From 95302ac672411a119a8a6763afcd477d2cdca05e Mon Sep 17 00:00:00 2001 From: bong-water-water-bong Date: Sun, 15 Nov 2026 03:25:43 -0400 Subject: [PATCH] token_rate_floor_device: a real default floor, so the row judges instead of skipping forever engine#2481 made `--floor 0` exit 77 (record-only asserts nothing, so 0 must not read as PASS) and the ORCH's instruction allowed either that OR a real default. The device row's `TOKEN_RATE_FLOOR` defaulted to **0**, so with the fix in place it is a permanent skip: registered, tested, never applied -- exactly engine#2418's defect, arrived at from the other side. Default is now **60**, and it is measured rather than guessed: this row's own model (Qwen3-0.6B-Q4_K_M) ran **144.5 tok/s** sequential through `1bit serve` on gfx1151 on 2026-11-15, so 60 keeps 2.4x headroom and still catches a 2.4x regression. TOKEN_RATE_FLOOR=0 remains available and now says what it means: 'record-only, asserts nothing; unset it to judge at the default 60'. Also fixed a latent bug in my own guard from the first attempt: it used `exit "$SKIP"` above the line that defines SKIP, so it died with 'SKIP: unbound variable' instead of skipping. Literal 77 now. Verified: floor=0 -> 77 with the reason; default -> 77 here (no binary/model = skip, not pass); 6 arms 0 failed; copyright --check rc=0. --- tools/ci/token_rate_floor_device.sh | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/tools/ci/token_rate_floor_device.sh b/tools/ci/token_rate_floor_device.sh index 1e2dee3f7..e55a79b41 100755 --- a/tools/ci/token_rate_floor_device.sh +++ b/tools/ci/token_rate_floor_device.sh @@ -22,7 +22,15 @@ # that silently passes because there was no model would be exactly the failure this tool exists to catch. set -u ROOT=${1:-$(cd "$(dirname "$0")/../.." && pwd)} -FLOOR=${TOKEN_RATE_FLOOR:-0} +# A REAL default, not 0. `--floor 0` means "record, do not judge" and exits 77 (SKIP), so a default of 0 would +# make this row a permanent skip: registered, tested, never applied -- engine#2418's defect. 60 is measured, not +# guessed: this row's own model (Qwen3-0.6B-Q4_K_M) ran **144.5 tok/s** sequential through `1bit serve` on +# gfx1151 (2026-11-15), so 60 keeps 2.4x headroom and still catches a 2.4x regression. Override with TOKEN_RATE_FLOOR. +FLOOR=${TOKEN_RATE_FLOOR:-60} +if [ "$FLOOR" = "0" ]; then + echo "SKIPPED: token_rate_floor device row did NOT run -- TOKEN_RATE_FLOOR=0 is record-only, which asserts nothing; unset it to judge at the default 60" >&2 + exit 77 +fi PORT=${TOKEN_RATE_PORT:-8191} SKIP=77