search

TeCSAR-UNCC / gem5-SALAM

BSD 3-Clause "New" or "Revised" License
84 stars 23 forks source link

Fix floating point conversions and comparisons; fix immediate value parsing #11

Closed Sacusa closed 3 years ago

Sacusa commented 3 years ago
  1. Conversions from floating-point to integer and vice-versa were incorrect.
  2. Floating-point extension and truncation were incorrect.
  3. Floating-point comparisons were performed without interpreting the values as floating point.
  4. Implemented BitCast instruction.
  5. Reading immediate values with a negative exponent was not handled.
Josh-Slycord commented 3 years ago

Hello Sacusa, Just wanted to let you know I have verified your branch and it will be merged in soon. I wanted to let you know in advance that in the next few days the master branch is going to be updated to the new gem5-SALAMv2 code base, which has a completely new parser written using the LLVM API and updated to version 9. It also supports internal function calls, which was a major limiting factor of the old parsing method. Additionally the new version can be compiled to support AP data types or standard data types, all still using the LLVM API codebase. We will continue to provide support for the old version as we work towards finalizing the new version, and in the future the transition to v2 should be very straight forward, as we are working to automate and formalize as much of the setup process as possible.

Sacusa commented 3 years ago

Hi Josh, thanks for the update! I didn't realize at the time of creation of this pull request that there is a dev branch which seems to have taken care of most of the bugs I mentioned. Looking forward to v2 of the simulator!