ik_llama.cpp/github-data/pull_requests/603 - Check if MMQ should be used before using it.md

🔀 #603 - Check if MMQ should be used before using it

Author	ikawrakow
State	❌ Closed
Created	2025-07-12
Updated	2025-07-13

---

Description

In #589 I added an optimization of the fused ffn_up/gate op to not repeat the quantization of the activations when ffn_up and ffn_gate are quantized with the same type. But the check to use the direct route did not consider the possibility that some quantization types do not have MMQ implementation (e.g., IQ1_M), which then results in an assert.

This PR adds the missing check, which should fix #596