The blue flit in not allowed to make progress because the gray packet has already reserved the links. Since the head of the gray lint packets is stuck since the link is busy, the blue flit packets are stuck as well.
This comment was marked helpful 0 times.
raphaelk
Question:
How is this different from Cut-through flow control algorithm if every packet buffer in switches had size of 1?
Are they basically similar thing except in different scale?
This comment was marked helpful 0 times.
raphaelk
I think it was briefly mentioned in class that the link between two nodes, where heads of two packets, being marked "reserved" could be little misleading because if the node where the head of gray pack is at had one more buffer, blue could use that idle link to make progress, while gray packets are holding their position (while head is being processed).
This comment was marked helpful 0 times.
abunch
@raphaelk I think the only real difference is what happens when the head/first packet blocks, in cut-through if the head is blocked then the tail and other body packets continue transferring while in wormholes if the head flit blocks the rest of the packet waits. In cases where there is no blocking I think your intuition on wormholes just being cut-through with smaller packets is right. I suppose the real thing to look at here is the difference between packets and flits.
The wikipedia article actually has an example (http://en.wikipedia.org/wiki/Wormhole_switching) and it even says "This, however, can be confusing since cut-through routing does the same thing" in reference to buffers and transmission, once again making me think the only difference is the sizes.
The blue flit in not allowed to make progress because the gray packet has already reserved the links. Since the head of the gray lint packets is stuck since the link is busy, the blue flit packets are stuck as well.
This comment was marked helpful 0 times.
Question: How is this different from Cut-through flow control algorithm if every packet buffer in switches had size of 1?
Are they basically similar thing except in different scale?
This comment was marked helpful 0 times.
I think it was briefly mentioned in class that the link between two nodes, where heads of two packets, being marked "reserved" could be little misleading because if the node where the head of gray pack is at had one more buffer, blue could use that idle link to make progress, while gray packets are holding their position (while head is being processed).
This comment was marked helpful 0 times.
@raphaelk I think the only real difference is what happens when the head/first packet blocks, in cut-through if the head is blocked then the tail and other body packets continue transferring while in wormholes if the head flit blocks the rest of the packet waits. In cases where there is no blocking I think your intuition on wormholes just being cut-through with smaller packets is right. I suppose the real thing to look at here is the difference between packets and flits.
The wikipedia article actually has an example (http://en.wikipedia.org/wiki/Wormhole_switching) and it even says "This, however, can be confusing since cut-through routing does the same thing" in reference to buffers and transmission, once again making me think the only difference is the sizes.
This comment was marked helpful 0 times.