#SPEED PROBLEM

9 messages · Page 1 of 1 (latest)

gray crater
#

I have a problem when generating images with SDXL models. It takes around 10 minutes or more for each interaction. Is there any way to speed this up? I have an NVIDIA GeForce RTX 3070 Ti Laptop GPU.

jolly frostBOT
#

Hi @gray crater. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

There could be several reasons why you're experiencing slow image generation with SDXL models. Here are a few things you could check or try:

  1. Check your VRAM usage: High VRAM usage can slow down image generation. You can check this in your system's activity monitor. If it's too high, you might want to close other applications that are using a lot of VRAM.

  2. Adjust your cache size: If the cache size in your invoke.yaml file is too high, it can slow down generation speed. Try setting the cache to a lower value, like 0.25, and see if that helps.

  3. Check your GPU usage: Make sure that Invoke AI is using your GPU for image generation. You can check this in your system's activity monitor. If it's not, you might need to adjust your settings to ensure that Invoke AI is using your GPU.

  4. Adjust the number of steps: The number of steps performed in each generation can affect the generation time. Higher step counts typically create better images but require more generation time. You might want to try reducing the number of steps to see if that speeds up the generation process.

  5. Check for software updates: Make sure you're using the latest version of Invoke AI. There might be updates or fixes that could improve performance.

  6. Reinstall Invoke AI: If none of the above solutions work, you might want to try reinstalling Invoke AI.

Remember, SDXL models are larger and more complex than other models, so they naturally take longer to generate images. However, if the generation time is excessively long, it might be worth reaching out to Invoke AI's support for further assistance.

#
#

@gray crater kapa.ai is still learning and improving, please let us know how it did by reacting below

gray crater
#

the problem isnt when generation a simple image. The problem starts when adding control nets or ip adapters

lilac sail
#

this card has 8GB of VRAM this is the minimum for SDXL, each Controlnet or Ip adapters is to much for the vram.

sleek pecan
lilac sail
#

Get a Hardware upgrade or dont use Controlnets/ip adapters. Using SD1.5 can help with that too, because there is that enough spare vram for Controlnets/ip adapters.