13:38 <sf-slack2> <acomodi> litghost, mithro: I have opened a PR to change the VTR mode selection mechanism in symbiflow (https://github.com/SymbiFlow/vtr-verilog-to-routing/pull/50) and I have also updated the current PR in VTR mainline as well (https://github.com/verilog-to-routing/vtr-verilog-to-routing/pull/517)

13:38 <tpb> Title: mode selection: check modes while committing rt nets by acomodi · Pull Request #50 · SymbiFlow/vtr-verilog-to-routing · GitHub (at github.com)

13:39 <sf-slack2> <acomodi> litghost, mithro: I am running the titan benchmarks to get to see how better the new solution is w.r.t. the backtraced check of `Differing modes`

14:36 <sf-slack2> <acomodi> I had troubles with running picosoc on HW using https://github.com/verilog-to-routing/vtr-verilog-to-routing/pull/559, but it turns out to be a timing issue, by adding an additional timing divider it now works

14:55 <sf-slack2> <acomodi> *clock divider

15:43 <elms> tiempo of day dobry

15:52 <mithro> mkurc / kgugala: Morning?

15:52 <mithro> Morning everyone...

15:52 <sf-slack2> <mgielda> morning!

15:53 <mithro> mkurc: I think your FASM attribute stuff is potentially workable but we need to land https://github.com/SymbiFlow/symbiflow-arch-defs/pull/703 first I think?

15:53 <tpb> Title: mux_gen + v2x: Support generating FASM annotations for muxes. by mithro · Pull Request #703 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)

15:56 <mithro> Meeting now, if your around afterwards I would like to sync up on the status of it

16:11 <sf-slack2> <kgugala> @mithro: morning

16:44 <duck2> morning

16:54 <sf-slack2> <mkurc> Morning!

16:56 <sf-slack2> <mkurc> @mithro Changes to muxgen and v2x are somewhat independent. I still need to make final fixes to make generated xmls pass the recently merged vpr test.

16:57 <mithro> mkurc: I did a pretty major refactor ot v2x to support the FASM values for muxgen output

17:14 <sf-slack2> <mkurc> @mithro I see now that I still have some unresolved issues in discussion with @litghost about my implementation of the v2x FASM support. I hope I'll be able to resolve all that by tomorrow.

17:14 <mithro> mkurc: I think for now it might be a good idea to stop work on it until I can resolve conflicts with my pull request and yours

17:15 <sf-slack2> <mkurc> @mithro Yep. But I see that my changes to the v2x are not so extensive. It might be easier to merge mine to yours.

17:15 <mithro> mkurc: Yes, I'll will do that

17:16 <mithro> mkurc: I want to get the pack pattern stuff working first

17:16 <sf-slack2> <mkurc> @mithro Ok.

17:17 <mithro> mkurc: Could you maybe look at finishing off https://github.com/SymbiFlow/symbiflow-arch-defs/pull/645/files ?

17:17 <tpb> Title: WIP - utils/vlog: More tests from timing tutorial. by mithro · Pull Request #645 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)

17:18 <mithro> mkurc: Looks like most of the stuff needed for that is filling out the documentation and figuring out why the test fails?

17:18 <sf-slack2> <mkurc> @mithro Ok, I can look into that.

20:16 <mithro> litghost: So, I'm just looking at pack-pattern generation at the moment

20:16 <mithro> litghost: Were are the most complicated pack-patterns in the xc7 target?

20:17 <litghost> mithro: I don't know if its the most complicated, but I'd start with the CE VCC and SR GND pack patterns

20:17 <litghost> mithro: They are awful when written by hand, and I would love to move them to something better

20:18 <litghost> mithro: Most of the other pack patterns aren't too bad

20:18 <litghost> mithro: https://github.com/SymbiFlow/symbiflow-arch-defs/blob/master/xc7/primitives/common_slice/common_slice.pb_type.xml#L73

20:18 <tpb> Title: symbiflow-arch-defs/common_slice.pb_type.xml at master · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)

20:18 <litghost> mithro: Ideally the pack pattern would be generated inside of a generate statement

20:19 <litghost> mithro: I was thinking as wire attributes, but I haven't really looked into it

20:20 <mithro> litghost: They track into https://github.com/SymbiFlow/symbiflow-arch-defs/blob/master/xc7/primitives/ff/ff.pb_type.xml#L69-L70 ?

20:20 <tpb> Title: symbiflow-arch-defs/ff.pb_type.xml at master · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)

20:20 <litghost> mithro: Yep

20:21 <litghost> mithro: Ideally we could annotate the sink and source of the pack pattern, and annotate the in between

20:21 <litghost> mithro: But I'm open to other suggestions

20:22 <mithro> litghost: This pack pattern starts on the SR_GND output and ends on the Flip-Flop S input?

20:23 <litghost> mithro: Or CE_VCC and ends at CE

20:23 <litghost> mithro: Yes

20:23 <mithro> litghost: And there are multiple ones because of the multiple modes that the FF can be in?

20:23 <litghost> mithro: There are 3 patterns per FF type, CE=VCC, SR=GND, and CE=VCC+SR=GND

20:24 <litghost> mithro: Two dimensions, FF blackbox type (e.g. FDRE vs FDSE) and pack combination (CE=VCC / SR=GND / CE=VCC+SR=GND)

20:24 <litghost> mithro: That last one (CE=VCC+SR=GND), it is unclear if that patterns is actually required, or whether the component parts are sufficient

20:27 <mithro> litghost: Probably as things can only be part of one pack-pattern...

20:27 <mithro> Be back in 30mins, meeting

20:27 <litghost> mithro: Okay, good luck

21:15 <mithro> litghost: There was something like "specialize carry chains" or similar somewhere?

21:16 <litghost> mithro: Still is

21:16 <litghost> mithro: I wanted to wait until we had timing results in to determine course of action on the specialization

21:16 <mithro> litghost: Ahh, it's not xc7 specific, its in utils/specialize_carrychains.py

21:16 <mithro> litghost: What does it do?

21:17 <litghost> mithro: Ensures carry chain names are unique based on their parentage

21:17 <litghost> mithro: So the SLICEM carry chain has a different pattern than the SLICEL carry chain

21:18 <litghost> mithro: It is unclear that we actually want the carry chain in SLICEM to be used over the SLICEL, except when LUT-RAM's or SRL are required

21:18 <litghost> mithro: Equiv tiles enable putting a SLICEL cluster at a SLICEM tile so we are limiting our resources

22:37 <mithro> Does anyone need me to review anything?

23:01 <litghost> elms requested a second reviewer on https://github.com/SymbiFlow/symbiflow-arch-defs/pull/708

23:01 <tpb> Title: Initial xc7 route timing import. by litghost · Pull Request #708 · SymbiFlow/symbiflow-arch-defs · GitHub (at github.com)