programmerjake | lkcl, read through most of the svp64 branch page in the wiki, lgtm except for one change i made...changed a mention of CUDA to OpenCL-style | 08:47 |
---|---|---|
lkcl | i can't quite believe how many times i re-read the spec pages and end up noticing new things every time. it's sometimes well over 20 times a day | 10:50 |
Veera[m] | lkcl: Can we consider Bugs: 750, 790 and 791 finished? | 10:53 |
lkcl | Veera[m], yeah i'd say so, great work. if you'd like to do one more, there is ls2? | 12:11 |
lkcl | i created this but haven't run it, and there's no wiki page for it either | 12:12 |
lkcl | https://git.libre-soc.org/?p=dev-env-setup.git;a=blob;f=hdl-dev-ls2;hb=HEAD | 12:12 |
Veera[m] | ls2: what I have to do? | 12:13 |
lkcl | basically run it, on the uoregon server, and check you can get "helloworld.bin" | 12:18 |
lkcl | the two crucial commands are, in the soc directory: | 12:18 |
lkcl | make microwatt_external_core | 12:19 |
lkcl | and | 12:19 |
lkcl | python3 src/ls2.py arty_a7 hello_world.bin | 12:19 |
lkcl | but external_core_top.v must be copied from soc (after the make microwatt_external_core command produces it) into the ls2 directory | 12:20 |
lkcl | and hello_world.bin must be made in microwatt | 12:20 |
Veera[m] | I will check and try! This info feels terse. But let me become familiar, then may get it. | 12:23 |
lkcl | i created a *real* bare-bones page here https://libre-soc.org/HDL_workflow/ls2/ | 12:26 |
lkcl | and if you look at the uoregon nextpnr-xilinx schroot, it is what i did there | 12:26 |
lkcl | ah | 12:26 |
lkcl | forgot to add nextpnr-xilinx to https://libre-soc.org/HDL_workflow/ls2/ will do that now | 12:26 |
lkcl | mostly it is about running all of the devscripts, but in the right order. | 12:28 |
Veera[m] | After trying all this the hello_world.bin bitstream has to be run in uoregon Silicon server and ssh consoled to confirm "Hello World" gets printed? | 12:33 |
lkcl | yes | 12:45 |
lkcl | i already have it done / running so i know that it works | 12:45 |
Veera[m] | So have to update the ls2 wikipage about this or telling it works is enough? | 12:48 |
lkcl | both - by following the ls2 wikipage, someone else needs to be able to do that and get it working | 12:50 |
lkcl | nothing like as sophisticated as this https://shenki.github.io/boot-linux-on-microwatt/ | 12:51 |
Veera[m] | Saw the page is nice! | 12:58 |
Veera[m] | Have you opened any bug report and allotted? budget! | 12:59 |
lkcl | do make a bugreport. let's say EUR 600 again? | 13:07 |
lkcl | yes, shenki's page is nice, that will be another one, another time | 13:07 |
Veera[m] | May I do this work in next batch. What title I have to give for bug report. Which NlNet Milestone? First SoC! | 13:10 |
Veera[m] | Let me see if I can work out this quickly! | 13:10 |
lkcl | yes first soc. if you can do it quickly i can tell the IBM India Education team it is ready for someone to try | 13:23 |
Veera[m] | Bug title: Document steps for ls2 microwatt hello world for fpga boards | 13:37 |
Veera[m] | Is this title Okay? | 13:37 |
lkcl | yes that's great | 13:40 |
lkcl | (titles can always be changed) | 13:40 |
Veera[m] | oh. ok. I will open bug report. And start work in a short while. thanks. | 13:42 |
ghostmansd | Hi folks, a crazy question: do we have some _correct_ mnemonic along with args which occupies more than 4 bytes? | 14:21 |
ghostmansd | I'm cleaning code a bit, and would like to convert any SVP64 instruction to "nop" for a while in gas | 14:22 |
ghostmansd | First, I check for `sv.` prefix; unless present, I pass the string to usual PPC | 14:23 |
ghostmansd | So, I guess, any mnemonic's string repr in our scheme has at least 4 bytes: "sv.X\0", right? | 14:24 |
ghostmansd | And I can simply override the original string with "nop" for now. | 14:24 |
ghostmansd | Obviously until stuff works. :-) | 14:25 |
ghostmansd | Yeah, it seems it works. Oddly I had to write all these thoughts here to realize that my assumptions were correct. | 14:26 |
lkcl | ghostmansd[m], you're the 2nd person to talk to themselves in 24hr :) | 15:45 |
lkcl | it's a magic self-solving IRC channel! | 15:46 |
lkcl | yes it would be perfect idea to put nop. the next trick would be to instead put ".long 0xNNNNNN; op" | 15:47 |
lkcl | ghostmansd[m], there's dozens of examples at the end here https://git.libre-soc.org/?p=openpower-isa.git;a=blob;f=src/openpower/sv/trans/svp64.py;hb=HEAD | 15:54 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!