- Replace Python lists with individual scalar variables (s0..s5, i0..i5, a0..a5)
- Replace min-heap sift-down with fully unrolled sorted insertion
(descending order, no dynamic indexing, no while loops)
- Replace raw SMEM pointer arithmetic with CuTeDSL SMEM tensors
(s_merge_s, s_merge_i, s_merge_a)
- Replace cute.where with cute.math.fmax
- Fix expert index calculation: col + tile_n_offset + subtile_idx * epi_n
- Top-6 accumulates across all N-tiles (for E=384 with 3 tiles of 128)
- Add iter_acc_early_release for overlapping accumulator
- Rewrite test to compare fused kernel vs 2-kernel reference path
- Remove stale memory doc
Major fixes:
- Added tiled_mma_sfb creation (always CtaGroup.ONE, rounded N)
- Added mma_tiler_sfb, cta_tile_shape_mnk_sfb, cluster_layout_sfb_vmnk
- Use blockscaled_utils.make_smem_layout_sfa/sfb (with sf_vec_size)
instead of sm100_utils (which doesn't support block-scaled SF layouts)
- Proper TMEM column accounting for SFA + SFB + accumulator
- Fixed make_blockscaled_trivial_tiled_mma argument order
(a_dtype, b_dtype, a_major, b_major, sf_dtype, sf_vec_size, cta_group, mma_inst_shape)
- Fixed SFB TMA atom to use tiled_mma_sfb and cluster_layout_sfb_vmnk
- Fixed SFB partition_SFB to use tiled_mma_sfb.get_slice
- Fixed SFB global tile partitioning to use mma_tiler_sfb
- Fixed mainloop_s2t_copy_and_partition to use TMEM fragments
(make_fragment_SFA/SFB) as the tSF parameter
- Updated run_nvfp4_fused_router wrapper to accept processed weight
tensors from Nvfp4Linear._mat_b and _scale_b
- Updated test to properly build Nvfp4Linear and use processed weights
The old code was a rough sketch that never worked — it was missing
the entire tiled_mma_sfb infrastructure, used wrong SMEM layout
functions, and had broken TMA atom setup for scale factors.
docs/p7_tmem_column_layout.md: Verified that tcgen05.ld 32x32b.x8 is
the correct instruction for multi-row softmax. Each call reads 8 KV
positions for 32 rows. No instruction change needed from single-row.
test_p7_multi_row_softmax.py: Tests T=1,4,32,64,128 at various HD and N.
Gate: cos >= 0.999996.
- Kernel loops over KV tiles internally with running max/sum rescale
- SMEM accumulator sOacc[hd] replaces TMEM accumulation across tiles
- P is UN-NORMALIZED for multi-tile (exp(s-max), not /sum)
- Per KV tile: QK→softmax→PV→TMEM→read→add to sOacc
- Final: O = sOacc / running_sum
- Single tile (n_kv_tiles=1): same as before, no rescale
- Updated CAPI, Python loader, production.py fast path
- Added multi-tile test cases (N=256, 512)
