1. CSE 291-a Interconnection Networks Lecture 10: Flow Control February 21, 2007 Prof. Chung-Kuan Cheng CSE Dept, UC San Diego Winter 2007 Transcribed by Thomas Weng

2. Topics Introduction Bufferless Flow Control Buffered Flow Control - Cut-through - Wormhole - Virtual Channel

3. Introduction Objective: Bandwidth and Latency Resources: Buffer and Channel Channel Switch Buffer Channel State

4. Unit of Messages Message is too long, cut up message RISN Basic chunk is Packet Header Message Packet Tail (indicates packet is done) Body (content) RI = Routing Information SN = Sequence Number Packet should be reasonably long so we minimize overhead.

5. Flit and Phit RISN Packet Tail Typical Range Phits FlitsPacket 8 bits 1-64 bits 64 bits 16-512 bits 1K 128 bits–512K Type Virtual Channel Flit (Flow Control Digit) Flit Phit (Physical Transfer Digit) H, T, or H&T, or Body Packet Phit

6. Bufferless Flow Control What if we have no buffer? Why not have buffer? Less power and improved latency. Three methods: 1. Drop the data (easiest way) 2. Misroute the data (treat data like hot potato) 3. Dropless approach (reservations)

7. Drop the Data Drop the data: Tell sender you dropped data 1) Nack – Negative Acknowledgement (Tell sender you dropped the data) 2) Ack – Acknowledgement (The sender resends if Ack is not received within timeout period) Reverse Channel: Ack and flow control signals are sent in a reverse channel 0 0 1 c b a o h 8 f 0 Drop 0 0 1 c h dropped data b a

8. Drop the Data (cont) Dropped data simplifies things, but pay a price by resending data. Latency is long if data was rejected. 0123 0 FHBBBTHBBBT RNA 1 FHBBBHBBBT RNA 2 FHBHBBBT RNA 3 FHBBBT RA Suppose channel 2 sent and was rejected

9. Dropless Flow Control Dropless Flow Control: Request propagates from source to destination and allocates the channel. Ack is transmitted back to the source. Packet is sent. A tail flit is sent to de-allocate the channel. 0RADT 1RADT 2RADT 3RADT 4RADT Channel Total time T 0 = 3Htr + L/b hops latency / hop length of packet bandwidth

10. Buffered Flow Control 1. Store & Forward 2. Cut-through 3. Wormhole 4. Virtual Channel

11. Store & Forward 1. Store & Forward Flow Control: Each node receives a packet and then sends it out. 0HBBBT 1HBBBT 2HBBBT 3HBBBT Channel T 0 = H(tr + L/b)

12. Cut-through 2. Cut-through Flow Control: Each node starts to send the packet without waiting for the whole packet to arrive. Cut-through is more efficient approach. 1) Good performance 2) Large buffer sizes, consumes more power T 0 = Htr + L/b 0HBBBT 1HBBBT 2HBBBT 3HBBBT 0HBBBT 1HBBBT 2|---- Not Ready ----|HBBBT 3HBBBT Suppose in the middle, we get stuck

13. Wormhole 3. Wormhole Flow Control: Main difference is that we just have a little buffer, don’t need to store the entire packet. In terms of performance & bandwidth, it’s better than cut-through. States: Idle, Wait, Active 0 1 LI T B B H 0 1 L H WU T B B

14. Wormhole (cont) 0 1 L BH WU T B 0 1 U BB AU TH 0 1 U T AU B B H HBBT HBBT In: Out:

15. Virtual Channel 4. Virtual Channel: Try to split channel in time domain. By doing so we can fully utilize channel since we don’t waste it by holding it. Wormhole + Virtual Channel = Winner

16. Virtual Channel (cont) 1 ABAB Input a a n a 1 a 2 a 3 a 4 Input b b n b 1 b 2 b 3 b 4 Interleaved a n b n a 1 b 1 a 2 b 2 a 3 b 3 a 4 b 4 Winner Takes All a n a 1 a 2 a 3 a 4 b n b 1 b 2 b 3 b 4 2 4 3