Multiple profiles in one checkpoint?

[nvidia]

Multiple profiles in one checkpoint?

[contentis]

Yes, the current logic selects engines on

1. How close is the requested resolution to the optimal resolution of an engine
2. If multiple engines have the same score, select the one with the least VRAM consumption.

[Poukpalaova]

I've created multiple Profile for a SD 1.5, but when I change the resolution to match another profile, in the console I still see it take the Profile 0.
How can I switch between profile?
I have a 4090 video card with latest drivers (545.84)
The first picture asked, was batch 4, 512x512, 1 word in positive prompt
And in the 2 picture asked, same as 1 but with an upscale to x2 that lead to a 1024x1024 that should have take my profile 3:

Loading TensorRT engine: F:\IMAGES\AIgenerated\SDXL 1.6\stable-diffusion-webui\models\Unet-trt\photon_v1_d902a082_cc89_sample=1x4x64x64+2x4x64x64+8x4x96x96-timesteps=1+2+8-encoder_hidden_states=1x77x768+2x77x768+8x154x768.trt
[I] Loading bytes from F:\IMAGES\AIgenerated\SDXL 1.6\stable-diffusion-webui\models\Unet-trt\photon_v1_d902a082_cc89_sample=1x4x64x64+2x4x64x64+8x4x96x96-timesteps=1+2+8-encoder_hidden_states=1x77x768+2x77x768+8x154x768.trt
Profile 0:
sample = [(1, 4, 64, 64), (2, 4, 64, 64), (8, 4, 96, 96)]
timesteps = [(1,), (2,), (8,)]
encoder_hidden_states = [(1, 77, 768), (2, 77, 768), (8, 154, 768)]
latent = [(-1945914624), (-1945909504), (-1945910272)]

100%|█████████████████████████████████████████████████████████████████████████████████| 20/20 [00:04<00:00, 4.79it/s]
Total progress: 100%|█████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 10.12it/s]
100%|█████████████████████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 13.27it/s]
Total progress: 100%|█████████████████████████████████████████████████████████████████| 20/20 [00:02<00:00, 9.66it/s]
100%|█████████████████████████████████████████████████████████████████████████████████| 20/20 [00:01<00:00, 12.42it/s]
0%| | 0/20 [00:00<?, ?it/s]Loading TensorRT engine: F:\IMAGES\AIgenerated\SDXL 1.6\stable-diffusion-webui\models\Unet-trt\photon_v1_d902a082_cc89_sample=1x4x64x64+2x4x128x128+8x4x128x128-timesteps=1+2+8-encoder_hidden_states=1x77x768+2x77x768+8x154x768.trt
[I] Loading bytes from F:\IMAGES\AIgenerated\SDXL 1.6\stable-diffusion-webui\models\Unet-trt\photon_v1_d902a082_cc89_sample=1x4x64x64+2x4x128x128+8x4x128x128-timesteps=1+2+8-encoder_hidden_states=1x77x768+2x77x768+8x154x768.trt
Profile 0:
sample = [(1, 4, 64, 64), (2, 4, 128, 128), (8, 4, 128, 128)]
timesteps = [(1,), (2,), (8,)]
encoder_hidden_states = [(1, 77, 768), (2, 77, 768), (8, 154, 768)]
latent = [(-1945966336), (-1945972992), (-1945967872)]

100%|█████████████████████████████████████████████████████████████████████████████████| 20/20 [00:14<00:00, 1.41it/s]
Total progress: 100%|█████████████████████████████████████████████████████████████████| 40/40 [00:23<00:00, 1.69it/s]
Total progress: 100%|█████████████████████████████████████████████████████████████████| 40/40 [00:23<00:00, 1.76it/s]

[carnivar10]

Thanks for the clarification, If the extension automatically changes the optimal resolution, is it possible to add the ability to set multiple resolutions and generate them at once when using the 'Export Engine'? I think that would be really useful.

[contentis]

This has been asked in #60 as well. Currently, there is no support to export multiple checkpoints or profiles at once.