| *** tpb has joined #symbiflow | 00:00 | |
| *** proteusguy has quit IRC | 00:58 | |
| *** _whitelogger has quit IRC | 01:06 | |
| *** _whitelogger has joined #symbiflow | 01:09 | |
| *** proteusguy has joined #symbiflow | 01:12 | |
| *** proteusguy has quit IRC | 01:28 | |
| *** proteusguy has joined #symbiflow | 01:41 | |
| *** craigo has joined #symbiflow | 02:01 | |
| *** davidw has joined #symbiflow | 03:11 | |
| *** davidw is now known as Guest54885 | 03:12 | |
| *** davidthings has quit IRC | 03:12 | |
| *** Guest54885 has quit IRC | 03:55 | |
| *** Guest54885 has joined #symbiflow | 03:56 | |
| *** Guest54885 has quit IRC | 03:59 | |
| *** davidthings has joined #symbiflow | 04:00 | |
| *** davidthings has quit IRC | 05:50 | |
| *** kraiskil has joined #symbiflow | 06:19 | |
| *** OmniMancer has joined #symbiflow | 06:20 | |
| *** Bertl_oO is now known as Bertl_zZ | 06:33 | |
| *** freemint has joined #symbiflow | 07:34 | |
| -_whitenotifier- [symbiflow-arch-defs] mkurc-ant opened issue #1119: Add a simulation model for the PLLE2_ADV - https://git.io/JezUw | 09:25 | |
| *** celadon has quit IRC | 09:41 | |
| *** celadon has joined #symbiflow | 09:43 | |
| *** kraiskil has quit IRC | 10:05 | |
| *** proteusguy has quit IRC | 10:34 | |
| *** proteusguy has joined #symbiflow | 10:34 | |
| *** proteus-guy has joined #symbiflow | 10:35 | |
| *** proteus-guy has quit IRC | 10:37 | |
| *** proteusguy has quit IRC | 10:37 | |
| *** proteusguy has joined #symbiflow | 10:37 | |
| *** zeigren has quit IRC | 10:42 | |
| *** hzeller[m] has quit IRC | 10:42 | |
| *** mrhat2010[m] has quit IRC | 10:42 | |
| *** nrossi has quit IRC | 10:42 | |
| *** galv[m] has quit IRC | 10:43 | |
| *** alexhw[m] has quit IRC | 10:43 | |
| *** synaption[m] has quit IRC | 10:43 | |
| *** xobs has quit IRC | 10:43 | |
| *** lromor[m] has quit IRC | 10:43 | |
| *** proteus-guy has joined #symbiflow | 10:43 | |
| *** clacktronics has quit IRC | 11:04 | |
| *** synaption[m] has joined #symbiflow | 11:07 | |
| *** clacktronics has joined #symbiflow | 11:12 | |
| *** clacktronics has joined #symbiflow | 11:14 | |
| *** clacktronics has quit IRC | 11:31 | |
| *** clacktronics has joined #symbiflow | 11:32 | |
| *** kraiskil has joined #symbiflow | 11:48 | |
| *** zeigren has joined #symbiflow | 11:49 | |
| *** hzeller[m] has joined #symbiflow | 11:49 | |
| *** galv[m] has joined #symbiflow | 11:49 | |
| *** nrossi has joined #symbiflow | 11:49 | |
| *** xobs has joined #symbiflow | 11:49 | |
| *** alexhw[m] has joined #symbiflow | 11:49 | |
| *** mrhat2010[m] has joined #symbiflow | 11:49 | |
| *** lromor[m] has joined #symbiflow | 11:49 | |
| *** citypw has joined #symbiflow | 12:08 | |
| *** freemint has quit IRC | 13:30 | |
| *** davidthings has joined #symbiflow | 13:33 | |
| *** freemint has joined #symbiflow | 13:45 | |
| *** Bertl_zZ is now known as Bertl | 14:38 | |
| *** OmniMancer has quit IRC | 15:44 | |
| *** freemint has quit IRC | 15:53 | |
| hackerfoo | litghost: I hadn't tested FILL_LIMIT yet; commenting that out at the top avoids the routing error. | 16:06 |
|---|---|---|
| litghost | Ah, sure | 16:07 |
| hackerfoo | But then it ran all night, so I need to diagnose that. | 16:07 |
| *** freemint has joined #symbiflow | 16:08 | |
| hackerfoo | I'm going to test without BREAK_ON_MISS, but that has been tested and worked before the rebase. | 16:09 |
| hackerfoo | I changed the hash algorithm too, but bad hashes shouldn't affect the result. | 16:12 |
| *** citypw has quit IRC | 16:33 | |
| *** david__ has joined #symbiflow | 16:33 | |
| *** davidthings has quit IRC | 16:34 | |
| *** david__ has quit IRC | 16:51 | |
| *** Bertl is now known as Bertl_oO | 17:13 | |
| *** freemint has quit IRC | 17:31 | |
| hackerfoo | litghost: Were there any recent changes that might affect routing speed? | 18:40 |
| hackerfoo | (upstream) | 18:40 |
| litghost | hackerfoo: Nothing obvious, why? | 18:40 |
| hackerfoo | It seems much slower, and I don't know why yet. | 18:41 |
| litghost | Slower with the same heap push count? | 18:42 |
| litghost | Or are you getting more heap push's? | 18:42 |
| litghost | And by what factor are we talking about? | 18:42 |
| litghost | 10% or 2x/3x/4x? | 18:42 |
| hackerfoo | Last night it took 10x per iteration without any major changes to the lookahead. | 18:44 |
| litghost | Do you have 10x heap push's? Or just 10x slower with same # of heap pushs? | 18:44 |
| hackerfoo | I think it was about 1.5x heap pushes. | 18:45 |
| litghost | That's pretty odd. If you recompile VTR at the previous git commit and use the same rrgraph and lookahead, does the behavior revert? | 18:46 |
| litghost | At the end of the day, the route time should only be affected by: | 18:47 |
| litghost | the circuit and it's pack/place | 18:47 |
| litghost | the lookahead | 18:47 |
| litghost | the rrgraph | 18:47 |
| litghost | VTR itself | 18:47 |
| litghost | You can fix all but one and retest (though you might need to supply "--verify_file_digests off" | 18:48 |
| litghost | The arch.xml is in there too, but I doubt it's changing at all | 18:49 |
| hackerfoo | Okay, thanks. I'll look into it more. | 18:49 |
| *** proteus-guy has quit IRC | 19:32 | |
| lromor[m] | Hi, my makefile is not working anymore when running: | 20:26 |
| lromor[m] | ``` | 20:26 |
| lromor[m] | ``` | 20:26 |
| * lromor[m] sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/FsVMjjQPcOYeaPlTHrdSzxNB > | 20:27 | |
| litghost | There is a defect in the CMake dependencies around python libraries, solution is to remove <build_dir>/xc7/archs/artix7/channels.db | 20:33 |
| litghost | Sorry for the confusion | 20:33 |
| litghost | Of course maybe I'm misunderstanding what you are trying to do | 20:33 |
| litghost | That particular error would arise if the arch.xml and rrgraph.xml got out of sync | 20:34 |
| lromor[m] | I' | 20:35 |
| litghost | That message was truncated, I don't follow what you were asking for | 20:35 |
| lromor[m] | * When generating the packed netlist (.net) I'm using those the arch timings and the rr graph. I guess there's a mismatch. | 20:36 |
| litghost | The mismatch inidicated in that error is a mismatch between the arch.xml and rrgraph.xml | 20:36 |
| litghost | I recommend re-running the following CMake target "file_xc7_archs_artix7_devices_rr_graph_xc7a50t-arty-swbut_test.lookahead.bin" | 20:37 |
| lromor[m] | I'm not sure at which point I should remove the channels.db as it's not a dependency of my vpr command. | 20:37 |
| litghost | I would remove channels.db and then rebuild up to "file_xc7_archs_artix7_devices_rr_graph_xc7a50t-arty-swbut_test.lookahead.bin" | 20:37 |
| litghost | Then take the new arch.xml/rrgraph.xml/lookaheads | 20:38 |
| lromor[m] | ok | 20:38 |
| litghost | Does that make sense? | 20:38 |
| lromor[m] | Yup! Thanks! | 20:38 |
| *** craigo has quit IRC | 20:39 | |
| litghost | One thing we will likely be adding the in next couple weeks once the ROI breakout is done is some kind of "make install" that exports the files needed to run VPR against various circuits | 20:39 |
| litghost | Outside of the symbiflow CMake system | 20:39 |
| litghost | I imagine that will make your flow easier | 20:40 |
| lromor[m] | That would ideal! Thank you for your work! | 20:46 |
| *** kraiskil has quit IRC | 21:55 | |
| lromor[m] | litghost: That fixed the issue! Thanks | 21:56 |
| litghost | Great! | 21:56 |
| *** kraiskil has joined #symbiflow | 22:24 | |
| *** freemint has joined #symbiflow | 22:27 | |
| *** kraiskil has quit IRC | 22:28 | |
| *** Bertl_oO is now known as Bertl_zZ | 22:34 | |
| hackerfoo | litghost: Routing works now, but it's still a little under 2x slower than before for some reason. I'm going to try to get the PR out today. | 23:00 |
| litghost | Sounds good | 23:01 |
| litghost | I have my debugging tools ready to go and tommorow hopefully I'll be able to see problems | 23:02 |
| mithro | litghost: Does this look good? https://github.com/SymbiFlow/prjxray-db/compare/master...mithro:master | 23:20 |
| tpb | Title: Comparing SymbiFlow:master...mithro:master · SymbiFlow/prjxray-db · GitHub (at github.com) | 23:20 |
| hackerfoo | litghost: The PR is ready for review. There are some things you can tweak at the top of connection_box_lookahead_map.cpp. The things we decide to keep can be made into flags. | 23:30 |
| litghost | mithro: https://github.com/SymbiFlow/prjxray-db/compare/master...mithro:master#diff-55623f36146a8aa2a84d52b0106f499aR1287 is wrong | 23:30 |
| tpb | Title: Comparing SymbiFlow:master...mithro:master · SymbiFlow/prjxray-db · GitHub (at github.com) | 23:30 |
| hackerfoo | There's also some nice ASCII-art of the cost maps that I've left on for now. | 23:30 |
| litghost | mithro: Everything else looks okay | 23:31 |
| litghost | hackerfoo: How does murax first iteration look with https://github.com/SymbiFlow/vtr-verilog-to-routing/pull/310 ? | 23:35 |
| tpb | Title: Parallel connection box lookahead by HackerFoo · Pull Request #310 · SymbiFlow/vtr-verilog-to-routing · GitHub (at github.com) | 23:35 |
| mithro | litghost: The INT_L.IMUX_L33.LOGIC_OUTS_L18 change in that pull request doesn't look right? | 23:37 |
| litghost | mithro: Ya, there was no change that should have caused that | 23:38 |
| litghost | mithro: So it's some form of instability | 23:38 |
| hackerfoo | litghost: I'm currently running it, but last time it ran for 1188s with 2.4e9 heap pushes and a CPD of 11.809. This was with BREAK_ON_MISS = true, so I'm running without that since I don't trust that optimization yet. | 23:39 |
| hackerfoo | It'll probably be the same, though. | 23:39 |
| litghost | hackerfoo: Okay, that matches about what I was getting with your wip-testing @ 0477b874ad1be65626c99ae13f707586a7dae116 | 23:41 |
| hackerfoo | The fill now follows a diamond pattern to match the Manhattan distance, and penalizes based on the distance. The max_fill's (sizes of largest hole) are still surprizingly high. | 23:41 |
| hackerfoo | So increasing sampling from 3x3 might help, because there is a lot of guessing right now. | 23:42 |
| litghost | hackerfoo: I don't expect increasing the sample grid will change much actually, as the holes are likely a function of the graph structure | 23:43 |
| litghost | hackerfoo: I do think a 2D interpolation may be better suited to hole filling, but I'm waiting to get data to support that | 23:43 |
| mithro | litghost: Should I just revert those changes? | 23:44 |
| hackerfoo | Turning on FILL_LIMIT limits the amount of filling, but then I get an error routing GND to a LUT. | 23:44 |
| litghost | mithro: Just the ones to INT_L | 23:44 |
| hackerfoo | Even setting it at 30. | 23:44 |
| litghost | hackerfoo: It's interesting, I didn't see the same issue on the prevoius search, but I cannot explain why yet | 23:45 |
| litghost | hackerfoo: I was able to use a fill limit of 3 and it worked | 23:45 |
| hackerfoo | Maybe I'll try it again, because I fixed some things. | 23:46 |
| hackerfoo | This might be useful to someone else: https://github.com/HackerFoo/hackerfoo-go-functions/blob/master/functions/symbiflow#L734-L745 | 23:54 |
| tpb | Title: hackerfoo-go-functions/symbiflow at master · HackerFoo/hackerfoo-go-functions · GitHub (at github.com) | 23:54 |
| hackerfoo | It lists all the wip branches that will and will not merge cleanly. | 23:55 |
| hackerfoo | It doesn't detect conflicts between them, though. | 23:56 |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!