Thursday, 2019-04-25

*** tpb has joined #symbiflow00:00
digshadowHi hackerfoo. Whats your question?00:00
hackerfoodigshadow: from the picture above of Vivado's routing of a RAM32X1S, it seems it's using DI2 for the input and O6 as the output.00:04
hackerfooThis seems to contradict information in the arch-defs repo.00:05
digshadowWhat is conflicting? Which is input vs output? How the pins are assigned? What does arch-defs have?00:06
hackerfooIt's more of an observation at this point. I'm going to try to change the definitions to get VPR to route it the same way.00:07
digshadowFWIW I almost exclusively dealt with bit definitions. If you have a routing question, I'm probably not the person to ask00:08
digshadowAlso timing00:08
digshadowEx: if you have questions about RAM32X1S bitstream encoding I can provide insight there00:08
hackerfooWell, is this what you expect? The data input line goes to DI2, and the data output to O6.00:11
digshadowhackerfoo: looking at the user guide, two observations. First, the DI naming is very confusing with net DID[0] => pin DID2, net DID[1] => pint DID1. That aside, it notes in a few things (ex: "Figure 2-11: 64 X 3 Simple Dual Port Distributed RAM (RAM64M)") that they go to pin DI1 when only one is used00:17
digshadowIf its unclear though, I'd probably make a minitest with a number of configurations (or using existing minitests) and take a look at the routing in the various configurations00:18
digshadowLet me see whats in the minitests dir...00:19
digshadowYeah if you synthesize prjxray/fuzzers/018-clb-ram/minitest, maybe modifying it to use the commented out top module00:21
digshadowyou should get a bunch of different examples00:21
digshadownvm no modification needed, just synthesize it00:21
*** _whitelogger has quit IRC00:35
*** _whitelogger has joined #symbiflow00:37
hackerfoodigshadow: Thanks. I'll look at it a little later if I can.00:44
*** futarisIRCcloud has joined #symbiflow01:10
*** Bertl_oO is now known as Bertl_zZ02:29
*** citypw has joined #symbiflow03:23
*** tmichalak1 has quit IRC05:33
*** kraiskil has joined #symbiflow05:35
*** proteusguy has quit IRC05:56
*** proteusguy has joined #symbiflow06:14
*** jevinskie has joined #symbiflow06:18
*** proteusguy has quit IRC06:18
*** kraiskil has quit IRC06:20
*** kraiskil has joined #symbiflow06:22
*** proteusguy has joined #symbiflow06:30
*** Vonter has quit IRC06:56
mithrokgugala: Can you take over https://github.com/SymbiFlow/symbiflow-arch-defs/pull/645 ?07:24
tpbTitle: WIP - utils/vlog: More tests from timing tutorial. by mithro · Pull Request #645 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)07:24
sf-slack2<kgugala> sure07:25
sf-slack2<kgugala> what is left there?07:25
sf-slack2<kgugala> Is it only making the CI green, or some features are  missing?07:25
mithrokgugala: It's a bit unclear, I think most of the work is figuring out what they are *suppose* to be doing :-P07:27
sf-slack2<kgugala> ah, OK07:27
mithrokgugala: Regarding https://github.com/SymbiFlow/symbiflow-arch-defs/issues/59907:29
tpbTitle: v2x solve issues to have working tests · Issue #599 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)07:29
mithrokgugala: I don't understand why the $dff cells which are generated by the 'proc' pass are getting ignored?07:31
mithrokgugala: Actually, this https://github.com/SymbiFlow/symbiflow-arch-defs/blob/master/utils/vlog/yosys/json.py#L231-L232 might be the problem...07:31
tpbTitle: symbiflow-arch-defs/json.py at master · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)07:32
mithrotmichalak : I merged https://github.com/SymbiFlow/prjxray/pull/788 as the CI was green07:36
tpbTitle: Disable currently flaky fuzzers to make CI go green again by tmichalak · Pull Request #788 · SymbiFlow/prjxray · GitHub (at github.com)07:36
sf-slack2<tmichalak> @mithro: great, I also solved the problem with 052-pip-clkin ( https://github.com/SymbiFlow/prjxray/pull/799). At least I think so.07:39
tpbTitle: 052-pip-clkin: stabilize and re-enable the fuzzer by tmichalak · Pull Request #799 · SymbiFlow/prjxray · GitHub (at github.com)07:39
sf-slack2<kgugala> @mithro yes, this may be problem07:39
sf-slack2<kgugala> I'll play around and see what is happening there07:40
mithrokgugala: Did you see my comments on https://github.com/SymbiFlow/symbiflow-arch-defs/issues/642 ?07:41
tpbTitle: Make v2x produce the new composable interconnect specification again (IE Revert #637). · Issue #642 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)07:41
sf-slack2<tmichalak> @mithro: currently I am running my tests for 054-pip-fan-alt (https://github.com/SymbiFlow/prjxray/pull/789) and started looking into reenabling 045-hclk-cmt-pips (https://github.com/SymbiFlow/prjxray/issues/790)07:41
tpbTitle: [WIP] 054-pip-fan-alt: stabilize the fuzzer by tmichalak · Pull Request #789 · SymbiFlow/prjxray · GitHub (at github.com)07:41
*** Vonter has joined #symbiflow07:42
sf-slack2<kgugala> @mithro yes, I've seen them07:45
mithrokgugala: Did they make sense?07:45
sf-slack2<kgugala> this makes sense07:46
sf-slack2<kgugala> the question is if that won't break the current flow?07:46
sf-slack2<kgugala> I'm using v2x to generate all the xmls and uipdate them with timings07:47
sf-slack2<kgugala> so introducing this change now may break arch-defs07:47
mithrokgugala: What do you mean?08:00
mithroelms: Look what I found -> https://github.com/mithro/symbiflow-arch-defs/commit/4c996d52d2f7652c42e46e909815e97b8d98a01308:00
tpbTitle: ice40: Adding real world timing values. · mithro/[email protected] · GitHub (at github.com)08:00
sf-slack2<acomodi> mithro: regarding https://github.com/SymbiFlow/symbiflow-arch-defs/issues/642, all the golden pb_types/model should be in the `new` format, am I correct? Than the `xmlsort.xsl` handles the translation to the `old` format to have the interconnects compatible with current VPR08:00
tpbTitle: Issues · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)08:00
mithroacomodi: Yes08:01
sf-slack2<acomodi> mithro: ok, got it, I'll revert the PR then08:01
sf-slack2<kgugala> @mithro In parallel I'm working on bel timings import into arch.xml. There I switched from handwritten xmls to the ones generated from verilog (I discussed that with @elms and @litghost). Now, with v2x handling multiple instances I should be able to finish this08:04
sf-slack2<kgugala> @mithro but if we switch to a new interconnect generation, wouldn't it break the flow?08:04
sf-slack2<kgugala> can vpr handle the new interconnect format?08:05
mithrokgugala: See what acomodi said --  The `xmlsort.xsl` handles the translation to the `old` format to have the interconnects compatible with current VPR08:05
mithrokgugala: Which is done as part of the architecture merging step08:05
sf-slack2<kgugala> oh08:05
sf-slack2<kgugala> OK08:05
sf-slack2<kgugala> I must have missed that08:05
mithrokgugala: VPR doesn't handle <xi:include> either08:05
sf-slack2<acomodi> mithro: here there is the revert PR https://github.com/SymbiFlow/symbiflow-arch-defs/pull/648, it needs some changes though as I have noticed it creates incorrect direct connections08:22
sf-slack2<tmichalak> kokoro seems to be stuck, force-run tag doesn't work, anyone else experiences this problem? I wonder if there is anything else I can do except wait...09:15
sf-slack2<mkurc> @tmichalak I made a PR to arch-defs an hour ago and the kokoro seems to be still running. I don't know if I can see logs of a pending build.09:18
sf-slack2<tmichalak> @mkurc: maybe the problem only applies to new runs only, let me know if your run finishes09:21
sf-slack2<acomodi> Could it be due to the fact that there are multiple PRs that are being run on kokoro?09:21
sf-slack2<tmichalak> @acomodi: actually I start to think that it is indeed the case09:22
sf-slack2<acomodi> for instance I opened a PR as well a half an hour ago09:22
sf-slack2<mkurc> Might be so. It would be nice to have a interface to see load of the build server.09:23
sf-slack2<tmichalak> @acomodi: I noticed that when running multiple PRs for fuzzers stabilization there were some hickups in kokoro's runs09:23
sf-slack2<tmichalak> @mkurc: I agree09:24
*** kraiskil has quit IRC09:53
*** citypw has quit IRC10:00
*** kraiskil has joined #symbiflow10:11
*** auscompgeek has joined #symbiflow10:17
sf-slack2<mkurc> @tmichalak Kokoro finished checking my PR successfully10:32
sf-slack2<tmichalak> @mkurc: okay thanks, so I have 2 PRs running, i.e. one running and the second one about to run with the use of the force-run tag. Unfortunately, nothing changes for 2 hours now.10:34
sf-slack2<acomodi> There is no way right now to run only one of the different checks right? (e.g. XC7-presubmit failed and I want to re-run only that one)10:37
sf-slack2<kgugala> I don't think so10:37
sf-slack2<kgugala> just rerun all of them10:37
*** proteusguy has quit IRC10:38
*** Vonter has quit IRC10:53
*** Vonter has joined #symbiflow10:57
*** proteusguy has joined #symbiflow11:27
*** proteusguy has quit IRC11:28
*** kraiskil has quit IRC11:44
*** kraiskil has joined #symbiflow12:01
*** Bertl_zZ is now known as Bertl12:22
sf-slack2<acomodi> update on equivalent tiles: I have update the PR https://github.com/SymbiFlow/vtr-verilog-to-routing/pull/36. Now the placer can swap the blocks in equivalent tiles as well. CI still fails as the arch files do not include the tiles tag.13:10
tpbTitle: Sign in to GitHub · GitHub (at github.com)13:10
sf-slack2<acomodi> By running the `ff_ce_sr` tests I got into a fasm2frames issue13:11
sf-slack2<acomodi> `prjxray.fasm_assembler.FasmInconsistentBits: FASM line "CLBLM_R_X5Y117.SLICEL_X1.DOUTMUX.O5" wanted to clear bit (131743, 35, 21) but was set by FASM line "CLBLM_R_X5Y117.SLICEL_X1.DOUTMUX.D5Q"`13:11
sf-slack2<acomodi> the tests that work right now are: `chain_packing`, `simple_ff` and `buttons`13:12
*** Bertl is now known as Bertl_oO13:29
hackerfooI was reading through the CLB user guide last night that the software is responsible for setting A6 high in 32 bit RAM modes to keep O5/O6 independent.13:57
hackerfooI also made VPR use O5 and got a different alternating pattern.13:59
*** kraiskil has quit IRC14:23
*** kraiskil has joined #symbiflow14:36
sf-slack2<acomodi> `counter` also is working15:21
*** citypw has joined #symbiflow15:29
*** proteusguy has joined #symbiflow15:33
*** citypw has quit IRC16:18
*** jevinskie has quit IRC16:21
*** _whitelogger has quit IRC17:29
*** _whitelogger has joined #symbiflow17:31
hackerfooDoes anyone have a partial reconfiguration license for Vivado? Any plans for SymbiFlow to do partial reconfiguration?18:46
hackerfooNevermind, it's mentioned right here: https://symbiflow.github.io/#xray18:49
tpbTitle: SymbiFlow - the GCC of FPGAs (at symbiflow.github.io)18:49
*** kraiskil has quit IRC18:49

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!