--- Log opened Fri May 03 00:00:42 2013 | ||
@juliusb | stekern: apologies, fire at will | 10:02 |
---|---|---|
@juliusb | (apologies for delay that is) | 10:02 |
mor1kx | [mor1kx] skristiansson pushed 6 new commits to master: https://github.com/openrisc/mor1kx/compare/ba04fb906c86...82b9bf12f292 | 10:42 |
mor1kx | mor1kx/master 9194bab Stefan Kristiansson: move all decode to execute signal registering to a seperate module... | 10:42 |
mor1kx | mor1kx/master 2328a3f Stefan Kristiansson: mor1kx-defines: fix typo in OR1K_ALU_OPC_SECONDARY_SELECT | 10:42 |
mor1kx | mor1kx/master 8e8e258 Stefan Kristiansson: decode: break down immediate decoding into seperate signals | 10:42 |
stekern | juliusb: np | 10:43 |
stekern | you need a patch for mor1kx-devenv with that SECONDARY_SELECT commit, expect to see a pull request soonish | 10:43 |
stekern | done | 10:54 |
stekern | this mux is nasty: https://github.com/openrisc/mor1kx/blob/master/rtl/verilog/mor1kx_ctrl_cappuccino.v#L757 | 11:07 |
stekern | that alone takes about 250 LC, so about 5% of a full blown cappuccino | 11:08 |
@juliusb | wow, try the whole {decode_condition{32}}&selected_thing | 13:23 |
@juliusb | all ORed together | 13:23 |
@juliusb | (as opposed to case statement, I see this a lot - I guess maybe it's less resource intensive?) | 13:23 |
stekern | where decoded_condition is spradr==OR1K... | 14:50 |
stekern | ? | 14:51 |
@juliusb | yea | 15:04 |
stekern | mmm, that would become a couple of xnors, ors and ands | 15:26 |
stekern | i'll give that a go | 15:27 |
stekern | should be {32{decode_condition}} of course, but it only saved 10 LC | 18:55 |
stekern | and it is a lot uglyer | 18:55 |
stekern | uglier too | 18:55 |
stekern | another option would be to put it in a ROM | 18:56 |
stekern | that would be 0 LC in FPGAs, but would occupy blockram | 18:56 |
stekern | ah, but the sr and epcr are there too.. | 19:02 |
--- Log closed Sat May 04 00:00:43 2013 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!