00:46 <cr1901> whitequark[cis]: Approx how many LUTs/FFs does a minimal 1xSPI stream of yours take?

00:46 <whitequark[cis]> this stream-based PHY is actually QSPI-only

00:46 <whitequark[cis]> i.e. it does not have an SPI-only option; it can always be used in 4X mode at runtime (and 2X)

00:48 <cr1901> How many LUTs/FFs does it take in 4xSPI mode then?

00:48 <whitequark[cis]> i haven't synthesized it yet

00:48 <whitequark[cis]> this is all done in simulation

00:49 <whitequark[cis]> debugging pipelining issues where you aim for 100% cycle utilization on hardware is so difficult as to be practically unviable

00:49 <cr1901> If it's small enough, I have an idea to replace the contrived sentinel demo I have in repo w/ something more useful (music player)

00:50 <whitequark[cis]> it's very heavily optimized

00:52 <cr1901> Just hit the SPI flash hard enough to stream in samples, manipulate them, and output to a delta sigma DAC (or if I'm incredibly lazy, bitbang I2S, and make the user provide the sound device)

00:53 <whitequark[cis]> ooh, right

00:53 <whitequark[cis]> I2S is another excellent candidate for this methodology

00:58 <whitequark[cis]> okay, i'm done with my re-engineered IOStream, QSPIEnframer, and QSPIDeframer

00:58 <whitequark[cis]> time to make QSPIClocker and QSPIBusController

00:59 <whitequark[cis]> or maybe QSPIClocker as a separate module doesn't make sense... hm

01:16 <_whitenotifier-9> [amaranth] whitequark opened issue #1416: Add a way to have fine grained control over `oe` for `io.Buffer`, `io.FFBuffer`, `io.DDRBuffer` - https://github.com/amaranth-lang/amaranth/issues/1416

01:19 <_whitenotifier-9> [amaranth] whitequark opened issue #1417: Add a `PortLike` that can be used in simulation to test designs which instantiate I/O buffers - https://github.com/amaranth-lang/amaranth/issues/1417

01:29 <whitequark[cis]> huh, clocking was surprisingly easy

01:30 <whitequark[cis]> https://paste.debian.net/1321024/

02:27 <whitequark[cis]> the new design works perfectly

02:29 <whitequark[cis]> that's running a command and then doing an X4 read without a single wasted cycle

02:36 <tpw_rules> nyaaaaan

02:36 <tpw_rules> but in utf16?

02:37 <whitequark[cis]> watch valid

02:37 <tpw_rules> or is that like good and bad nybble alignment

02:37 <whitequark[cis]> https://gist.github.com/whitequark/5ef7f42c82da32331f7bf72e2d78ceb1

02:38 <tpw_rules> how is the new testbench api working out

02:39 <whitequark[cis]> it's excellent

02:40 <whitequark[cis]> there are a few new pitfalls (mainly, if you are doing assert ctx.get(...) == ... you must do so before any ctx.set() in case you have comb feedback in the design, and you also must use ctx.tick().sample() if you want to interact with streams

02:40 <tpw_rules> what about streams causes the last thing?

02:40 <whitequark[cis]> it's not stream-specific, it's also about comb feedback

02:41 <tpw_rules> don't you have to sample signals specifically? i guess the stream you would sample?

02:41 <whitequark[cis]> it's just that you can get away with sampling at only approximately the right time in many cases (like flags) but you have to get things right if you have a dense flow of data through a stream

02:43 <tpw_rules> are these in the docs yet?

02:43 <whitequark[cis]> what are "these"?

02:43 <tpw_rules> these new pitfalls

02:43 <whitequark[cis]> that sounds like a work question. I'm on vacation so you won't get an answer

02:44 <tpw_rules> fair enough

02:44 <tpw_rules> excited to try it out

02:52 <whitequark[cis]> nice :3

05:50 <whitequark[cis]> the QSPI gateware works very well :D

06:12 <Darius> whitequark[cis]: victory!

06:14 <whitequark[cis]> the main annoyance was that the w25q32 doc lied to me

06:24 <Darius> impossible

15:42 <cr1901> https://github.com/cr1901/smolarith/actions/runs/9540185140/job/26291567203?pr=17#step:6:91 Oops lol

15:43 <cr1901> (Python by default will suppress DeprecationWarning outside of main. This includes doctests. Putting "warnings.simplefilter("default")" in my sphinx conf.py fixes this

15:43 <cr1901> I def welcome other solutions tho

18:27 <_whitenotifier-9> [yosys] qarlosalberto commented on issue #35: frontend error - https://github.com/YoWASP/yosys/issues/35#issuecomment-2183249369