pub const DENSE_GEMV_BATCHM_MAX_M: u32 = 16;Expand description
2026-09-25: Rows dense_gemv_bf16_batchm computes: MAX_M in
kernels/gb10/common/dense_gemv_bf16_batchm.cu. The kernel clamps a
larger M, so dense_gemv_batchm refuses above this value. Each row’s
accumulation does not depend on M, so a row’s result is the same at every
batch width.
Other readers: the row thresholds of the GLM-5.3 DSA, KDA and MLP dispatch
(glm5next_dsa/layer.rs, glm5next_kda/mod.rs, glm5next_mlp/forward.rs,
glm5next_layer/levers.rs), the verify_k workspace size
(glm5_next_load/loader.rs) and the clamp in
target_defaults::resolve_batchm_max.