2023-02-01 14:28:16 +00:00
import sys
2022-10-04 09:32:22 +00:00
import contextlib
2023-05-21 18:55:14 +00:00
from functools import lru_cache
2022-09-11 05:11:27 +00:00
import torch
2023-08-09 07:25:35 +00:00
from modules import errors , shared
2023-02-01 14:28:16 +00:00
if sys . platform == " darwin " :
from modules import mac_specific
2022-09-12 13:34:13 +00:00
2022-11-12 07:00:49 +00:00
2022-11-12 03:02:40 +00:00
def has_mps ( ) - > bool :
2023-02-01 14:28:16 +00:00
if sys . platform != " darwin " :
2022-11-12 03:02:40 +00:00
return False
2023-02-01 14:28:16 +00:00
else :
return mac_specific . has_mps
2022-09-11 15:48:36 +00:00
2022-11-12 07:00:49 +00:00
2022-11-27 10:08:54 +00:00
def get_cuda_device_string ( ) :
if shared . cmd_opts . device_id is not None :
return f " cuda: { shared . cmd_opts . device_id } "
2022-10-22 11:04:14 +00:00
2022-11-27 10:08:54 +00:00
return " cuda "
2022-10-22 11:04:14 +00:00
2022-11-27 10:08:54 +00:00
2023-01-27 08:28:12 +00:00
def get_optimal_device_name ( ) :
2022-11-27 10:08:54 +00:00
if torch . cuda . is_available ( ) :
2023-01-27 08:28:12 +00:00
return get_cuda_device_string ( )
2022-09-11 15:48:36 +00:00
2022-11-12 03:02:40 +00:00
if has_mps ( ) :
2023-01-27 08:28:12 +00:00
return " mps "
return " cpu "
2022-09-11 15:48:36 +00:00
2023-01-27 08:28:12 +00:00
def get_optimal_device ( ) :
return torch . device ( get_optimal_device_name ( ) )
2022-09-11 20:24:24 +00:00
2022-12-03 15:06:33 +00:00
def get_device_for ( task ) :
if task in shared . cmd_opts . use_cpu :
return cpu
return get_optimal_device ( )
2022-09-11 20:24:24 +00:00
def torch_gc ( ) :
2023-07-08 14:13:18 +00:00
2022-09-11 20:24:24 +00:00
if torch . cuda . is_available ( ) :
2022-11-27 10:08:54 +00:00
with torch . cuda . device ( get_cuda_device_string ( ) ) :
2022-11-26 23:25:16 +00:00
torch . cuda . empty_cache ( )
torch . cuda . ipc_collect ( )
2023-07-10 18:18:34 +00:00
if has_mps ( ) :
mac_specific . torch_mps_gc ( )
2022-09-12 13:34:13 +00:00
def enable_tf32 ( ) :
if torch . cuda . is_available ( ) :
2022-12-03 13:01:23 +00:00
# enabling benchmark option seems to enable a range of cards to do fp16 when they otherwise can't
# see https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/4407
2023-05-10 08:05:02 +00:00
if any ( torch . cuda . get_device_capability ( devid ) == ( 7 , 5 ) for devid in range ( 0 , torch . cuda . device_count ( ) ) ) :
2022-11-08 02:06:48 +00:00
torch . backends . cudnn . benchmark = True
2022-12-03 12:57:52 +00:00
2022-09-12 13:34:13 +00:00
torch . backends . cuda . matmul . allow_tf32 = True
torch . backends . cudnn . allow_tf32 = True
errors . run ( enable_tf32 , " Enabling TF32 " )
2022-09-12 17:09:32 +00:00
2023-08-03 04:18:55 +00:00
cpu : torch . device = torch . device ( " cpu " )
device : torch . device = None
device_interrogate : torch . device = None
device_gfpgan : torch . device = None
device_esrgan : torch . device = None
device_codeformer : torch . device = None
dtype : torch . dtype = torch . float16
dtype_vae : torch . dtype = torch . float16
dtype_unet : torch . dtype = torch . float16
2023-01-25 04:51:45 +00:00
unet_needs_upcast = False
2022-09-12 17:09:32 +00:00
2022-11-12 07:00:49 +00:00
2023-01-27 15:19:43 +00:00
def cond_cast_unet ( input ) :
return input . to ( dtype_unet ) if unet_needs_upcast else input
def cond_cast_float ( input ) :
return input . float ( ) if unet_needs_upcast else input
2023-08-02 21:00:23 +00:00
nv_rng = None
2022-10-10 13:11:14 +00:00
def autocast ( disable = False ) :
if disable :
return contextlib . nullcontext ( )
2022-10-04 09:32:22 +00:00
if dtype == torch . float32 or shared . cmd_opts . precision == " full " :
return contextlib . nullcontext ( )
return torch . autocast ( " cuda " )
2022-10-25 06:01:57 +00:00
2022-11-12 07:00:49 +00:00
2023-01-25 05:23:10 +00:00
def without_autocast ( disable = False ) :
return torch . autocast ( " cuda " , enabled = False ) if torch . is_autocast_enabled ( ) and not disable else contextlib . nullcontext ( )
2023-01-16 19:59:46 +00:00
class NansException ( Exception ) :
pass
def test_for_nans ( x , where ) :
2023-01-17 08:04:56 +00:00
if shared . cmd_opts . disable_nan_check :
return
2023-01-16 19:59:46 +00:00
if not torch . all ( torch . isnan ( x ) ) . item ( ) :
return
if where == " unet " :
message = " A tensor with all NaNs was produced in Unet. "
if not shared . cmd_opts . no_half :
2023-01-25 05:23:10 +00:00
message + = " This could be either because there ' s not enough precision to represent the picture, or because your video card does not support half type. Try setting the \" Upcast cross attention layer to float32 \" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. "
2023-01-16 19:59:46 +00:00
elif where == " vae " :
message = " A tensor with all NaNs was produced in VAE. "
if not shared . cmd_opts . no_half and not shared . cmd_opts . no_half_vae :
message + = " This could be because there ' s not enough precision to represent the picture. Try adding --no-half-vae commandline argument to fix this. "
else :
message = " A tensor with all NaNs was produced. "
2023-01-27 10:08:00 +00:00
message + = " Use --disable-nan-check commandline argument to disable this check. "
2023-01-16 19:59:46 +00:00
raise NansException ( message )
2023-05-21 18:55:14 +00:00
@lru_cache
def first_time_calculation ( ) :
"""
just do any calculation with pytorch layers - the first time this is done it allocaltes about 700 MB of memory and
spends about 2.7 seconds doing that , at least wih NVidia .
"""
x = torch . zeros ( ( 1 , 1 ) ) . to ( device , dtype )
linear = torch . nn . Linear ( 1 , 1 ) . to ( device , dtype )
linear ( x )
x = torch . zeros ( ( 1 , 1 , 3 , 3 ) ) . to ( device , dtype )
conv2d = torch . nn . Conv2d ( 1 , 1 , ( 3 , 3 ) ) . to ( device , dtype )
conv2d ( x )
2023-08-09 05:43:31 +00:00