TV Layout Formula — Interactive Visualizer

mma.sync.m16n8k16 — TV_layout_C: shape ((4,8),(2,2)) : stride ((16,1),(8,64))

col = tid % 8 row = (tid / 8) * 2 + (vid % 2) + (vid / 2) * 8

Thread ID (tid): 0

T_inner (tid%8) = 0, T_outer (tid/8) = 0

Value Index (vid): 0

V_inner (vid%2) = 0, V_outer (vid/2) = 0

(tid%8)*1 = 0

(tid/8)*16 = 0

(vid%2)*8 = 0

(vid/2)*64 = 0

offset 0

    → col = 0,  
    row = 0  
    → C[0, 0]
  

Current (tid, vid)

Same thread (all 4 values)

Same value index (all 32 threads)

Thread 0's Registers

C[0, 0]

C[1, 0]

C[8, 0]

C[9, 0]