Nail biting
Google tomorrow.
Intel on March 3.
Stanford on March 15-16.
Microsoft either March 11-12 or 21-22, I hope the former.
No word from other schools. Other systems students and I have concluded that the committees are using an O(n²) algorithm, hence their slow replies. Clearly they need systems researchers to show them better scheduling approaches. I commented to David that I was stressing out about not hearing. He said, "Well, you can call some people, see how much they can tell you." I frowned a bit, being recalcitrant to call people I only sort-of know. He smiled. "Or you could just chill out."
Nice breakthrough on my thesis today. I've been spinning on Chapter 3: it's been very wishy-washy, talking about the three layers and their requirements. David suggested a performance model. Now I decompose the cost of a reprogramming system as P(n) + E(n,t), where P is the cost of installing code (generally O(n)) and E is the cost of executing the code. This makes it clear why concise code is better: not only does it reduce the transmission cost, it reduces the interpretation overhead (assuming a uniform cost per instruction). Chapter 3 now has some meat in it and is starting to be precise about what I am trying to say.
Intel on March 3.
Stanford on March 15-16.
Microsoft either March 11-12 or 21-22, I hope the former.
No word from other schools. Other systems students and I have concluded that the committees are using an O(n²) algorithm, hence their slow replies. Clearly they need systems researchers to show them better scheduling approaches. I commented to David that I was stressing out about not hearing. He said, "Well, you can call some people, see how much they can tell you." I frowned a bit, being recalcitrant to call people I only sort-of know. He smiled. "Or you could just chill out."
Nice breakthrough on my thesis today. I've been spinning on Chapter 3: it's been very wishy-washy, talking about the three layers and their requirements. David suggested a performance model. Now I decompose the cost of a reprogramming system as P(n) + E(n,t), where P is the cost of installing code (generally O(n)) and E is the cost of executing the code. This makes it clear why concise code is better: not only does it reduce the transmission cost, it reduces the interpretation overhead (assuming a uniform cost per instruction). Chapter 3 now has some meat in it and is starting to be precise about what I am trying to say.