Repository Issue Activity (beta)

#1615 vulkan: vkGetDeviceQueue2 returns VK_NULL_HANDLE on older AMD drivers, first vkQueueSubmit aborts

#1611 Performance: Vulkan ggml_vk_get_op_batch_size for GGML_OP_GET_ROWS always returns 0 on Apple M2 Fedora Asahi

#1448 `ggml_conv_transpose_2d_p0` silently ignores batch dimension (ne[3] > 1)

#1609 Convert ggml to Rust

#1608 cpu: support GGML_CPU_ALL_VARIANTS with static linking

#1545 ggml-metal: a transient command-buffer failure (e.g. iOS background-GPU rejection) permanently latches the whole backend into an error state

#1427 BLAS test failure on arm64

#1566 Brew build of llama.cpp b9970+ fails due to missing definitions in ggml-org/ggml own ggml.h

#1554 libggml 0.13.1 fails in MUL_MAT_ID() on i915 (Intel comet lake) on freebsd-16

#1565 libggml isn't throwing an error if intel gpu fence timeout happens but DOES if GGML_VK_PERF_LOGGER=1

#1557 CUDA backend crashes with "an illegal memory access" on SM 8.9 (RTX 4060 Laptop) when processing ViT-style small matmul shapes

#1555 Question about requirements.txt dependencies

#1486 Windows build fails due to MLX backend (undefined symbols, no Windows build tags)

#1546 N/A

#1542 Meta: gguf-org impersonation

#1537 support prebuild

#1535 ggml-0.14.0 build fails on OpenBSD/riscv64

#1532 ggml-metal: `im2col` `_ext` kernel dispatched with base-kernel grid geometry corrupts large 1-D convs

#1530 error: no member named 'nr' in 'ggml_backend_meta_split_state'

#1529 GGML_OP_CONV_TRANSPOSE_1D is Way Too Slow at TTS Scale (CUDA/HIP/Metal)

#1506 mul_mat_id produces wrong output with K-quantized source weights

#1514 Three backward-pass gaps blocking standard ViT training: LayerNorm, GELU, conv_2d/CONT

#1491 ggml_rms_norm_back: backend-sched compute produces different (wrong) result than legacy compute_with_ctx

#1525 Recommended way to add graph-internal distributed collectives for downstream ggml projects?

#233 ggml : implement a spellcheck model (xfspell, t5-spellchecker, etc)