UPSTREAM PR #1180: feat: support for SDXS-512 model

[loci-dev]

Mirrored from leejet/stable-diffusion.cpp#1180

Once I wrote "three TinySD models should be enough" ( #939 ), but I changed my mind on user's request in December 2025 ( #603 ) and because SDXS-512 is so incredible fast. (That makes SDXS very handy for my sdcpp-on-android project.)

The main challenge for me was that SDXS does not use AutoEncoderKL as primary VAE (as most else) but AutoEncoderTiny, see also https://huggingface.co/IDKiro/sdxs-512-dreamshaper . I also hope that including SDXS into sd.cpp will convince @IDKiro to release into the public his SDXS-1024 one day.

[loci-review]

Explore the complete analysis inside the Version Insights

I've successfully generated a summary report for your project. The report shows a comprehensive performance analysis comparing the base version (aaf16831-eb1a-11f0-81f2-dbb430499cb5) to the target version (4e61d151-eba5-11f0-81f2-dbb430499cb5) for the stable-diffusion.cpp repository (Pull Request #12).

Key Highlights:

• Top Performance Change: The std::_Hashtable::begin function shows a dramatic 310% throughput improvement
• Mixed Results: Some functions gained throughput at the cost of response time, while others improved response time with reduced throughput
• Primary Impact Areas: Most changes affect C++ STL container operations (vectors, hashtables, red-black trees)
• Affected Binaries: Both sd-server and sd-cli show performance changes

The report includes detailed metrics for the top 10 functions by performance change, along with key observations and recommendations for further investigation.

[loci-review]

Explore the complete analysis inside the Version Insights

I've generated a comprehensive summary report for your project. The analysis shows significant performance regressions in Pull Request #12 for the stable-diffusion.cpp repository.

Key Highlights:

⚠️ Critical Issues Identified:

• Response times increased by up to 226.5% in some functions
• STL vector and iterator operations are severely impacted
• Both sd-server and sd-cli binaries are affected

🎯 Most Affected Areas:

• Vector operations (especially vector::end())
• Iterator functions
• Memory allocation routines

💡 Main Recommendation:
Consider investigating compiler optimization changes or reverting PR #12 until the root cause of these performance regressions is identified and resolved.

The report includes detailed metrics, affected components, and actionable recommendations for addressing these performance issues.