probe(review): fp8 serve row + 3-ref fixture + fp8 quality-collapse diagnosis
Build Wheel / build (3.12) (pull_request) Failing after 50s
Build Wheel / build (3.11) (pull_request) Failing after 50s
pre-commit / pre-commit (pull_request) Failing after 54s

Review follow-up on PR #4: run --quantization fp8 through the served
path on the 3090 (the flag issue 1 validated for capacity), add the
untested 3-reference fixture, and measure all counts.

Capacity: confirmed — all of 1/2/3/4 references fit and run (4-ref
44.1-44.4s, server peak 23,522MB, nvidia-smi max 23,983MiB; Marlin
W8A16 is faster than BF16, 3.3x the Diffusers nocache baseline).

Quality: every fp8 output is structurally broken on this commit and
card (~97% transparent RGBA, grey RGB, PSNR 7-9dB) — on the edit path
at every reference count and on T2I alike (issue 1's fp8 artifacts
reinspected: same collapse; that run validated rc+image, not output
structure). img_mlp exclusion does not fix it; transformer-only fp8
also collapses, pinning the culprit on the Marlin-quantized DiT.
3090 verdict stays not-built; upstream fix order is fp8 output
correctness before any capacity role for it.
This commit is contained in:
admin committed 2026-10-08 06:27:36 -04:00
1 parent 67eb7fdb48
commit 4f5f99d8c6
6 files changed
+1102 -16

No files matched your search

+1
View File
@@ -33,6 +33,7 @@ CFG = 1.0
FUNCTIONAL_FIXTURES = (
"edit-expression-ysolde", # 1 reference
"edit-compose-pair", # 2 references
"edit-compose-three", # 3 references (derived; see fixtures.json)
"edit-compose-four", # 4 references
)
MEASURED_FIXTURE = "edit-compose-four"