Time is already telling. Specifically, the time now (second week of October) versus Cointerra's tape-out target date ("first week of October") is very telling.
I notice you never explain how Cointerra can make physical claims about their chip's GH/W/sec before they have put it into physical simulation. Maybe you should take on that challenge instead of going for cheap shots at me?
I think the point youre missing is that tape-out is the end of the physical design and simulation process but that that you get substantially correct data way before you tape-out. Once physical design has been finished and layout has been done, you can do power and timings simulations and get pretty accurate data. Every single other asic company in the bitcoin world has done the same thing. You dont need to wait til you tape out to get that data. You get it during the physical design process, weeks before you actually tape-out.
The typical physical design process starts when the rtl is complete and the functional simulation already works.. And takes about 3 months (though knc & hashfast have both proved you can do it in closer to 2 months if you really motivate everyone at the physical design house to work fast and sleep less.). Cointerra is also trying to do it in less than 3 months... We'll see if they make it.
Btw, im not a complete stranger to silicon design, though its been a long time since ive done it. My team designed one of the first (possibly The first) gpus, and also an early risc microprocessor that competed with arm, though we didnt alas beat them. Once standardised in a cellphone the rest was history.
And as for being late to tape out... I was aware of the tape-out goals of several of the asic companies, and whether they revealed the dates publically or not, few, if ever, tape-out on the day they expected to. Things happen, its not an exact science. Of course i had hoped ct wouldve taped out last week.. But im sure theyre busy trying to make up lost time and tape out just as soon as they can. Theyre still expecting to deliver systems in december (and seeing as knc did it (well done guys), and hashfast are expecting to do it (fingers crossed), i see no reason why cointerra cant also expect to go from tapeout to actual shipping systems in 2 months as well). Those guys all paid extra to their fabs to get the 'super rocket runs' or whatever the fab calls it when they charge shitloads extra to make things happen faster. Literally the fabs employ guys to hand carry the wafers during the mask production phase and move things along much quicker - almost twice as fast - as their regular priced mask layer production.
I just think youre plain wrong to keep writing that cointerra cant design rtl worth shit. Its insane. Ct has a chip spec that thats more GH per die mm than any of the others, and that didnt happen by accident. Maybe it took them too much extra time to optimise the circuitry so much better than the others.. Maybe that was time wasted when they couldve been shipping earlier like the others are... But instead, ct's differentiator is that they'll have a very efficient chip. And of course i wouldve liked to see it hashing weeks sooner, but i also dont only intend to hash with it for a week or two, but will have it switched on for perhaps the whole year... So i want it to be the fastest gh's, for the fewest dollars, and use the lowest KwH... And i think the extra time ct spent optimising the design will have longterm payback.
Sorry crumbs for another wall!
