09:37 <lkcl_> somlo, yes, i saw the CSRs after investigating a bit more - pythondata_cpu_microwatt provides its own version of timer0_value_read (and write?)

09:38 <lkcl_> however they happen not to be used

09:39 <lkcl_> am still getting these:

09:39 <lkcl_> Memtest at 0x0000000040000000...

09:39 <lkcl_> [bus: 0x0]: 0xaaaa0000 vs 0xaaaaaaaa

09:40 <lkcl_> that's after putting in a nmigen-based wishbone 64-to-32-bit downconverter. yet in sim it's fine

09:48 <daveshah> Is picorv32 reliably working fine at the same frequency?

10:06 <lkcl_> daveshah: yes

10:06 <lkcl_> 55mhz

10:09 <lkcl_> i've put in a system for dumping GPRs (implemented microwatt's DMI interface)

10:10 <lkcl_> and a way to activate that on a "range" of values of PC

10:10 <lkcl_> then ran it on both microwatt and libresoc and did a diff

10:11 <lkcl_> that told me that somewhere a couple of GPRs end up being different

10:12 <lkcl_> now of course i need to do a binary search tracking down which function causes that difference

10:15 <lkcl_> am just going to have to be veeeery patient.

10:34 <lkcl_> daveshah: the only main difference which *might* affect timing (somehow?) is that instruction read is still 64-bit. however i am beginning to suspect something else going on, and will start running some of microwatt unit tests

10:50 <lkcl_> ah! i have errors i can track down! alriiiight. progress

15:30 <lkcl_> daveshah: running a microwatt unit test, found 4 ALU bugs (whoops) - andc, orc, modsd, divsd. will come back to memtest once i'm confident these are dealt with :)

15:31 <daveshah> I think Dolu ran VexRiscv in Verilator lockstep with a reference C++ model to debug Linux in litex startup

15:31 <daveshah> any divergence in register value would stop the sim

15:32 <lkcl_> niice

15:33 <lkcl_> i'm doing something similar in the unit tests

15:33 <lkcl_> we have a python-based cycle-accurate POWER9 simulator being developed co-simultaneously with the HDL

15:34 <lkcl_> except, running large programs takes... a while :)

15:49 <futarisIRCcloud> https://spoolqueue.com/new-design/fpga/migen/litex/2020/08/11/acorn-cle-215.html

15:49 <tpb> Title: Running Linux on the Acorn CLE-215+ | The Spool Queue (at spoolqueue.com)

15:50 <futarisIRCcloud> Didn't see this blog post from a couple weeks ago

19:12 <lkcl_> futarisIRCcloud: wow, he got linux up and running on it, including a uart and sdmmc - despite the fact that those PCBs have what... 12 pins? :)

19:38 <_florent_> futarisIRCcloud: nice, thanks for sharing, i'm also interested in getting LitePCIe working in Root mode, i'll get in touch with him to see if we could collaborate on this