llama.cpp Update b10034 Excludes MoE Kernels for Adreno A7x Devices

Core summary

One sentence to understand this update

llama.cpp update b10034 addresses an issue with Adreno A7x devices by excluding certain MoE kernels due to compiler miscompilation and weight corruption.

Impact & opportunity

What this could mean

This fix improves stability and prevents errors for users running MoE models on Adreno A7x GPUs, ensuring more reliable performance.