1 Commits

Author SHA1 Message Date
project6-dev
9c97a24edf diag: verify_ixformer_attn.py — test ixformer native attention with head_dim=256
The 10-50x slowdown is from bypassing ixformer SDPA and using Python
matmul fallback. Test if ixformer actually crashes on head_dim=256
or if the bypass was premature.
2026-08-13 04:25:09 +00:00