1580158
/
stable-diffusion-webui-distributed
mirror of https://github.com/papuSpartan/stable-diffusion-webui-distributed.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

use correct bench payload. remove unnecessary warning in debug when iteration > 1 (thought I already pushed this).

pull/17/head
unknown 2023-07-01 22:51:02 -05:00
parent 7cb452829c
commit c22af2d772
No known key found for this signature in database
GPG Key ID: CA376082283AF69A
3 changed files with 17 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
scripts/extension.py
View File

@ -159,7 +159,7 @@ class Script(scripts.Script):
try:
images: json = job.worker.response["images"]
# if we for some reason get more than we asked for
if job.batch_size < len(images):
if (job.batch_size * p.n_iter) < len(images):
logger.debug(f"Requested {job.batch_size} images from '{job.worker.uuid}', got {len(images)}")
if donor_worker is None:

14
scripts/spartan/Worker.py
View File

@ -10,13 +10,14 @@ from threading import Thread
from webui import server_name
from modules.shared import cmd_opts
import gradio as gr
from scripts.spartan.shared import benchmark_payload, logger, warmup_samples
from scripts.spartan.shared import logger, warmup_samples
from enum import Enum
import json
import base64
import queue
from modules.shared import state as master_state
from modules.api.api import encode_pil_to_base64
import scripts.spartan.shared as sh
class InvalidWorkerResponse(Exception):
@ -128,9 +129,6 @@ class Worker:
Stores the payload used to benchmark the world and certain attributes of the worker.
These things are used to draw certain conclusions after the first session.
Args:
benchmark_payload (dict): The payload used in the benchmark.
Returns:
dict: Worker info, including how it was benchmarked.
"""
@ -209,7 +207,7 @@ class Worker:
# if worker has not yet been benchmarked then
eta = (num_images / self.avg_ipm) * 60
# show effect of increased step size
real_steps_to_benched = steps / benchmark_payload['steps']
real_steps_to_benched = steps / sh.benchmark_payload['steps']
eta = eta * real_steps_to_benched
# show effect of high-res fix
@ -219,7 +217,7 @@ class Worker:
# show effect of image size
real_pix_to_benched = (payload['width'] * payload['height'])\
/ (benchmark_payload['width'] * benchmark_payload['height'])
/ (sh.benchmark_payload['width'] * sh.benchmark_payload['height'])
eta = eta * real_pix_to_benched
# show effect of using a sampler other than euler a
@ -431,7 +429,7 @@ class Worker:
float: Images per minute
"""
return benchmark_payload['batch_size'] / (seconds / 60)
return sh.benchmark_payload['batch_size'] / (seconds / 60)
results: List[float] = []
# it's seems to be lower for the first couple of generations
@ -442,7 +440,7 @@ class Worker:
self.response = None
return 0
t = Thread(target=self.request, args=(benchmark_payload, None, False,), name=f"{self.uuid}_benchmark_request")
t = Thread(target=self.request, args=(sh.benchmark_payload, None, False,), name=f"{self.uuid}_benchmark_request")
try: # if the worker is unreachable/offline then handle that here
t.start()
start = time.time()

20
scripts/spartan/World.py
View File

@ -17,7 +17,8 @@ from pathlib import Path
from modules.processing import process_images, StableDiffusionProcessingTxt2Img
import modules.shared as shared
from scripts.spartan.Worker import Worker, State
from scripts.spartan.shared import logger, warmup_samples, benchmark_payload
from scripts.spartan.shared import logger, warmup_samples
import scripts.spartan.shared as sh
class NotBenchmarked(Exception):
@ -170,7 +171,6 @@ class World:
"""
Attempts to benchmark all workers a part of the world.
"""
from scripts.spartan.shared import benchmark_payload
workers_info: dict = {}
saved: bool = os.path.exists(self.worker_info_path)
@ -186,8 +186,8 @@ class World:
if saved:
with open(self.worker_info_path, 'r') as worker_info_file:
workers_info = json.load(worker_info_file)
benchmark_payload = workers_info['benchmark_payload']
logger.info(f"Using saved benchmark config:\n{benchmark_payload}")
sh.benchmark_payload = workers_info['benchmark_payload']
logger.info(f"Using saved benchmark config:\n{sh.benchmark_payload}")
saved = False
workers = self.get_workers()
@ -200,8 +200,8 @@ class World:
with open(self.worker_info_path, 'r') as worker_info_file:
try:
workers_info = json.load(worker_info_file)
benchmark_payload = workers_info['benchmark_payload']
logger.info(f"Using saved benchmark config:\n{benchmark_payload}")
sh.benchmark_payload = workers_info['benchmark_payload']
logger.info(f"Using saved benchmark config:\n{sh.benchmark_payload}")
except json.JSONDecodeError:
logger.error(f"workers.json is not valid JSON, regenerating")
rebenchmark = True
@ -238,7 +238,7 @@ class World:
for worker in unbenched_workers:
workers_info.update(worker.info())
workers_info.update({'benchmark_payload': benchmark_payload})
workers_info.update({'benchmark_payload': sh.benchmark_payload})
# save benchmark results to workers.json
json.dump(workers_info, worker_info_file, indent=3)
@ -348,8 +348,8 @@ class World:
# wrap our benchmark payload
master_bench_payload = StableDiffusionProcessingTxt2Img()
for key in benchmark_payload:
setattr(master_bench_payload, key, benchmark_payload[key])
for key in sh.benchmark_payload:
setattr(master_bench_payload, key, sh.benchmark_payload[key])
# Keeps from trying to save the images when we don't know the path. Also, there's not really any reason to.
master_bench_payload.do_not_save_samples = True
@ -362,7 +362,7 @@ class World:
process_images(master_bench_payload)
elapsed = time.time() - start
ipm = benchmark_payload['batch_size'] / (elapsed / 60)
ipm = sh.benchmark_payload['batch_size'] / (elapsed / 60)
logger.debug(f"Master benchmark took {elapsed:.2f}: {ipm:.2f} ipm")
self.master().benchmarked = True