UPSTREAM PR #18469: lora: count lora nodes in graph_max_nodes

[loci-dev]

Mirrored from ggml-org/llama.cpp#18469

Fix ggml-org/llama.cpp#18050

Fix ggml-org/llama.cpp#18466

Supersede ggml-org/llama.cpp#18375

[loci-review]

Explore the complete analysis inside the Version Insights

I've successfully retrieved the performance summary report for your llama.cpp project (Pull Request #748). Here are the key highlights:

Summary

The analysis compares two versions of the llama.cpp library and shows mixed performance results:

Major Improvements ⬆️

• __iter_equals_val: +316.54% throughput increase
• begin() function: +289.33% throughput increase
• operator= (unique_ptr): +97.86% throughput increase

Notable Regressions ⬇️

• end() functions: -75.41% throughput decrease
• swap: -51.02% throughput decrease
• build_rope_shift: -9.64% throughput decrease

The report suggests that while certain STL iterator and vector operations have been significantly optimized, some other operations have experienced performance degradation. The recommendations include investigating the regressions and profiling real-world workloads to ensure overall application performance improvements.

Would you like me to provide more details about any specific function or aspect of this performance analysis?

[loci-review]

Explore the complete analysis inside the Version Insights

I've successfully retrieved the summary report for your project. The report shows a performance comparison for the llama.cpp repository (pull request #748) between two versions.

Key Highlights:

⚠️ Critical Performance Issues Detected:

• The iterator operator- function shows an 82.73% increase in response time
• Multiple STL container operations show significant performance degradations (43-69%)
• All top 10 functions show performance regressions

Most Affected Areas:

1. Iterator operations
2. Vector operations (end functions)
3. Memory management (swap, unique_ptr)
4. Tree-based container operations

Recommendation: This pull request introduces significant performance regressions in core STL operations and may require review before merging.