fatal: Memory allocation failure

Begin to load 1 model
[Unload] Trying to free 11766.16 MB for cuda:0 with 0 models keep loaded ...
[Unload] Unload model IntegratedAutoencoderKL
[Memory Management] Current Free GPU Memory: 10827.83 MB
[Memory Management] Required Model Memory: 4897.05 MB
[Memory Management] Required Inference Memory: 1024.00 MB
[Memory Management] Estimated Remaining GPU Memory: 4906.78 MB
Moving model(s) has taken 0.64 seconds
100%|██████████████████████████████████████████████████████████████████████████████████| 35/35 [00:51<00:00,  1.46s/it]
To load target model IntegratedAutoencoderKL███████████████████████████████████████████| 35/35 [00:57<00:00,  1.46s/it]
Begin to load 1 model
[Unload] Trying to free 18153.59 MB for cuda:0 with 0 models keep loaded ...
[Unload] Unload model KModel
[Memory Management] Current Free GPU Memory: 10776.85 MB
[Memory Management] Required Model Memory: 159.56 MB
[Memory Management] Required Inference Memory: 1024.00 MB
[Memory Management] Estimated Remaining GPU Memory: 9593.29 MB
Moving model(s) has taken 1.25 seconds
fatal   : Memory allocation failure
Warning: Ran out of memory when regular VAE decoding, retrying with tiled VAE decoding.
  0%|                                                                                            | 0/1 [00:00<?, ?it/s]Press any key to continue . . .

Started happening after recent commit if im not mistaken. https://github.com/lllyasviel/stable-diffusion-webui-forge/commit/4e3c78178a7eae73311fb269cdeaf7a0b40644d1 Happens when i upscale images with high res fix. I get python error window or chrome tab crashing.

RTX 4070 Super 12GB VRAM
32GB DDR5 Memory
Windows 11

Start by comparing commit 4e3c78178a7eae73311fb269cdeaf7a0b40644d1 with its parent, then reproduce high-res fix upscaling using the reported RTX 4070 Super and 12GB VRAM setup. Trace the regular and tiled VAE decoding path around the memory allocation failure; done means the workflow completes without the Python or browser crash.