Both w and k are 64 element arrays. The w[t] element is used each round where t is the current round {0...63}. The diagram indicates just that. It simply omits rounds 2 to 62 because all the rounds are identical and it would make the diagram massive otherwise. Diagrams seem to not be your thing did you look at the pseudocode I linked to. It describes it in very 'english like' syntax. The SHA-2 family really isn't that complex. ECDSA on the other hand ...
h = g
g = f
f = e
e = d + h + Ch(e,f,g) + Sum1(e) + w[t] + k[t]
d = c
c = b
b = atic
a = h + Ch(e,f,g) + Sum1(e) + w[t] + k[t] + Sum0(a) + Maj(a, b, c)
http://en.wikipedia.org/wiki/SHA-2#PseudocodeSpecifically "a = h + Ch(e,f,g) + Sum1(e) +
w[t] + k[t] + Sum0(a) + Maj(a, b, c)" So on round 0 w[0] and k[0] is used, on round 1 w[1] and k[1] is used ... on round 63 w[63] and k[63] is used.