M post: the history encode relies on a driver conversion fold without a start-up check

Offen
#232 1 Kommentar 0 Reaktionen 0 zugewiesene Personen Auf GitHub ansehen

Maintainer antworten meist innerhalb von 1 Tag

Dieses Issue hat noch niemand übernommen.

Bewertung

Schwierigkeit
5/5
Geschätzter Aufwand
Über eine Woche
Anfängerfreundlichkeit
15/100
Issue-Typ
Feature
Klarheit
Größtenteils klar
Aktivitätsstatus
Aktiv
Tech-Stack
cpp, python

Rechercherichtung

Read #231 first to see the Model L guard this mirrors, then the history-encode block in clean/dx12/kernels/post.hlsl around lines 1127-1132 where the mad(log2(..), kEncK, 0.0f) result feeds cvt3_f16. Pair that with the earlier round-to-odd encode to build the needs=nofold port, add the needs=fold port, and write the start-up self-test that runs both real post shaders over the crafted midpoint frame and compares raw bytes. Add the isa= rules in build_dx12.py last; done when the probe selects a port from real shader output and both sites are guarded.

Vom Indexierungsmodell aus dem Issue-Text verfasst.

Beschreibung

Model M's D3D12 post kernels (clean/dx12/kernels/post.hlsl, around lines 1127-1132) write the history texel as
mad(log2(..), kEncK, 0.0f) fed into the driver's round-to-nearest-even matrix conversion (cvt3_f16). The gfx1201
D3D12 compiler folds that into v_fma_mixlo_f16 .., kEncK, 0, which is HIP's own instruction, so today's bytes match HIP.

The exactness depends on that fold. A driver that does not fold would round the FP32 product first and then convert
it: two roundings, one FP16 ulp off wherever the FP32 product lands exactly on an FP16 midpoint. The frame checks would
likely not see it, because synthetic data rarely hits such midpoints.

Model L's post is getting the same guard in #231. M's post should get it too:

  • a needs=fold / needs=nofold pair of post ports, with the previous round-to-odd form as the no-fold fallback;
  • a start-up self-test that runs the REAL fold and no-fold post shaders on one small crafted frame whose history-encode
    inputs sit on FP16 midpoints from both sides, ties, +-0, subnormals and the floor, and compares their raw bytes
    (equal: fold port; different or any error: no-fold port). A separate probe shader is not enough, because the driver
    can compile the real post shader differently;
  • an isa= rule in build_dx12.py on each port that relies on the fold, tied to the actual sites.
Vorherrschende Sprache
C++
Sterne
12
Forks
2
Ø Merge
3 Std. 10 Min.
Gemergte PRs (30 T.)
171

Entwicklungsumgebung

Erste Schritte

  1. Lesen Sie das ganze Issue und danach den Beitragsleitfaden des Projekts.
  2. Schreiben Sie ins Issue, dass Sie es übernehmen — das erspart doppelte Arbeit.
  3. Forken Sie das Repository und arbeiten Sie in einem Branch.
  4. Öffnen Sie einen Pull Request, der die Issue-Nummer nennt.

Mehr aus phantoslayer21/wronggpu

Alle Issues in phantoslayer21/wronggpu

Ähnliche Issues

Weitere Issues zu C++

Neue Issues direkt in Ihr Postfach

Eine kurze Übersicht über anfängerfreundliche GitHub-Issues.