2020-09-09 04:03 PM
Solved! Go to Solution.
2020-09-09 04:25 PM
Probably the STM32H7A/H7B WLCSP132 chips. (4.57 X 4.37 mm)
2020-09-09 04:25 PM
Probably the STM32H7A/H7B WLCSP132 chips. (4.57 X 4.37 mm)
2020-09-09 04:31 PM
still 10x10mm, was hoping to get a smaller package.
the L412 QFN-48 is 5x5 but only single precision..
2020-09-10 07:56 PM
The F767 has the FPU-D
https://www.st.com/resource/en/datasheet/stm32f767zi.pdf
Most of these ICs have a lot of FLASH/RAM that contribute to the die size, the FPU-D is perhaps 3x size of FPU-S, I couldn't get the ARM guys to share specific info.
LQFP100 H730AB
https://www.st.com/resource/en/datasheet/stm32h730ab.pdf
2020-09-11 12:49 AM
One thing the OP didn't make clear was whether it is acceptable to do the floating-point in software.
All the compilers provide software double-precision floating-point where it's needed. It will increase execution-time (hence power-consumption), and use up more FLASH. But often this is a good overall solution.
But even the smallest stm32l0 can do it. A quick check found stm32l052T8 is 2.61 x 2.88 mm but I don't know if that's the smallest.
2020-09-12 03:12 AM