PR #14285 kv-cache : use ggml_set_rows

kv-cache : use ggml_set_rows #14285

ggerganov merged 7 commits into master from gg/kv-cache-use-set-rows

github-actions added ggml

Base automatically changed from gg/model-rework-out-ids to master 178 days ago

ggerganov force pushed to 5f87f289 178 days ago

ggerganov force pushed to db0cd695 178 days ago

ggerganov force pushed from a0c0fb6e 177 days ago

ggerganov force pushed to d1da9927 177 days ago

github-actions added examples

ggerganov force pushed to 14554a82 176 days ago

ggerganov marked this pull request as ready for review 176 days ago

ggerganov force pushed 176 days ago

ggerganov force pushed to e1aba6af 176 days ago

github-actions added testing

github-actions added Apple Metal

ggerganov force pushed from 5983eb1f 175 days ago

ggerganov force pushed 175 days ago

ggerganov force pushed to 36f8e20d 175 days ago

ggerganov marked this pull request as draft 171 days ago

ggerganov force pushed from 36f8e20d 170 days ago

ggerganov force pushed 170 days ago

ggerganov force pushed to 06bb08ac 170 days ago

ggerganov force pushed from 06bb08ac 168 days ago

ggerganov force pushed 168 days ago

ggerganov force pushed to 45341236 167 days ago

ggerganov marked this pull request as ready for review 166 days ago

ggerganov requested a review from

slaren 166 days ago

slaren approved these changes on 2025-07-01

github-actions added Vulkan

github-actions added SYCL

github-actions added Ascend NPU

github-actions added OpenCL

kv-cache : use ggml_set_rows

cd811b7a

graph : separate k and v indices

ac8f3474

cont : remove redundant ifs

2ac5be3a

kv-cache : improve find_slot impl

a70293bc

kv-cache : bounds-check when accessing slot_info indices

f3da97e6

kv-cache : add comments

5495ea96

ggml : add TODOs for adding GGML_OP_SET_ROWS support in the backends

30b4d4e1

ggerganov force pushed to 30b4d4e1 165 days ago

ggerganov merged a70c8a0c into master 165 days ago

ggerganov deleted the gg/kv-cache-use-set-rows branch 165 days ago

Reviewers

slaren

Assignees

No one assigned

Labels

testing Vulkan examples ggml SYCL Apple Metal Ascend NPU OpenCL

Milestone

No milestone

llama.cpp kv-cache : use ggml_set_rows #14285 Merged

kv-cache : use ggml_set_rows #14285

llama.cpp
kv-cache : use ggml_set_rows
#14285

Merged