Suggestions/Comments | Overview of Design
We were very disappointed with the OMNILAB software. A strange problem we had was loading new
test vectors once we made changes to the command file. We used "convert" to create a new stimulus
file, but getting OMNILAB to realize the changes required some voodoo. Also, once we got our
test to run, verification was made difficult by the inability to group signals into vectors.
Bit by bit verification takes much, much longer than comparing two hex numbers.
We were also surprised to discover that we would need all 24 of our stimulus pins to test our
setup with SRAM. We had no trigger pin left, and triggering off of one of our control signals
was often very difficult.
Our speed tests further confirmed our expectations of our chip's performance. We observed
complete functionality at 4.25 MHz, just as predicted. And at 8.5 MHz, well above our predicted
maximum of 4.55 MHz, we saw problems created by delays in retrieving instructions from memory.
Suggestions/Comments:
We encountered many frustrating problems in the lab unrelated to the functionality of our chip.
The computers, 486 machines running Windows 95, were often painfully slow, and there were frequent
problems printing our results.
Overview of Design:
We were proud to verify that our design worked exactly as we had planned last semester. Every
aspect of our chip, including the entire SRAM interface, worked perfectly. We were a little
uncertain coming into the testing because we had no SRAM model last semester with which to
test our chip. But our analysis of the SRAM's timing discipline proved to be correct.