convert : refactor rope scaling handling (ggml-org#18013) * refactor rope scaling handling * ws-- * missed a couple * use find_hparam

llama_context: synchronize before reallocating output buffer (ggml-or……g#17974)

ggml-alloc : fix reuse-parent logic for misaligned sizes (ggml-org#17884)

cuda : add missing support check for xielu (ggml-org#17895) 

metal: SSM kernel improvements (ggml-org#17876) * feat: Add a batched version of ssm_conv This was done using Claude Code. It found a number of optimizations around how the threads were organized, resulting in a huge performance boost! Branch: Mamba2SSD Signed-off-by: Gabe Goodhart <[email protected]> * feat: Optimized SSM_SCAN kernel for metal This used Claude Code and resulted in a modest performance improvement while maintaining correctness. Branch: Mamba2SSD Signed-off-by: Gabe Goodhart <[email protected]> * test: Add test-backend-ops perf tests for SSM_CONV Branch: SSMKernelImprovements Signed-off-by: Gabe Goodhart <[email protected]> * test: Real representitive tests for SSM_CONV Branch: SSMKernelImprovements Signed-off-by: Gabe Goodhart <[email protected]> * refactor: Use function constant for ssm_conv batch size Branch: SSMKernelImprovements Signed-off-by: Gabe Goodhart <[email protected]> * test: backend op tests for ssm_scan from granite4 1b-h Branch: SSMKernelImprovements Signed-off-by: Gabe Goodhart <[email protected]> * style: remove commented out templates Branch: SSMKernelImprovements Signed-off-by: Gabe Goodhart <[email protected]> * feat: float4 version of ssm_conv_batched Branch: SSMKernelImprovements Signed-off-by: Gabe Goodhart <[email protected]> * fix: Add missing ggml_metal_cv_free Signed-off-by: Gabe Goodhart <[email protected]> Co-authored-by: Georgi Gerganov <[email protected]> --------- Signed-off-by: Gabe Goodhart <[email protected]> Co-authored-by: Georgi Gerganov <[email protected]>

model-conversion : add token ids to prompt token output [no ci] (ggml……-org#17863) This commit adds the token ids to the printed prompt outputs. The motivation for this is that is can be useful to see the actual token ids alongside the token strings for debugging.

common : change --color to accept on/off/auto, default to auto (ggml-……org#17827)

ci : transform release binary root dir in tar to llama-bXXXX (ggml-or……g#17773) * transform release binary root dir in tar to llama-bXXXX * bsdtar supports -s instead of --transform

ggml-cpu: remove duplicate conditional check 'iid' (ggml-org#17650) 

vulkan: Reduce temporary memory usage for TOP_K (ggml-org#17623) - Compute row size for the temp buffer based on the output of the first pass. - Update shader addressing math to use the output row size - Pass the output row size as "ncols_output", what used to be "ncols_output" is now "k" For the common case of K=40 and src0=(200000,1,1,1), this reduces the temporary buffer from about 3.2MB to 500KB.