daniel/h3-blackwell-runtime
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Complete Vortex attention interface design

Browse source
This commit is contained in:
Daniel Maddern 2026-08-26 13:40:59 +07:00
parent 7d0608e857
commit ff8f7672b3
1 changed files with 16 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
research/vortex_exact_attention/DESIGN.md
View file

@ -166,3 +166,19 @@ boundary rather than a temporal schedule rearrangement.
Stop immediately on changed KV order, any material score tensor, nonzero output Stop immediately on changed KV order, any material score tensor, nonzero output
difference, forced local spilling, inability to beat the `220 ms` model gate, or difference, forced local spilling, inability to beat the `220 ms` model gate, or
fallback to Sage2 for main computation. fallback to Sage2 for main computation.
## Interface And Future Ulysses Layout
The isolated API consumes projection-strided BF16 NHD views directly. Its work
descriptor must carry `q_start`, `q_count`, `kv_count`, `head_start`,
`head_count`, source token/head strides, destination token/head strides, and a
future rank destination. No HND transpose or contiguous input copy is allowed.
Output stores land directly in token-major `[token, local_head*128+d]` rows for
the local output projection. Ragged local token counts and nonzero destination
offsets are legal, but each rank must receive the globally ordered KV sequence
before attention. Independently normalized KV partials must never be merged.
VEA-B is optimized first for the single-GPU Q128 shape. A future SM120/SM100 or
Ulysses schedule may choose Q64 or a different producer geometry, but it must
retain this numerical and descriptor contract and pass separately tuned gates.