| *** tpb has joined #yosys | 00:00 | |
| *** emeb has left #yosys | 00:13 | |
| *** emeb_mac has joined #yosys | 00:28 | |
| *** rohitksingh has quit IRC | 01:33 | |
| *** rohitksingh has joined #yosys | 01:55 | |
| *** citypw_ has joined #yosys | 02:09 | |
| *** rohitksingh has quit IRC | 02:13 | |
| *** rohitksingh has joined #yosys | 02:15 | |
| *** rohitksingh has quit IRC | 03:30 | |
| *** nengel has quit IRC | 03:44 | |
| *** nengel has joined #yosys | 03:46 | |
| *** voxadam has quit IRC | 05:02 | |
| *** voxadam has joined #yosys | 05:03 | |
| *** Jybz has joined #yosys | 05:08 | |
| *** rohitksingh has joined #yosys | 05:32 | |
| *** Jybz has quit IRC | 06:33 | |
| *** nengel has quit IRC | 06:48 | |
| *** nengel has joined #yosys | 06:50 | |
| *** emeb_mac has quit IRC | 07:06 | |
| *** adjtm_ has quit IRC | 07:17 | |
| *** jakobwenzel has joined #yosys | 08:14 | |
| *** dys has joined #yosys | 08:48 | |
| *** fsasm has joined #yosys | 08:49 | |
| *** vidbina has joined #yosys | 08:52 | |
| *** stzsch has quit IRC | 10:01 | |
| *** peepsalot has quit IRC | 10:58 | |
| *** finnb has quit IRC | 11:00 | |
| *** finnb has joined #yosys | 11:01 | |
| *** cr1901_modern has quit IRC | 11:33 | |
| *** nengel has quit IRC | 11:38 | |
| *** cr1901_modern has joined #yosys | 11:59 | |
| *** like2wise has joined #yosys | 13:06 | |
| *** vidbina has quit IRC | 14:58 | |
| *** fsasm has quit IRC | 15:06 | |
| *** develonepi3 has joined #yosys | 15:22 | |
| *** az0re has quit IRC | 15:27 | |
| *** vidbina has joined #yosys | 15:37 | |
| *** citypw_ has quit IRC | 15:42 | |
| *** vidbina has quit IRC | 16:16 | |
| *** dh73 has joined #yosys | 16:44 | |
| *** fsasm has joined #yosys | 17:07 | |
| *** rohitksingh has quit IRC | 17:13 | |
| *** dys has quit IRC | 17:37 | |
| *** rohitksingh has joined #yosys | 17:42 | |
| *** rjo has quit IRC | 18:07 | |
| *** promach3 has quit IRC | 18:07 | |
| *** fevv8[m] has quit IRC | 18:07 | |
| *** Thorn has quit IRC | 18:07 | |
| *** emily has quit IRC | 18:08 | |
| *** nrossi has quit IRC | 18:09 | |
| *** jryans has quit IRC | 18:09 | |
| *** pepijndevos[m] has quit IRC | 18:10 | |
| *** rjo has joined #yosys | 18:25 | |
| *** nengel has joined #yosys | 18:25 | |
| *** nrossi has joined #yosys | 18:26 | |
| *** pepijndevos[m] has joined #yosys | 18:26 | |
| *** fevv8[m] has joined #yosys | 18:30 | |
| *** kuldeep has quit IRC | 18:36 | |
| *** rohitksingh has quit IRC | 19:03 | |
| *** jryans has joined #yosys | 19:15 | |
| *** vidbina has joined #yosys | 19:37 | |
| *** fevv8[m] has quit IRC | 19:48 | |
| *** nrossi has quit IRC | 19:48 | |
| *** rjo has quit IRC | 19:48 | |
| *** pepijndevos[m] has quit IRC | 19:48 | |
| ZirconiumX | dh73: I unintentionally reinvented parts of your yosys-bench branch while trying to write something similar-ish | 20:20 |
|---|---|---|
| ZirconiumX | But there are bits that you might want to consider | 20:21 |
| *** emily has joined #yosys | 20:21 | |
| dh73 | Well, thank you. Those scripts was for a very specific tasks but is in the todo list to make them much better, so any input on that will be for sure considered | 20:23 |
| ZirconiumX | It seems your script is more or less "how well does this do in general", whereas mine is "how do these two versions compare?" | 20:24 |
| ZirconiumX | So I'm using what I believe to be the right tool for the job: a sequential probability ratio test | 20:24 |
| *** X-Scale` has joined #yosys | 20:24 | |
| *** X-Scale has quit IRC | 20:25 | |
| dh73 | Pareto? | 20:25 |
| *** X-Scale` is now known as X-Scale | 20:26 | |
| ZirconiumX | Which means I can compare two versions and minimise the average running time while keeping bounded error | 20:26 |
| dh73 | Pareto images between implementation of design space? | 20:26 |
| ZirconiumX | Wald, actually :P | 20:26 |
| ZirconiumX | https://en.wikipedia.org/wiki/Sequential_probability_ratio_test | 20:26 |
| tpb | Title: Sequential probability ratio test - Wikipedia (at en.wikipedia.org) | 20:26 |
| ZirconiumX | (but a generalised version of this) | 20:26 |
| dh73 | I see. Also yes, those scripts are just to compare current implementations without a record. Your approach sounds good. Since Nina added the scratchpad for abc, one of the future implementations of the benchmark suite is to have data of QoR among Yosys versions. I think is a very good tool to have | 20:32 |
| ZirconiumX | As a data point, I can reject ABC9 as at most no better than ABC1 for iCE40 in 49 place-and-routes | 20:34 |
| ZirconiumX | Which is not that much CPU time actually | 20:34 |
| daveshah | > As a data point, I can reject ABC9 as at most no better than ABC1 for iCE40 in 49 place-and-routes | 20:36 |
| daveshah | presumably only for a few designs? | 20:36 |
| ZirconiumX | At the moment I'm only testing picorv32_large.v, so potentially | 20:36 |
| ZirconiumX | It'd be feasible to round-robin the scripts while testing | 20:37 |
| ZirconiumX | Although Yosys seems to crash on things like OpenSPARC | 20:37 |
| daveshah | Certainly my experience of ABC9 for ECP5 is that it made a few designs worse and more designs better | 20:37 |
| daveshah | I have had OpenSPARC work with Yosys in the past | 20:38 |
| daveshah | Is this the T2 variant that is in yosys-bench? | 20:39 |
| ZirconiumX | Yeah | 20:39 |
| daveshah | What synth command? I think I've only tested that one with synth_xilinx | 20:40 |
| daveshah | and that was a few months ago | 20:40 |
| dh73 | Yeah, OpenSPARC have tons of clock domains, so abc gets crazy on them. You need some modifications. I think dave did modifications once. As far as I remember, most of the crashes happened when using `-dff` | 20:40 |
| ZirconiumX | ECP5 I think | 20:41 |
| ZirconiumX | It takes a while to reproduce :P | 20:41 |
| *** nengel has quit IRC | 20:52 | |
| *** nrossi has joined #yosys | 21:00 | |
| *** fevv8[m] has joined #yosys | 21:00 | |
| *** rjo has joined #yosys | 21:00 | |
| *** jryans has joined #yosys | 21:00 | |
| *** promach3 has joined #yosys | 21:00 | |
| *** pepijndevos[m] has joined #yosys | 21:00 | |
| *** fsasm has quit IRC | 21:03 | |
| GenTooMan | when is the main branch of yosys updated? When a version change occurs with a set of bug fix criteria? | 21:09 |
| daveshah | No, almost everything goes into master | 21:11 |
| *** emeb_mac has joined #yosys | 21:15 | |
| *** Thorn has joined #yosys | 21:25 | |
| GenTooMan | no tags then? Ok .. hopefully the master works then. | 21:27 |
| daveshah | There are tags but releases are infrequent unfortunately | 21:30 |
| daveshah | In general the releases are more buggy than master | 21:30 |
| GenTooMan | Ok ... thanks I was wondering if I needed to update yosys as I have 0.9+932 however I haven't an idea of what (build #?) it's at | 21:36 |
| GenTooMan | oh well execute pull and hope for the best :D | 21:44 |
| *** strobokopp has joined #yosys | 22:16 | |
| *** dh73 has quit IRC | 22:49 | |
| *** vidbina has quit IRC | 23:04 | |
| *** likewise has quit IRC | 23:07 | |
| GenTooMan | mm not too happy having to edit yosys makefile every time I build it. | 23:31 |
| *** like2wise has quit IRC | 23:36 | |
| ZirconiumX | GenTooMan, Makefile.conf is a thing | 23:37 |
| ZirconiumX | You can override the Yosys Makefile rules by putting them in Makefile.conf | 23:37 |
| GenTooMan | Well things happen this sway, yosys has a bad habit of assuming (like most applications) where you can install it. So they define the prefix wrong for the linux distribution I have. When I do a "git pull" I have to delete the Makefile so I can update the source. Makefile.conf likely is in the same category no? | 23:42 |
| awygle | Makefile.conf isn't tracked in the repository | 23:43 |
| awygle | so it won't have that problem | 23:43 |
| GenTooMan | I'll check if it allows me to override PREFIX in the makefile | 23:44 |
| *** FL4SHK has quit IRC | 23:45 | |
| *** FL4SHK has joined #yosys | 23:47 | |
| *** FL4SHK has quit IRC | 23:57 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!