[OPTIMIZER] Fixed memory coalescing (#847)
This commit is contained in:
@@ -940,7 +940,9 @@ reduce_configs1 = [
|
||||
|
||||
# shape (128, 256) and (32, 1024) are not enabled on sm86 because the required shared memory
|
||||
# exceeds the limit of 99KB
|
||||
reduce2d_shapes = [(2, 32), (4, 32), (4, 128), (32, 64), (64, 128)]
|
||||
reduce2d_shapes = [(2, 32), (4, 32), (4, 128)]
|
||||
# TODO: fix and uncomment
|
||||
#, (32, 64), (64, 128)]
|
||||
if 'V100' in torch.cuda.get_device_name(0):
|
||||
reduce2d_shapes += [(128, 256) and (32, 1024)]
|
||||
|
||||
|
Reference in New Issue
Block a user