Oh dear
Damn. Only answered 4 questions and a fragment.
Q1: Comparative Architectures - 16/20
Easy question on processor features, divided into seven parts.
Q2: VLSI Design - 14/20
Nice question on minority gates - boolean algebra, NAND circuit, direct CMOS circuit, logical effort. Though I think I screwed up the overall logical effort and I'm really not too certain about some of my calculations in that part.
Q3: Digital Communication II - 15/20
TCP. Went fairly well. Did the RTT calculations for TCP Vegas and TCP Reno, wibbled about buffering, got fast retransmit/fast recovery a bit confused, waffled a touch on the last part but included a diagram to show what happens to transmission rate with a) rare and b) moderate packet loss.
Q8: Optimising Compilers - 17/20
Fairly easy question on flowgraphs, SSA and Reaching Definitions (which aren't really covered in the course, but v.similar to AVAIL).
Sadly, those took well over two hours and abortive attempts at Computer Systems Modelling and Denotational Semantics only left me with five minutes to do:
(Q5: Advanced Systems Topics - 7/20)
Only attempted a, b, c(i) here and with very sketchy handwriting. Named appropriate examples (CAN, Napster, Gnutella) and discussed topology, queries and a bit of generality; said unstructured networks were better for churn as there's no overlaid structure to regenerate; said unstructured P2P could explicitly weight sources by ping and then ran out of time before I could fail to answer how to deal with latencies in structured networks.
Gives an overall estimate of 69%. Which probably means low-mid 2i allowing for overestimates. Bugger. It probably went better than Paper 4 last year (for which I only got 60%), so a good (75%+) paper today and I should still be in with a reasonable chance of a first.
So then I went out and got drunk... and, unfortunately, ended up being given two shots of tequila.
Q1: Comparative Architectures - 16/20
Easy question on processor features, divided into seven parts.
Q2: VLSI Design - 14/20
Nice question on minority gates - boolean algebra, NAND circuit, direct CMOS circuit, logical effort. Though I think I screwed up the overall logical effort and I'm really not too certain about some of my calculations in that part.
Q3: Digital Communication II - 15/20
TCP. Went fairly well. Did the RTT calculations for TCP Vegas and TCP Reno, wibbled about buffering, got fast retransmit/fast recovery a bit confused, waffled a touch on the last part but included a diagram to show what happens to transmission rate with a) rare and b) moderate packet loss.
Q8: Optimising Compilers - 17/20
Fairly easy question on flowgraphs, SSA and Reaching Definitions (which aren't really covered in the course, but v.similar to AVAIL).
Sadly, those took well over two hours and abortive attempts at Computer Systems Modelling and Denotational Semantics only left me with five minutes to do:
(Q5: Advanced Systems Topics - 7/20)
Only attempted a, b, c(i) here and with very sketchy handwriting. Named appropriate examples (CAN, Napster, Gnutella) and discussed topology, queries and a bit of generality; said unstructured networks were better for churn as there's no overlaid structure to regenerate; said unstructured P2P could explicitly weight sources by ping and then ran out of time before I could fail to answer how to deal with latencies in structured networks.
Gives an overall estimate of 69%. Which probably means low-mid 2i allowing for overestimates. Bugger. It probably went better than Paper 4 last year (for which I only got 60%), so a good (75%+) paper today and I should still be in with a reasonable chance of a first.
So then I went out and got drunk... and, unfortunately, ended up being given two shots of tequila.