Fixed CUDA setup bugs, including #81.

CHANGELOG.md

@@ -149,3 +149,9 @@ Bug fixes:
 
 Bug fixes:
  - Fixed a bug in the CUDA Setup which led to an incomprehensible error if no GPU was detected.
+
+### 0.35.4
+
+Bug fixes:
+ - Fixed a bug in the CUDA Setup failed with the cuda runtime was found, but not the cuda library.
+ - Fixed a bug where not finding the cuda runtime led to an incomprehensible error.

bitsandbytes/cextension.py

@@ -116,7 +116,7 @@ try:
         CUDASetup.get_instance().generate_instructions()
         CUDASetup.get_instance().print_log_stack()
         raise RuntimeError('''
-        CUDA Setup failed despite GPU being available. Inspect the CUDA SETUP outputs to fix your environment!
+        CUDA Setup failed despite GPU being available. Inspect the CUDA SETUP outputs aboveto fix your environment!
         If you cannot find any issues and suspect a bug, please open an issue with detals about your environment:
         https://github.com/TimDettmers/bitsandbytes/issues''')
     lib.cadam32bit_g32
@@ -124,8 +124,6 @@ try:
     lib.get_cusparse.restype = ct.c_void_p
     COMPILED_WITH_CUDA = True
 except AttributeError:
-    warn(
-        "The installed version of bitsandbytes was compiled without GPU support. "
-        "8-bit optimizers and GPU quantization are unavailable."
-    )
+    warn("The installed version of bitsandbytes was compiled without GPU support. "
+        "8-bit optimizers and GPU quantization are unavailable.")
     COMPILED_WITH_CUDA = False

bitsandbytes/cuda_setup/main.py

@@ -17,6 +17,7 @@ evaluation:
 """
 
 import ctypes
+import torch
 
 from .paths import determine_cuda_runtime_lib_path
 from bitsandbytes.cextension import CUDASetup
@@ -29,8 +30,11 @@ def check_cuda_result(cuda, result_val):
         cuda.cuGetErrorString(result_val, ctypes.byref(error_str))
         CUDASetup.get_instance().add_log_entry(f"CUDA exception! Error code: {error_str.value.decode()}")
 
+
+# https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART____VERSION.html#group__CUDART____VERSION
 def get_cuda_version(cuda, cudart_path):
-    # https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART____VERSION.html#group__CUDART____VERSION
+    if cuda is None: return None
+
     try:
         cudart = ctypes.CDLL(cudart_path)
     except OSError:
@@ -72,7 +76,6 @@ def get_compute_capabilities(cuda):
     # bits taken from https://gist.github.com/f0k/63a664160d016a491b2cbea15913d549
     """
 
-
     nGpus = ctypes.c_int()
     cc_major = ctypes.c_int()
     cc_minor = ctypes.c_int()
@@ -99,11 +102,11 @@ def get_compute_capability(cuda):
     capabilities are downwards compatible. If no GPUs are detected, it returns
     None.
     """
-    ccs = get_compute_capabilities(cuda)
-    if ccs:
+    if cuda is None: return None
+
     # TODO: handle different compute capabilities; for now, take the max
-        return ccs[-1]
-    return None
+    ccs = get_compute_capabilities(cuda)
+    if ccs: return ccs[-1]
 
 
 def evaluate_cuda_setup():
@@ -113,28 +116,31 @@ def evaluate_cuda_setup():
     #print('Welcome to bitsandbytes. For bug reports, please submit your error trace to: https://github.com/TimDettmers/bitsandbytes/issues')
     #print('For effortless bug reporting copy-paste your error into this form: https://docs.google.com/forms/d/e/1FAIpQLScPB8emS3Thkp66nvqwmjTEgxp8Y9ufuWTzFyr9kJ5AoI47dQ/viewform?usp=sf_link')
     #print('='*80)
-    #if not torch.cuda.is_available():
-        #print('No GPU detected. Loading CPU library...')
-        #return binary_name
-
-    binary_name = "libbitsandbytes_cpu.so"
+    if not torch.cuda.is_available(): return 'libsbitsandbytes_cpu.so', None, None, None, None
 
     cuda_setup = CUDASetup.get_instance()
     cudart_path = determine_cuda_runtime_lib_path()
-    if cudart_path is None:
-        cuda_setup.add_log_entry("WARNING: No libcudart.so found! Install CUDA or the cudatoolkit package (anaconda)!", is_warning=True)
-        return binary_name
-
-    cuda_setup.add_log_entry((f"CUDA SETUP: CUDA runtime path found: {cudart_path}"))
     cuda = get_cuda_lib_handle()
     cc = get_compute_capability(cuda)
-    cuda_setup.add_log_entry(f"CUDA SETUP: Highest compute capability among GPUs detected: {cc}")
     cuda_version_string = get_cuda_version(cuda, cudart_path)
 
+    failure = False
+    if cudart_path is None:
+        failure = True
+        cuda_setup.add_log_entry("WARNING: No libcudart.so found! Install CUDA or the cudatoolkit package (anaconda)!", is_warning=True)
+    else:
+        cuda_setup.add_log_entry((f"CUDA SETUP: CUDA runtime path found: {cudart_path}"))
 
     if cc == '' or cc is None:
-        cuda_setup.add_log_entry("WARNING: No GPU detected! Check your CUDA paths. Processing to load CPU-only library...", is_warning=True)
-        return binary_name, cudart_path, cuda, cc, cuda_version_string
+        failure = True
+        cuda_setup.add_log_entry("WARNING: No GPU detected! Check your CUDA paths. Proceeding to load CPU-only library...", is_warning=True)
+    else:
+        cuda_setup.add_log_entry(f"CUDA SETUP: Highest compute capability among GPUs detected: {cc}")
+
+    if cuda is None:
+        failure = True
+    else:
+        cuda_setup.add_log_entry(f'CUDA SETUP: Detected CUDA version {cuda_version_string}')
 
     # 7.5 is the minimum CC vor cublaslt
     has_cublaslt = cc in ["7.5", "8.0", "8.6"]
@@ -145,16 +151,13 @@ def evaluate_cuda_setup():
 
     # we use ls -l instead of nvcc to determine the cuda version
     # since most installations will have the libcudart.so installed, but not the compiler
-    cuda_setup.add_log_entry(f'CUDA SETUP: Detected CUDA version {cuda_version_string}')
 
-    def get_binary_name():
-        "if not has_cublaslt (CC < 7.5), then we have to choose  _nocublaslt.so"
-        bin_base_name = "libbitsandbytes_cuda"
-        if has_cublaslt:
-            return f"{bin_base_name}{cuda_version_string}.so"
+    if failure:
+        binary_name = "libbitsandbytes_cpu.so"
+    elif has_cublaslt:
+        binary_name = f"libbitsandbytes_cuda{cuda_version_string}.so"
     else:
-            return f"{bin_base_name}{cuda_version_string}_nocublaslt.so"
-
-    binary_name = get_binary_name()
+        "if not has_cublaslt (CC < 7.5), then we have to choose  _nocublaslt.so"
+        binary_name = f"libbitsandbytes_cuda{cuda_version_string}_nocublaslt.so"
 
     return binary_name, cudart_path, cuda, cc, cuda_version_string

setup.py

@@ -18,7 +18,7 @@ def read(fname):
 
 setup(
     name=f"bitsandbytes",
-    version=f"0.35.3",
+    version=f"0.35.4",
     author="Tim Dettmers",
     author_email="dettmers@cs.washington.edu",
     description="8-bit optimizers and matrix multiplication routines.",