Eval bug: Segmentation fault on image encoder quantization

[agNihit928]

Name and Version

version: 4644 (d774ab3)
built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu

Operating systems

Linux

GGML backends

CUDA

Hardware

Intel CPU + T4

Models

Qwen2-VL

Problem description & steps to reproduce

Getting a segmentation fault when trying the newly updated quantization support the image encoder with this output on debugging:

First Bad Commit

This is new commit happened today only, which presented a newly added feature which is causing the bug

Relevant log output

Thread 1 "llama-llava-cli" received signal SIGSEGV, Segmentation fault.
0x00007ffff7cf6720 in quantize_row_q8_0_ref ()