pub const W4A16_GEMV_SW_OUTS_PER_BLOCK: u32 = 8;
2026-09-25: Outputs per 256-thread block of the single-warp kernels (w4a16_gemv_sw and the _sw kernels in w4a16_gemv_fused.cu); equals #define N_PER_BLOCK_SW.
w4a16_gemv_sw
_sw
#define N_PER_BLOCK_SW