UPSTREAM PR #1187: Update ggml: fixed CUDA large memcpy assertion

[loci-dev]

Mirrored from leejet/stable-diffusion.cpp#1187

fix #1106

[loci-agentic-ai]

Explore the complete analysis inside the Version Insights

I've successfully retrieved the summary report for your project. The analysis shows that Pull Request #16 for the stable-diffusion.cpp repository introduces significant performance improvements, particularly in C++ STL container operations.

Key Highlights:

• Major improvements in tree/iterator operations with response time increases of 214-218% and throughput increases of 284-289%
• Vector operations show mixed but generally positive results
• Top performer: The begin function shows a 289% throughput increase
• Areas to watch: The end() function shows decreased throughput despite improved response time

The report includes detailed metrics for the top 10 functions by response time change, comparing the base version (7f8d5301) with the target version (6debeeb1) across both the sd-cli and sd-server binaries.