Eval bug: Gemma MoE model crashing

[Offset0x]

Name and Version

version: 9459 (07ac3ce)
built with Clang 19.1.5 for Windows x86_64 (latest version)

Operating systems

Windows

GGML backends

CUDA

Hardware

RTX 5060 Ti

Models

No response

Problem description & steps to reproduce

D:/a/beellama.cpp/beellama.cpp/ggml/src/ggml-backend.cpp:272: GGML_ASSERT(offset + size <= ggml_nbytes(tensor) && "tensor read out of bounds") failed

In the previous version it was working, and I was getting a 1.5x+ speed up on the MoE model too, but now it seems only dense is working.

First Bad Commit

No response

Relevant log output

Logs

[Anbeeld]

Just pushed pre-release WIP version of v0.3.0: https://github.com/Anbeeld/beellama.cpp/tree/v0.3.0

The branch contains a full rebase to the latest llama.cpp (with MTP and upstream spec architecture), but also a lot of DFlash fixes (multi-slot, multi-GPU, speculation in tool calls, etc.)

Please build and try that branch, and report back on what's the state of the issue there. All the follow up fixes will be on v0.3.0 too, as it's a bit pointless to do patches within v0.2.0 because of upstream rebasing.