Skip to content

PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度

fallback compute type on b70 if needed (#1904)

Yi Liu0c845c3d24f1
diff --git a/auto_round_extension/ark/auto_round_kernel/qlinear.py b/auto_round_extension/ark/auto_round_kernel/qlinear.py
index 749d2db1c..8384f29ef 100644
--- a/auto_round_extension/ark/auto_round_kernel/qlinear.py
+++ b/auto_round_extension/ark/auto_round_kernel/qlinear.py
@@ -17,6 +17,8 @@ import math
import torch
import torch.nn as nn
+from .utils import fallback_compute_type_if_needed
+
try:
import auto_round_kernel
@@ -207,6 +209,7 @@ class QuantLinear(nn.Module):
self.sdt = "fp16"
self.cdt = "int8"
self.torch_dt = torch.float16
+ self.cdt = fallback_compute_type_if_needed(self.cdt, self.qweight.device.index)
else:
self.sdt = "fp32"
self.cdt = "auto"
diff --git a/auto_round_extension/ark/auto_round_kernel/utils.py b/auto_round_extension/ark/auto_round_kernel/utils.py
new file mode 100644
index 000000000..d1a2294f8
--- /dev/null
+++ b/auto_round_extension/ark/auto_round_kernel/utils.py
@@ -0,0 +1,58 @@
+# Copyright (c) 2026 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import logging
+import re
+import subprocess
+from functools import lru_cache
+
+import torch
+
+logger = logging.getLogger(__name__)
+
+B70_DEVICE_ID = "0xe223"
+
+
+@lru_cache(maxsize=None)
+def is_oneapi_ge_2026() -> bool:
+ try:
+ output = subprocess.check_output(
+ ["icpx", "--version"],
+ stderr=subprocess.STDOUT,
+ text=True,
+ )
+ except (subprocess.CalledProcessError, FileNotFoundError):
+ return False
+
+ match = re.search(r"Compiler\s+(\d{4})\.", output)
+ return bool(match and int(match.group(1)) >= 2026)
+
+
+@lru_cache(maxsize=None)
+def is_b70(device: int = 0) -> bool:
+ try:
+ name = torch.xpu.get_device_properties(device).name
+ except Exception:
+ return False
+
+ return B70_DEVICE_ID in name.lower()
+
+
+@lru_cache(maxsize=None)
+def fallback_compute_type_if_needed(compute_dtype: str, device: int = 0) -> str:
+ if compute_dtype.lower() == "int8" and is_b70(device) and not is_oneapi_ge_2026():
+ logger.warning("XMX int8 is not supported on B70 with oneAPI < 2026. Falling back to fp16.")
+ return "fp16"
+
+ return compute_dtype