| *** tpb <[email protected]> has joined #symbiflow | 00:00 | |
| -_whitenotifier-700a- [f4pga-arch-defs] kazkojima opened issue #2463: VPR fails to place long carry chains on xc7 - https://github.com/SymbiFlow/f4pga-arch-defs/issues/2463 | 09:22 | |
| tpb | <marzoul> Hi, I am trying to launch prjxray on a virtex-7 device, just for the fun or it ;-) | 14:55 |
|---|---|---|
| tpb | <marzoul> It needs some tweaking in many scripts tcl / py, making fuzzers steps pass one by one | 14:55 |
| tpb | <marzoul> Then I hit an error in fuzzer 005-tilegrid that I don't actually know what could be the source | 14:58 |
| tpb | <marzoul> I believe it is relatively at end of fuzzer005 | 14:58 |
| tpb | <marzoul> The script inadd_tdb.py fails because it is trying to import these ill-formed frame coordinates (debug print added by me): | 14:58 |
| tpb | <marzoul> Issue with: CMT_TOP_L_LOWER_B_X250Y113.DWORD:29.DBIT:15.DFRAME:1d <const1> <5259 candidates> | 14:58 |
| tpb | <marzoul> Issue with: CMT_TOP_L_UPPER_T_X250Y148.DWORD:24.DFRAME:1c <const0> | 14:58 |
| tpb | <marzoul> Issue with: CLK_HROW_BOT_R_X137Y130.DWORD:6.DFRAME:1a <const0> | 14:58 |
| tpb | <marzoul> Issue with: HCLK_CMT_L_X250Y130.DWORD:5.DFRAME:1c <const0> | 14:58 |
| tpb | <marzoul> Issue with: HCLK_IOI3_X1Y130.DWORD:0.DFRAME:21 <const0> | 14:58 |
| tpb | <marzoul> If this rings a bell, please share candidate root causes :-) | 15:54 |
| tpb | <mithro> @marzoul Can you log a bug on GitHub? Then someone who has more experience like acomodi or mkurc can probably respond with help if they understand what you are seeing | 17:44 |
| tpb | <mithro> @marzoul The people who were recently doing work to get kintex 7 going might also have some advice here -- they all went and hid on discord or something, maybe someone can point you in their direction (I know they have commented on GitHub issues recently). | 17:45 |
| tpb | <sf-slack> <pgielda> Also, would be great to move discussions to CHIPS Alliance slack (#f4pga channel) or #f4pga on libera which is bridged to it. We plan to retire this slack instance at some point. | 17:59 |
| tpb | <mithro> We should be able to set up a redirect from this channel to #f4pga IRC channel | 19:04 |
| tpb | <sf-slack> <pgielda> It would have to be both ways (#f4pga <-> symbiflow). But its easier to just switch... Otherwise its just confusing... | 19:07 |
| tpb | <mithro> pgielda: Not a bridge, a redirect -- IE when you try and join the #symbiflow IRC channel it sends you to the #f4pga IRC channel | 19:08 |
| tpb | <sf-slack> <pgielda> This makes sense! | 19:23 |
| tpb | <sf-slack> <torsten.reuschel> @marzoul: How did you set up the region of interest for your tilegrid, e.g. https://github.com/f4pga/prjxray/blob/master/settings/kintex7.sh#L13? @hansfbaier recently noted that the ROI coordinates need to be defined are the _site coordinates_, not the _tile coordinates_. | 19:57 |
| tpb | <sf-slack> <torsten.reuschel> @mithro I sure hope that no one went hiding.. the blinky repo at https://github.com/kintex-chatter/xc7k325t-blinky-nextpnr seems to have attracted attention and contributions from additional people interested specifically in Kintex7 support. My understanding and intent of contributing over there is to make the repo superfluous eventually. I see it as a playground for all levels of open-source FPGA tools from | 20:04 |
| tpb | <sf-slack> which to launch Kintex7-specific pull requests to the upstream projects. Have I missed an equivalent location/community/repo at Symbiflow/F4PGA? | 20:04 |
| tpb | <mithro> torsten.reuschel: I think we are totally happy for discussion about these topics to be had directly on the #symbiflow / #f4pga channels like here on IRC. Also would be more than happy to enable / create something on GitHub under these organizations if people want too | 20:06 |
| tpb | <mithro> Test test | 20:41 |
| tpb | <mithro> People are also *very* welcome to have discussions about nextpnr in this channel and on other SymbiFlow spaces. The interchange work is happening on both nextpnr and vpr in parallel too. We are all pretty friendly, for example Google+Lattice+Antmicro+gatecat collaborating on the Crosslink NX support in nextpnr for a ChromeOS project which needed a fully open source toolchain. | 20:41 |
| tpb | <mithro> Centos 8 was End of Lifed in December 2021 (see https://www.centos.org/centos-linux-eol/) but surprisingly Centos 7 will get maintenance updates until 2024-06-30! | 21:13 |
| tpb | <sf-slack> <hansfbaier> @mithro I am currently working on prjxray support for Kintex's high performance banks. I have a 030-iob18 fuzzer that already runs and delivers results, but all the candidates are empty, and I believe the reason for this is, that bit2fasm does not disassemble many of the bits relevant to that fuzzer: | 22:37 |
| tpb | <sf-slack> <hansfbaier> ```# In frame 0x004424a6 186 bits were not converted. { unknown_bit = "004424a6_2_0", unknown_segment = "0x00442480", unknown_segbit = "38_64" } { unknown_bit = "004424a6_2_12", unknown_segment = "0x00442480", unknown_segbit = "38_76" } { unknown_bit = "004424a6_2_16", unknown_segment = "0x00442480", unknown_segbit = "38_80" } { unknown_bit = "004424a6_2_24", unknown_segment = "0x00442480", unknown_segbit = "38_88" | 22:37 |
| tpb | <sf-slack> } { unknown_bit = "004424a6_2_26", unknown_segment = "0x00442480", unknown_segbit = "38_90" } { unknown_bit = "004424a6_2_30", unknown_segment = "0x00442480", unknown_segbit = "38_94" } { unknown_bit = "004424a6_3_0", unknown_segment = "0x00442480", unknown_segbit = "38_96" } { unknown_bit = "004424a6_3_2", unknown_segment = "0x00442480", unknown_segbit = "38_98" } { unknown_bit = "004424a6_3_14", unknown_segment = "0x00442480", | 22:37 |
| tpb | <sf-slack> unknown_segbit = "38_110" } { unknown_bit = "004424a6_3_18", unknown_segment = "0x00442480", unknown_segbit = "38_114" } { unknown_bit = "004424a6_3_30", unknown_segment = "0x00442480", unknown_segbit = "38_126" } { unknown_bit = "004424a6_4_14", unknown_segment = "0x00442480", unknown_segbit = "38_142" } { unknown_bit = "004424a6_4_16", unknown_segment = "0x00442480", unknown_segbit = "38_144" } { unknown_bit = "004424a6_4_30", | 22:37 |
| tpb | <sf-slack> unknown_segment = "0x00442480", unknown_segbit = "38_158" } { unknown_bit = "004424a6_5_30", unknown_segment = "0x00442480", unknown_segbit = "38_190" } { unknown_bit = "004424a6_6_0", unknown_segment = "0x00442480", unknown_segbit = "38_192" }``` | 22:37 |
| tpb | <sf-slack> <hansfbaier> @mithro This looks like a chicken-and-egg problem. How am I supposed to get out of that? | 22:38 |
| tpb | <mithro> hansfbaier: Log a GitHub issue and hopefully someone like acomodi / mkurc or similar can respond with what you need to do -- I'm afraid I don't know myself | 22:49 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!