[Libre-soc-dev] daily kan-ban update 27apr2022

[Luke Kenneth Casson Leighton]

On Thu, Apr 28, 2022 at 10:23 AM Jacob Lifshay <programmerjake at gmail.com> wrote:

> I went back to working on the division algorithm, I wrote the HDL
> version of it but ran out of time today, so committed it in a broken
> state with the appropriate tests set to @unittest.skip.

no great, good call.  the 192-bit-long DIV FSM is causing
massive adder chains on the Arty Xilinx FPGA (because we can't use
CARRY4 blocks), and anything that would use a DSP Mult block
(such as Goldschmidt) is a Good Thing.

yesterday i got much further with linux microwatt 5.7 booting and
suspect one of two things: (a) lwarx/stwcx. are missing and giving
grief (b) a mismatch between the L1 cache sizes + number of TLB
entries and the devicetree file is giving grief.

l.