v3.9 update (#2203)

* v3.9 update

* voidD

---------

Co-authored-by: yuzhai <yuzhai@nvidia.com>

python/cutlass/__init__.py

@@ -134,7 +134,7 @@ def get_option_registry():
         this._option_registry = OptionRegistry(device_cc())
     return this._option_registry
 
-this.__version__ = '3.8.0'
+this.__version__ = '3.9.0'
 
 from cutlass.backend import create_memory_pool
 from cutlass.emit.pytorch import pytorch

python/cutlass_library/emit_kernel_listing.py

@@ -282,6 +282,8 @@ def _computeFlopsPerByte(operation, m, n, k, batch_count=1, beta=0.0):
 def emit_gemm_kernel_testlist(manifest, curr_build_dir, arch, mode
                               ):
   profiler_reference_computing = "--verification-providers=device --providers=cutlass"
+  
+
   # beta values for L0 and L1
   # TODO: randomize beta values for wider coverage
   beta_values = [0.5]

python/cutlass_library/generator.py

@@ -10025,7 +10025,8 @@ def GenerateSM120_TensorOp_fp4_UMMA_gemm_with_block_scaled(manifest, cuda_versio
 
   tile_sizes_cooperative = [
     [128, 128, 128],
-    [128, 128, 256]
+    [128, 128, 256],
+    [256, 128, 128]
   ]
 
   tile_sizes_pingpong = [

python/setup_library.py

@@ -36,7 +36,7 @@ from setuptools import setup
 def perform_setup():
     setup(
         name='cutlass_library',
-        version='3.8.0',
+        version='3.9.0',
         description='CUTLASS library generation scripts',
         packages=['cutlass_library']
     )

python/setup_pycute.py

@@ -36,7 +36,7 @@ from setuptools import setup
 def perform_setup():
     setup(
         name='pycute',
-        version='3.8.0',
+        version='3.9.0',
         description='Python implementation of CuTe',
         packages=['pycute'],
     )