July 12 - 16

Statistics

On Monday, I made some adjustments to the power files for the filter cache for statistics. The power consumption of a cache includes consideration for line numbers of the cache. Since the filter cache isn't a normal cache (technically it's more of a glorified hash table), it doesn't store any data. However, all of the power functions assume that the filter cache stores as much data as any other cache, which decreases the actual amount of lines. I also had to add in some default values for power statistics, since the preset values did not fit the values for the filter cache.

New Filter Cache Scheme

The rest of this week I've been working on a new scheme for the filter cache. Cesare explained the scheme to me about two weeks ago. Instead of storing all of the addresses that are being written to memory in the filter cache, we store all of the addresses that are read. It's a slightly more complicated filter cache scheme, but its not too bad. I started by making a diagram of what would happen to the filter cache on memory reads and memory writes in different situations to make sure that I understood what was going on before I started to implement anything.

Surprisingly, the new filter cache wasn't that bad to implement! (Actually, one of the more difficult parts was trying to come up with a name for the new scheme.) I've definitely gotten a lot more comfortable with working with the code inside the simulator over the past few weeks, because I barely needed to think about which methods and functions to use and where to put the code. (In the past, those were the most difficult parts.) I was also able to edit the simplest benchmark to get it to test what I wanted to on my own - it was exciting! By Wednesday morning, I had finished the implementation of the new filter cache scheme and was confident that it worked. I ran some preliminary results just to test the scheme out a little bit more (if they didn't make sense, I would know that something could be wrong.) The results looked promising at a first glance!

Statistics^2

Cesare and I talked about some new statistics that we should add to the simulator before running a suite of benchmarks. I added the statistics to the file. Then, Cesare showed me how to update the scripts that create really nice, easy-to-import-into-Excel files of the statistics so I could include the new statistics that I created. After doing that, I was supposed to run the simulator once to see if the files parsed correctly.

Simulator Troubleshooting

Of course, it wouldn't be a true attempt at running the simulator with the scripts if something didn't go wrong. The scripts generated a command string that wasn't valid, and hence would not actually run the benchmark. Apparently, this was caused by a missing space in one of the strings. After that was fixed, the parsing additions that I added worked correctly (Phew!). I did learn some useful commands for committing the code I've worked on to the SVN repository, as well as checking out the new code from SVN and updating some files.

On Friday morning, I worked on adding a new switch to the simulator for the new scheme (read-filter) and made some changes so the new scheme name could be included in the statistics file. With Cesare's help, I edited one of the scripts that runs multiple simulator jobs so that the new filter cache scheme could be included. I added another statistic to the statistics module, and then started to run a very large simulation!

Going by so quickly!

On Thursday, (I'm a little out of order - sorry), Professor Bahar mentioned the possibility of finding a student research competition to enter a poster of what I've been working on in. I would love to go to a conference and present a poster so I was very, very excited! I also can't believe that I've finished my 7th week here already - this summer is going by so fast!

Weekend

Hamilton Reunion!!

This weekend I traveled to Connecticut to my friend Leigh's house. Leigh is one of my close friends from Hamilton. Four of my other good friends were meeting us at her house on Saturday. Since a lot of us went abroad last semester, we hadn't all been together since December. I have been so, so excited for this weekend for a very long time!

I arrived at Leigh's by train on Friday night. We woke up and made brownies (with a lot of extra chocolate), and then hung out and talked until Sarah and Lou arrived! Sarah, Lou, Leigh, and I, went downtown to see an art festival. There were a lot of gorgeous paintings to look at! Then we got lunch at a cafe and talked for a few hours. Afterwards, it was time to pick Sarah and Mallory up from the train station!

We piled on Leigh's couch after Sarah and Mallory arrived!

After getting back to Leigh's house, we all piled up on Leigh's couch and talked (and ate, of course.) We ate a lovely dinner outside with Leigh's friend from home, Phillipa, and friend from Spain, Christina. After dinner (and a lot more talking), we all crammed into Leigh's room for a very large sleepover. We talked and told riddles until we all fell asleep.

It was so, so, so nice to be able to see everyone from Hamilton again. Now I'm getting very excited about going back to school for my senior year in the fall!