stable-diffusion-webui/modules/lowvram.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

166 lines
6.1 KiB
Python
Raw Normal View History

2024-06-24 01:15:46 -06:00
from collections import namedtuple
import torch
2023-08-22 09:49:08 -06:00
from modules import devices, shared
module_in_gpu = None
cpu = torch.device("cpu")
2024-06-24 01:15:46 -06:00
ModuleWithParent = namedtuple('ModuleWithParent', ['module', 'parent'], defaults=['None'])
def send_everything_to_cpu():
global module_in_gpu
if module_in_gpu is not None:
module_in_gpu.to(cpu)
module_in_gpu = None
2023-08-22 09:49:08 -06:00
def is_needed(sd_model):
return shared.cmd_opts.lowvram or shared.cmd_opts.medvram or shared.cmd_opts.medvram_sdxl and hasattr(sd_model, 'conditioner')
def apply(sd_model):
enable = is_needed(sd_model)
shared.parallel_processing_allowed = not enable
if enable:
setup_for_low_vram(sd_model, not shared.cmd_opts.lowvram)
else:
sd_model.lowvram = False
def setup_for_low_vram(sd_model, use_medvram):
if getattr(sd_model, 'lowvram', False):
return
sd_model.lowvram = True
parents = {}
def send_me_to_gpu(module, _):
"""send this module to GPU; send whatever tracked module was previous in GPU to CPU;
we add this as forward_pre_hook to a lot of modules and this way all but one of them will
be in CPU
"""
global module_in_gpu
module = parents.get(module, module)
if module_in_gpu == module:
return
if module_in_gpu is not None:
module_in_gpu.to(cpu)
module.to(devices.device)
module_in_gpu = module
# see below for register_forward_pre_hook;
# first_stage_model does not use forward(), it uses encode/decode, so register_forward_pre_hook is
# useless here, and we just replace those methods
first_stage_model = sd_model.first_stage_model
first_stage_model_encode = sd_model.first_stage_model.encode
first_stage_model_decode = sd_model.first_stage_model.decode
def first_stage_model_encode_wrap(x):
send_me_to_gpu(first_stage_model, None)
return first_stage_model_encode(x)
def first_stage_model_decode_wrap(z):
send_me_to_gpu(first_stage_model, None)
return first_stage_model_decode(z)
2023-07-12 14:52:43 -06:00
to_remain_in_cpu = [
(sd_model, 'first_stage_model'),
(sd_model, 'depth_model'),
(sd_model, 'embedder'),
(sd_model, 'model'),
]
is_sdxl = hasattr(sd_model, 'conditioner')
is_sd2 = not is_sdxl and hasattr(sd_model.cond_stage_model, 'model')
2024-06-24 01:15:46 -06:00
if hasattr(sd_model, 'medvram_fields'):
to_remain_in_cpu = sd_model.medvram_fields()
elif is_sdxl:
2023-07-12 14:52:43 -06:00
to_remain_in_cpu.append((sd_model, 'conditioner'))
elif is_sd2:
to_remain_in_cpu.append((sd_model.cond_stage_model, 'model'))
else:
to_remain_in_cpu.append((sd_model.cond_stage_model, 'transformer'))
# remove several big modules: cond, first_stage, depth/embedder (if applicable), and unet from the model
stored = []
for obj, field in to_remain_in_cpu:
module = getattr(obj, field, None)
stored.append(module)
setattr(obj, field, None)
# send the model to GPU.
sd_model.to(devices.device)
2023-07-12 14:52:43 -06:00
# put modules back. the modules will be in CPU.
for (obj, field), module in zip(to_remain_in_cpu, stored):
setattr(obj, field, module)
# register hooks for those the first three models
2024-06-26 22:35:53 -06:00
if hasattr(sd_model, "cond_stage_model") and hasattr(sd_model.cond_stage_model, "medvram_modules"):
2024-06-24 01:15:46 -06:00
for module in sd_model.cond_stage_model.medvram_modules():
if isinstance(module, ModuleWithParent):
parent = module.parent
module = module.module
else:
parent = None
if module:
module.register_forward_pre_hook(send_me_to_gpu)
if parent:
parents[module] = parent
elif is_sdxl:
2023-07-12 14:52:43 -06:00
sd_model.conditioner.register_forward_pre_hook(send_me_to_gpu)
elif is_sd2:
sd_model.cond_stage_model.model.register_forward_pre_hook(send_me_to_gpu)
sd_model.cond_stage_model.model.token_embedding.register_forward_pre_hook(send_me_to_gpu)
parents[sd_model.cond_stage_model.model] = sd_model.cond_stage_model
parents[sd_model.cond_stage_model.model.token_embedding] = sd_model.cond_stage_model
2023-07-12 14:52:43 -06:00
else:
sd_model.cond_stage_model.transformer.register_forward_pre_hook(send_me_to_gpu)
parents[sd_model.cond_stage_model.transformer] = sd_model.cond_stage_model
2023-07-12 14:52:43 -06:00
sd_model.first_stage_model.register_forward_pre_hook(send_me_to_gpu)
sd_model.first_stage_model.encode = first_stage_model_encode_wrap
sd_model.first_stage_model.decode = first_stage_model_decode_wrap
2024-06-26 22:35:53 -06:00
if getattr(sd_model, 'depth_model', None) is not None:
sd_model.depth_model.register_forward_pre_hook(send_me_to_gpu)
2024-06-26 22:35:53 -06:00
if getattr(sd_model, 'embedder', None) is not None:
sd_model.embedder.register_forward_pre_hook(send_me_to_gpu)
2023-07-14 00:56:01 -06:00
if use_medvram:
sd_model.model.register_forward_pre_hook(send_me_to_gpu)
else:
diff_model = sd_model.model.diffusion_model
# the third remaining model is still too big for 4 GB, so we also do the same for its submodules
# so that only one of them is in GPU at a time
stored = diff_model.input_blocks, diff_model.middle_block, diff_model.output_blocks, diff_model.time_embed
diff_model.input_blocks, diff_model.middle_block, diff_model.output_blocks, diff_model.time_embed = None, None, None, None
sd_model.model.to(devices.device)
diff_model.input_blocks, diff_model.middle_block, diff_model.output_blocks, diff_model.time_embed = stored
# install hooks for bits of third model
diff_model.time_embed.register_forward_pre_hook(send_me_to_gpu)
for block in diff_model.input_blocks:
block.register_forward_pre_hook(send_me_to_gpu)
diff_model.middle_block.register_forward_pre_hook(send_me_to_gpu)
for block in diff_model.output_blocks:
block.register_forward_pre_hook(send_me_to_gpu)
def is_enabled(sd_model):
2023-08-22 09:49:08 -06:00
return sd_model.lowvram