LLM inference in C/C++
C++ ggml
ci : add older, min and dry-run options to ccache-clear (#27504)
Georgi Gerganov3a653fea932e
diff --git a/.github/actions/ccache-clear/action.yml b/.github/actions/ccache-clear/action.ymlindex d20f6bf848..2420045edf 100644--- a/.github/actions/ccache-clear/action.yml+++ b/.github/actions/ccache-clear/action.yml@@ -1,23 +1,88 @@# note: place this as the last step of the job, so the new cache is saved by "Post ccache" right after the old one is clearedname: "ccache-clear"-description: "Delete all GitHub Actions caches matching a key prefix"+description: "Delete GitHub Actions caches matching a key prefix, oldest first"inputs:key:description: "Cache key prefix to match and delete"required: true+ older:+ description: "Only delete caches created more than this long ago (e.g. 90m, 1h, 1d). By default all matching caches are deleted"+ required: false+ default: ""+ min:+ description: "Stop deleting if fewer than this many caches would remain (e.g. 1). By default there is no minimum"+ required: false+ default: "0"+ dry-run:+ description: "Only print the caches that would be deleted, without deleting them"+ required: false+ default: "false"runs:using: "composite"steps:- name: Clear cachesshell: bash+ env:+ CLEAR_KEY: ${{ inputs.key }}+ CLEAR_OLDER: ${{ inputs.older }}+ CLEAR_MIN: ${{ inputs.min }}+ CLEAR_DRY_RUN: ${{ inputs.dry-run }}run: |- CACHES=$(gh cache list --key "ccache-${{ inputs.key }}" --json id,key --jq '.[] | "\(.id) \(.key)"' 2>/dev/null)+ # Convert a duration (e.g. 90m, 1h, 1d, plain seconds) to seconds+ to_seconds() {+ local val="$1"+ [[ "$val" =~ ^[0-9]+$ ]] && { echo "$val"; return 0; }+ local num="${val%?}" unit="${val: -1}" mult+ [[ "$num" =~ ^[0-9]+$ ]] || return 1+ case "$unit" in+ s) mult=1 ;;+ m) mult=60 ;;+ h) mult=3600 ;;+ d) mult=86400 ;;+ *) return 1 ;;+ esac+ echo $((num * mult))+ }++ [[ "$CLEAR_MIN" =~ ^[0-9]+$ ]] || { echo "Invalid min value: $CLEAR_MIN" >&2; exit 1; }+ [[ "$CLEAR_DRY_RUN" =~ ^(true|false)$ ]] || { echo "Invalid dry-run value: $CLEAR_DRY_RUN" >&2; exit 1; }++ CACHES=$(gh cache list --key "ccache-$CLEAR_KEY" --json id,key,createdAt --jq '.[] | [.createdAt, .id, .key] | @tsv' 2>/dev/null | LC_ALL=C sort)if [ -z "$CACHES" ]; then- echo "No caches found with key prefix: ${{ inputs.key }}"+ echo "No caches found with key prefix: $CLEAR_KEY"exit 0fi- while read -r id key; do- echo "Deleting cache: $id ($key)"- gh cache delete "$id"++ TOTAL=$(( $(wc -l <<< "$CACHES") ))++ echo "Found $TOTAL cache(s) with key prefix: $CLEAR_KEY (oldest first):"+ while IFS=$'\t' read -r CREATED ID KEY; do+ printf ' %s %s %s\n' "$CREATED" "$ID" "$KEY"+ done <<< "$CACHES"++ CUTOFF=""+ if [ -n "$CLEAR_OLDER" ]; then+ OLDER_SECONDS=$(to_seconds "$CLEAR_OLDER") || { echo "Invalid older value: $CLEAR_OLDER (expected e.g. 90m, 1h, 1d)" >&2; exit 1; }+ CUTOFF=$(( $(date +%s) - OLDER_SECONDS ))+ fi++ # Caches are sorted oldest first+ DELETED=0+ while IFS=$'\t' read -r CREATED ID KEY; do+ if [ -n "$CUTOFF" ] && [ "$(date -d "$CREATED" +%s)" -ge "$CUTOFF" ]; then+ echo "Rest are not older than $CLEAR_OLDER, stopping"+ break+ fi+ if [ $((TOTAL - DELETED - 1)) -lt "$CLEAR_MIN" ]; then+ echo "Keeping at least $CLEAR_MIN cache(s), stopping"+ break+ fi+ if [ "$CLEAR_DRY_RUN" = "true" ]; then+ echo "Would delete cache: $ID ($KEY)"+ else+ echo "Deleting cache: $ID ($KEY)"+ gh cache delete "$ID"+ fi+ DELETED=$((DELETED + 1))done <<< "$CACHES"diff --git a/.github/workflows/build-cpu.yml b/.github/workflows/build-cpu.ymlindex ffced86c6a..f39304cabf 100644--- a/.github/workflows/build-cpu.yml+++ b/.github/workflows/build-cpu.yml@@ -117,6 +117,18 @@ jobs:./bin/llama-convert-llama2c-to-ggml --copy-vocab-from-model ./tok512.bin --llama2c-model stories260K.bin --llama2c-output-model stories260K.gguf./bin/llama-completion -m stories260K.gguf -p "One day, Lily met a Shoggoth" -n 500 -c 256+ # note: real deletion only on push to master (same condition as the ccache save),+ # dry-run otherwise (the token is read-only on PRs from forks)+ - name: ccache-clear+ uses: ./.github/actions/ccache-clear+ env:+ GH_TOKEN: ${{ github.token }}+ with:+ key: cpu-${{ matrix.os }}+ older: 1h+ min: 1+ dry-run: ${{ github.event_name != 'push' || github.ref != 'refs/heads/master' }}+windows:name: windows / ${{ matrix.build }}runs-on: windows-2025diff --git a/.pi/gg/SYSTEM.md b/.pi/gg/SYSTEM.mdindex 6a757c8694..47883081cf 100644--- a/.pi/gg/SYSTEM.md+++ b/.pi/gg/SYSTEM.md@@ -17,6 +17,7 @@ Coding:Pull requests (PRs):- New branch names are prefixed with "gg/"- Before opening a pull request, ask the user to confirm the description+- Don't explicitly wrap lines in the PR description (each paragraph and bullet is a single line)- When creating a pull request, look for the repository's PR template and follow it- For the AI usage disclosure section, write "YES. pi:llama.cpp/[MODEL]"- Ask the user to tell you what model was used and write it in place of [MODEL]