woutermans 5198fd38b3 Hardware e4m3 in the texture tails: enc0 tail and dec0 head ...

On the gfx12 native-FP8 path the enc0 tail and the dec0 head encode their results with v_cvt_pk_fp8_f32 on the f32 copies of the f16 values (e4m3x4_f32 in tex_common.h) instead of ZLUDA's integer e4m3x2. The clamps keep the conversion in range: +-2pi for enc0 (the f16 bound hi[0], same in f32) and +-448 for dec0, both as a median of three, where a NaN gives the lower bound. The codes equal e4m3x2 for every non-NaN f16, checked over all 65536 patterns on an RX 9070 XT; a NaN in dec0 now encodes as -448 instead of the NaN code. Harness frames are byte-identical in both sets (LDR and HDR, motion vectors at render and display resolution). dec0: 0.21 -> 0.15 ms at 1280x720 render and 0.74 -> 0.53 ms at 2560x1440; enc0 changes by 1-2% (its time is ZLUDA's compile of the feature code). -DD4R_TEX_SOFT_E4M3 (D4R_TEX_CFLAGS) keeps e4m3x2.

2026-10-06 23:25:14 +02:00

..

architecture.md

Fix DLSS quality-change corruption and frame completion ordering

2026-10-02 19:14:07 -04:00

building.md

Port K kernel tuning and M gfx1101 flags

2026-10-05 14:48:37 -04:00

native-kernels.md

Hardware e4m3 in the texture tails: enc0 tail and dec0 head

2026-10-06 23:25:14 +02:00

performance.md

docs: RDNA4 fast and accuracy sets, FP8 and 16-bit, before and after

2026-10-06 15:18:11 +02:00