01:19 <azonenberg> whitequark: re gtkwave nonsense, wow

01:19 <azonenberg> Would you be interested in a vcd importer for glscopeclient? like, would you actually use it?

01:19 <whitequark> some of the worst code i've ever seen in my life

01:19 <whitequark> hmm

01:19 <azonenberg> it's on the todo right now but not a priority as nobody has indicated it would be super valuable to them

01:20 <whitequark> i don't think it would help much in glscopeclient

01:21 <whitequark> gtkwave, for all its faults, does nail the UI patterns mostly right: a tree view plus a viewer for *lots* of signals

01:22 <azonenberg> Yeah. Improving glscopeclient's scalability to huge numbers of signals is on the todo list but that's not really its specialty at the moment

01:22 <azonenberg> it's mostly targeting analog, signal integrity, and complex serial protocol analysis

02:14 <promach3> mwk: how exactly is DQSBUFM.v used in DDR3 memory ?

09:38 <daveshah> promach3: have a look at the litedram ECP5 DDR PHY. All too many of mine & florent's tears have gone into making the ECP5 DQS primitives play nicely

09:39 <daveshah> very briefly, DQSBUFM handles the DQS input and generates strobe and FIFO pointer signals (for clock domain crossing from DQS to ECLK/SCLK domain) for the DQ IO primitives

09:49 <promach3> daveshah: https://github.com/enjoy-digital/litedram/tree/master/litedram/phy is in python ?

09:49 <daveshah> yes, it's Migen

09:50 <daveshah> https://github.com/ultraembedded/core_ddr3_controller/blob/master/src_v/phy/ecp5/ddr3_dfi_phy.v is a simpler PHY that just uses regular IO and not the DQS stuff, but I haven't personally tested it or even looked at in detail

10:16 <promach3> daveshah: what do you exactly mean by "not the DQS stuff" ?