sync : ggml#3125

Merged

ggerganov merged 10 commits intomasterfrom

sync-ggml-25-05-07

May 7, 2025

Commits on May 7, 2025

vulkan : kernels for depthwise 2D convolution (CONV_2D_DW) (ggml/1204)

Acly
authored and
ggerganov
committed
vulkan: Handle src1 batch dimension in non-contiguous mat-vec-mul shader (llama/13191)

jeffbolznv
authored and
ggerganov
committed
vulkan: Add bfloat16 support (llama/12554)

jeffbolznv
authored and
ggerganov
committed
build : fix build info on windows (llama/13239)

slaren
authored and
ggerganov
committed
ggml: Don't assert fail when tensor data changes (llama/13222)

jessegross
authored and
ggerganov
committed
rpc : avoid uninitialized memory in serialize_tensor (llama/13210)

justinsb
authored and
ggerganov
committed
ggml : Enable MMA for BF16 in llamafile_sgemm (llama/13148)

shalinib-ibm
authored and
ggerganov
committed
vulkan : fix lint (llama/0)
ggerganov
committed
sync : ggml
ggerganov
committed
cli : avoid std::exchange
ggerganov
committed