Today, I started looking at the wrappers for a single multiplier. Love and I talked for a while about how to implement one and what type of data to collect and keep track of. Love and I had a slight miscommunication with the data the first time around, but after that, things went pretty smoothly. He told me to look at about wrapper placement and the different combinations. For example, would it be for both input wrappers to be put on the same side versus opposite sides or side by side? He also told me to look at the number of frames and bit files that they generate. I will start twiddling with the files tomorrow.

Today, I ran the code to generate the multipliers (8,16,32) bits overnight and hopefully the data will be good tomorrow. I ran some over lunch today but it turned out that I had mistyped some of my code, so none of the .bit files were correct. I met with Professor Eli and gave her an update on what I've been up to. For now, I will work on the wrappers and hold off on the other merged modules.

I wrote the scripts to look at the 8-bit multipliers with wrappers. So far, I'm not sure what the best result is, but hopefully I'll find out tomorrow after I look at the output. I wonder if there will be a big difference whatsoever. I ran some pre-lim tests on the bigger boards and I'm worried about the time it will take. The last 32-bit one I ran took 3 and a half hours to finish.

Today I continued looking at the 8-bit multipliers with wrappers. I looked at my scripts this morning and talked to Love about it in the afternoon. I'm not quite finished yet, but I talked to him about what I had found thus far. He told me about one of the implementation methods and how one module (plus wrapper) would take up one frame's worth of space and then the above frame would be the only one being reconfigured. So, when I run further tests, sometimes I might need to make the width more so that the wrappers can fit at the top and bottom. One thing that I have noticed with the multiplier+wrappers is that depending on where the wrappers are placed, the entire routing configuration is different, so then when I measure the difference in frames, it says the entire thing is different, not just the wrapper itself. I think I will need to find an alternate method to figure out the differences in frames.

Today, I ran through some of the multiplier data and tried to make some trends through my data to see if there was some correlation. So far, nothing seems to be very obvious. I'll keep looking at the 8-bits and see if there are any trends that I can make out.

I changed some things on the website and hopefully the changes were worth it. I found some code that was wrong so I fixed (hopefully) all of the mistakes.

Last Week