Current performance (no of cycles) for a single update step of a 3-dimensional state vector
| Software(Microblaze without floating point unit) | Software(Microblaze with floating point unit) | Software(Intel i5 9th Gen) |
|---|---|---|
| 415750 @ 100MHz | 187581 @ 100MHz | 1761 ns (average of 10 runs) |
Hardware(100MHz):
| Write | Read | Computation | DSP | FF | LUT | |
|---|---|---|---|---|---|---|
| Floating Point | 120 | 108 | 365 | 70 | 10834 | 9078 |
| Floating Point(Optimized) | 120 | 108 | 307 | 114 | 9597 | 9251 |
| Fixed Point | 120 | 108 | 249 | 36 | 4767 | 5237 |
- Executing test C code in Microblaze ✔️ (Arun)
- Adding and using a performance counter IP to MicroBlaze ✔️ (Arun)
- Generate C code for Kalman filter using MATLAB ✔️ (Snehan)
- Execute Kalman Filter code in Microblaze with and without FPU ✔️ (Arun)
- Design test ip with AXI in Vivado HLS and interface with Microblaze ✔️ (Sakthi)
- Mesaure performance of C code in Intel i5 processor ✔️ (Snehan)
- Design ip for kalman filter using floating point ✔️ (Sakthi)
- Write C code in microblaze for using the kalman filter ip generated by HLS ✔️ (Arun)
- Change from floating point to fixed point ✔️ (Arun)
- Add optimization directives ✔️ (Arun)
- Report ✔️ (Sakthi, Arun, Snehan)
- Video (In Progress) (Sakthi, Arun, Snehan)