """
+        backend_lib.serializeModel(self._mm, ctypes.c_char_p(xml_path.encode()), ctypes.c_char_p(bin_path.encode()))
 
 
 class BaseNPUBackendWithPrefetch(BaseNPUBackend):

intel_npu_acceleration_library/backend/bindings.py

@@ -67,8 +67,9 @@ def init_common(lib: ctypes.CDLL):
     Args:
         lib (ctypes.CDLL): Intel® NPU Acceleration Library runtime library
     """
-    lib.saveModel.argtypes = [handler, ctypes.c_char_p]
+    lib.saveModel.argtypes = [handler, ctypes.c_char_p, ctypes.c_bool]
     lib.saveCompiledModel.argtypes = [handler, ctypes.c_char_p]
+    lib.serializeModel.argtypes = [handler, ctypes.c_char_p, ctypes.c_char_p]
 
     # Set input activations
     lib.set_activation.argtypes = [handler, ctypes.c_void_p, ctypes.c_int]