Search results

LLVM 22 hangs on building libggml with GGML_CPU_ALL_VARIANTS

Status: Open.

#1499 In ggml-org/ggml;

· catap opened on May 19, 2026

ggml-cuda: missing __syncthreads() in ssm-scan.cu

Status: Open.

#1498 In ggml-org/ggml;

· FrankZhuQS opened on May 19, 2026

ggml-cuda: cpy ignores destination strides (writes contiguously into a strided view_2d slot)

Status: Open.

#1497 In ggml-org/ggml;

· OriPekelman opened on May 18, 2026

ggml-cuda: add ggml_backend_cuda_buffer_from_ptr for UVA-direct mmap inference

Status: Open.

#1495 In ggml-org/ggml;

· OriPekelman opened on May 18, 2026

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

Status: Open.

#1491 In ggml-org/ggml;

· OriPekelman opened on May 14, 2026

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

Status: Open.

#1486 In ggml-org/ggml;

· SkybuckFlying opened on May 12, 2026

Refactor large source files into smaller, modular units

Status: Open.

#1482 In ggml-org/ggml;

· Zabis13 opened on May 10, 2026

cmake: include path for ggml.h not exported if GGML_BACKEND_DL is set

Status: Open.

#1467 In ggml-org/ggml;

· SchaichAlonso opened on Apr 28, 2026

ggml-cuda: flash-attn MMA picker has no Blackwell (sm_120) entry — silently uses Ampere config

Status: Open.

#1466 In ggml-org/ggml;

· Zbig9000 opened on Apr 27, 2026

ggml-hexagon: HAP_power_set_HMX uses &ctx instead of ctx in htp_iface_open(), causing large HMX GEMM slowdown

Status: Open.

#1452 In ggml-org/ggml;

· happyyzy opened on Mar 31, 2026

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

Status: Open.

#1448 In ggml-org/ggml;

· janblumenkamp opened on Mar 30, 2026

Try to return a local variable in mnist_model_init_from_file

Status: Open.

#1447 In ggml-org/ggml;

· vtpx opened on Mar 28, 2026