[misc] add verbose arg for zero and op builder (#3552)

* [misc] add print verbose

* [gemini] add print verbose

* [zero] add print verbose for low level

* [misc] add print verbose for op builder

op_builder/builder.py

@@ -7,7 +7,7 @@ import os
 import time
 from abc import ABC, abstractmethod
 from pathlib import Path
-from typing import List
+from typing import List, Optional
 
 from .utils import check_cuda_availability, check_system_pytorch_cuda_match, print_rank_0
 
@@ -138,7 +138,7 @@ class Builder(ABC):
         # make sure system CUDA and pytorch CUDA match, an error will raised inside the function if not
         check_system_pytorch_cuda_match(CUDA_HOME)
 
-    def load(self, verbose=True):
+    def load(self, verbose: Optional[bool] = None):
         """
         load the kernel during runtime. If the kernel is not built during pip install, it will build the kernel.
         If the kernel is built during runtime, it will be stored in `~/.cache/colossalai/torch_extensions/`. If the
@@ -149,6 +149,8 @@ class Builder(ABC):
         Args:
             verbose (bool, optional): show detailed info. Defaults to True.
         """
+        if verbose is None:
+            verbose = os.environ.get('CAI_KERNEL_VERBOSE', '0') == '1'
         # if the kernel has be compiled and cached, we directly use it
         if self.cached_op_module is not None:
             return self.cached_op_module