Ok, so Q&A (first page of) is typed up below. I dug a bit further than the question specified in a few places.
The pic below has an amusing story. We all know English is not the first language for anyone, but Sam, and he is probably the worst speller of the lot.
So, their tag line:
'Bitcoins make the world go around', is as they realised afterwards; a spelling mistake, and should read 'Bitcoin makes the world go round'.
Alas, too late and they had ordered Mars's boards by that point, all with the spelling error, which I had Sam so proudly pose with. We were all given a pcb board, complete with spelling mistake, as a memento. Who knows? Perhaps one day it itself may have value, if not only historical...
Bare in mind any err, umm, or ahhs are due to Marcus having to speak a foreign language to his own. He wasn't hestitating.
Q&A
–
ChipGeek
Note: These questions come from a guy (me) that has worked in (non bitcoin) ASIC design companies for over 25 years so my focus is from that perspective.
My questions are:
1) Is your ASIC design a hardcopy of the existing FPGA design or is it a semi-custom standard cell design? (They've said "standard cell" before but other information makes it look like hardcopy.)Marcus: Ah no it's definitely a standard cell, we've done hardcopies before like this, yes and structured ASIC, yes and gate array ASICs, but this specific design is standard cell ASIC 28nm.
2) Have you done a (semi-) custom ASIC design of this complexity before? Have you taken wafers through all of the post-fab production steps: wafer test, bumping, packaging, package test and binning?Marcus: Thats a lot of questions in one question.
Sam: Have you done an ASIC Before...?
Marcus: Yes we have done many ASICs before.
Sam: (confidently dropping ASICs fabricated on PCBs onto the desk): That one, that one, that one.
Marcus: We have done designs that are much more complex. One with 187 clock domains, that's the...that's the most tough design. I...hope that we will never see such a design again, so yes we have done more complex designs.
Me: What Was that?
Marcus: That was a hardcopy. 40Nm hardcopy.
Me: But what was that for?
Marcus: Ah that's a confidential large customer.
Me: Ok, I thought that might be the case, alright...
Sam: How many clock domains are going to be in our chips?
Marcus: It's basically, well...it's going to be two clock domains, but...but there's a lot of different complexity changes, differences in between as well...
Another forum member interupts: Large scale binning and fabrication...
Marcus: Fabrication and all that, that is done by the fab, so they handle that, that's their business. We provide them with our jailcode, we've written our jailcode in a way that makes the back end design faster, and the fab time faster. There's multiple ways you can do that...if you know what you're doing.
3) Are you doing pre-package wafer test?Marcus: No.
4) Are you doing post-packaging testing on a real production tester (Teradyne or similar)?Marcus: No. We...and...and...I..I..I need to say a little bit about why; ah because we will have a self built in test that will automaically test...the...because the chips are so large, so that we can compensate for any losses in the Bitcoin engines. If there are any failing ones then we can compensate for that.
Me: Physically large?
Marcus: The...the die size of the...the?
Me: Yeah
Marcus: The die size will be...very large.
Me: But does that...ummm...
Marcus: That, that means that some of the parts in the ASIC might work and some will not, but we can compensate for that.
Another member: We can compensate for that.
Me: So does that mean you yield less per wafer?
Marcus: So yeah, you will always have a yield problem, and when you increase the die size, the yield problem becomes larger of course.
Me: ok
5) (This question only makes sense if they are NOT doing hardcopy.) Is your package the same as the FPGA package or a custom one? If custom, have you done custom packages before?Marcus: Thats, it's not going to be the same package as the FPGA, as this is going to be a super large package. When we are talking about ASIC packages you have custom ASIC packages, and you have standard ASIC packages, ah, right now we're leaning towards, we have two options. We can use a very very large customised package solution, but we also have an option, so we...we don't have to decide that right now, but very soon we have to pick what kind of package, we should use. But it's, it's with ah, it definitely has to do with power, and heat dissipation. That's what's critical.
Me: So your now, your now going through a different route for heat dissipation, as you're not going down the liquid cooling anymore. Which was the original...
Sam: Datacentres told us not to. Please don't do it, we don't want to host that, we really don't want to see our customers demanding water cooling in our datacentres, blah...blah...blah, so...
Marcus interjects: But it's just the mounting...it's...it's a different mounting scheme, really. We mount that (show's big ass heatsink and fan) we can also mount water cooling
Sam: That's the heatsink (points)
Marcus: So err...we can choose, but for the first productions we'd rather use this ah, this massive active heatsink and over cool it and run the chip a little bit slower to make sure we can, we, we, we don't overheat it.
Me: So, I mean is it fair to assume, then, anything that you would house in future will go through this method, but you might offer a water cooling method if they want to do it at home if they want and have...?
Sam: Basically...basically yeah, at the moment, because we see it going towards industrial centres, rather than into ah single bedrooms like graphics cards are.
Me: But it's only going to be other than the cool factor, the ambient volume and, and being able to overclock it...
Sam: And then the only problem with these things (pointing at Mars fans) is they're not very big. That makes a lot more noise than this will ever make (pointing at huge fan on monster heatsink).
ElGaboHow many miners will you able to produce a day? Not exactly, their realistic perceived capacity + - 10%.Sam: We can effectively ramp up the production to meet our demands. The numbers we are talking about making 100's effectively, maybe low thousands are umm their not even...their not even bothered by that number. The factories we are using are used to doing this kind of thing, on this scale, large production.
Me: How many daily, then...?
Sam: How many, how many men do we ask them for then...
Marcus interjects: Do we need 10 or 20 guys...? What will be our maximum. They have the space, they have the equipment, everything.
Sam: The factory is not the limiting factor, it's the parts in the door really. As long as we can always get the parts...
Marcus interjects: Yep
Sam: ...we can build the boxes
Marcus: Yep.
Me: Ok, umm...
Marcus: If there would be 50,000 boxes, then it would be a limiting factor. But if it's 1000-2500 boxes, or even 5,000 then it's not a problem.
Me: Ok, so are we going to see...10, 20, 50 a day, what do you reckon...?
Marcus: More
Sam: More
Marcus: Many, many more.
Andreas: More
Sam: Many more.
Me: No, no I just...
Sam: In the hundreds.
Marcus: Exactly
Me: Okay, okay, no, no no...I, I just need to get these guys a figure...ok?
Marcus: Hundreds, in the hundreds.
–Ok, more tomorrow...night!
EDIT: Jailcode (where mentioned) = RTL Code. 
)
