For reference, my AMD 290x's top out at 373 H/s per gpu. 12.3, asm 1, intensity needs to be around 6. Anything else and it diminishes significantly.

http://image.prntscr.com/image/341d9b53d4614ab6af042b136a599d89.png

Good to know. Haven't been a fan of KNC's design language since the beginning.

Alrighty, that was a weird break but a lot of things happened behind the scenes. I'm now moving away from the Titans into some other areas, and will be liquidating all of the things I purchased/made. This includes my titan hardware AND the titan bridge board source files, so you guys can finally start making and fixing your own controllers.

I'm liquidating 15 Titan cubes all of them working with less than 30% of them having some non-working dies for 4 BTC total.

Individually if you want just 1 cube (will be random from the bunch), it'll be 0.25 BTC each. If you want to target a specific cube it'll be 0.50 BTC.

The Titan bridge board files from the working batch will be 5 BTC to whoever wants to be the person to deal with that from now on.

I'm just tired of all of this and if I'm going to move on, than there's no better people to keep it going than you guys. PM me if interested. Aside from the bridge problem, it looks like all other issues can be fixed to have a fully working set (minus the unfixable dies which I absolutely cannot help with at this time).

I agree. The quicker the low market sweeps up and empties the trash holders that will settle for less, the quicker this coin can build upon itself.

It feels like you guys are being hostile when i'm not. There's only so much I can do when it comes to crowdfunded products. You are providing a service to fix the boards for a price. That is not the same and comparable to what I did in making them. 0.2 BTC would literally cost a customer more than the price of the board itself and all the parts put together by anyone at the current rate of BTC to fiat right now. My point is, for the big bulk buyers that got it for 0.15 and the smaller buyers that got it for 0.20 paid way earlier in the year with the price being of lower value.

So unless you're saying that your fix is of higher value/comparable to someone designing a PCB, buying all the parts, and putting it together as a product, it just seems like you're taking advantage of people's need for the fix. I'd like to see you try and design it and source all the parts and have it manufactured (while controlling the fab) and still only charge 0.2 btc for your "tools and time". You know full well I can't just release a company trade secret yet you add the "/unwilling" end as if i'm purposefully trying to hide everything. If that were the case, I wouldn't even have made batch 1's and none of the big miners would be running their titans. I take offense.

Searing, I know you're venting, but you clearly don't understand the circumstances. A refund makes sense and is feasible when you make a product, have it mass produced, and sell it. Refunding can be done on a singular basis and can be taken out of the owners funds. A crowdfunded product was only possible using the funds from everyone combined to get the product to even exist. Refunds aren't feasible as there's no mass production, there's no development teams to deal with it easily, and there certainly isn't the capital to handle it. Again, you're treating me like i'm a self-contained company. I'm just a guy that's trying to help KNC customers. It worked the first time, failed the second time. So I learn from it. If I had the capital to offer refunds, then a crowdfunding thing wouldn't even be necessary.

I'm literally making titan mining still doable as it's rare that any big buying customers with more than 1 set, have their original bridges without burning them.

On my third batch I did say, if I get a bulk enough order than i'll be shipping the rest to batch 2 buyers for free and i'm sticking to my word. I'm waiting for a couple more people to get on board before I start however.

I'm also working on an extremely cheap closed loop liquid cooling solution that seems to be working nicely. Looks like I can reproduce stable 65-70 C temps on the VRMs reliably. I'll post some pics soon.

Fixing the batch 2's for 0.2 btc would be worth it if it were batch one. I sold batch 2 for 0.15 BTC. You might want to price match the fix so batch 2 customers don't feel like they are paying for a whole new board.

I used the second latest Crimson version. 16.6. With a custom modded BIOS and straps, I get about 176 - 178 Sol/s for a single 290x in a 3x rig.

Your 290x's should be getting minimum 175 Sol/s.

I'm considering your offer, though 10 BTC is a little high (after conversion rates in my shizzy country). When I can afford it, i'd totally take you up on that if you don't have a buyer yet. I think there's only a small handful of Titan owners that can even make use of it. Maybe an auction would be more relevant?

No I meant the reference to getting them working without modification. I think if the current design is dealt with, I can move on to getting rid of the need for it in a controller altogether. Can't even modify the size of the PCB without the main board getting in the damn way.

**I'll post some high quality images of my working batch 1's and batch 2's for comparison.

There's a big difference. With -i 0 my 290x can hit 135 Sol/s. With -i 2 it hit's 148 Sol/s. Honestly didn't expect that. Though since I have 3 modded 290x's in a single rig, the fluctations between them all are too wide a margin.

Looks like the best way is to have different -i values for all gpus even if they are exactly the same.

ouu that sounds nice. Let us know. Though keep in mind, the main board does not source up the chain unfortunately.

Yeah the batch 2's are the ones with issues that I just had issues. I think the issue you're experience might be an isolated situation. But you're welcome to try on some more (as I have a couple more laying around).


In my opinion, I think all of KnC's specs should be public after all this, and i'm trying my best here. I'm not a part of KnC and can't even grant NDA's like that. So lightfoot would be out of luck if he were even into getting it. but that's more hoops to jump through than it's worth. I'm working on some things to make it easier for everyone on the software side of Titan customers. I wish I could do something about the neptune customers, they are in the worst position of all.

They aren't upside down funny enough. They are identical to my batch 1. My batch ones worked even on faulty Pi's. The batch 2's are the exact same designs with no change except textual (which is far away from any danger).                                                                 

I was thinking of a way to short the controller board to prevent it from providing power to the Pi, and just supplying external power to the Pi. This would require two power sources but would be way safer on the bridge. Waiting on a new shipment of Pi's to see if I can test out those issues and come up with a surefire way of fixing this without having to add anything to any of the boards.

Well things work a little differently with intellectual property that's assigned to an entity or person. Simply liquidating their assets doesn't mean their patents or trademarks or company secrets get laid out. If that were the case, the Titan wafer specs and board pcb layouts would have already been available for free on the net.

Anyway, I never said anything was screwed up on your end. I already admitted that the second batch was botched because of a retracing/outsourcing problem with that manufacturer (which is why i'm switching to a different local fab). What I mean by working Pi, is a Pi that has no noticeable voltage fluctuations.

I've noticed the most severe Pi's that cause the bridges to burn, is the ones that ALSO have an "iffy" RJ45 jack (ethernet port). I've tested almost 100 Pi 1.1 and 1.2's and the evidence is pretty obvious. Sometimes the port just fails, and then miraculously "fix itself". This is all due to the poor optimizations of power within the Pi. This isn't to excuse KnC's design of their boards though. Their boards are poorly designed.

I'm actually under an NDA due to my manipulation of the cyclone, which is why I haven't chimed in to tell anyone any of the parts. However, I assure you as soon as the sale goes over and KnC is cleared up (legally), I can divulge anything you want.

As for why the Pi is fucking with the boards themselves is just poor choice of product. If they stuck with a beaglebone this stupid problem wouldn't have been an issue in the first place. The old Pi's have a voltage flutter that the bridges absolutely cannot tolerate. This is very much a Pi problem and is present in all Pi's of version 1.1 and 1.2 when dealing with their I/O hardware connections.

I can tell you there's some double tracing going on with the board and it tends to happen when you design a PCB like that (which is why my Batch 2's were messed up after first test use). I'll be back and forth with some people, but as soon as I can I will post the exact part.

My local manufacturer seemed to have screwed up the second batch with part sourcing which actually doesn't happen often. Unfortunately, I finally realized there is nothing I can do to the boards to prevent the Pi's from smelting them like a lightning bolt. However, for working Pi's i'll be making more bridges.

If anyone wants a bridge for a WORKING Pi, PM me. Everyone should know, that my first batch worked fine, but my larger second batch did not. So i'm going to be doing smaller batches from now on so they can be looked at with more detail before shipping. I'm also limiting customers to only those that have proof they have a working Pi, or those that accept that refunds aren't possible.

Again, I realize everyone that ordered batch 2 became royally screwed, and i'll be mailing out replacements from the third batch. If enough people want to buy the bridges to warrant a third batch.

PS. I'm also working on a way to get a PC to push the titans instead of the controller (as i've reverse engineered some of the stuff on the board and figured some things out using a cyclone IV exploit, more on that later). I just want to get rid of the Pi's and the bridge altogether instead of having to rely on them. Maybe a simple rewire board that anyone can make or that can be provided super cheap and a main controller pc/laptop.

See now that's a normal claim. I'll be happy to upload a detailed video of me plugging them in and showing them working. Stay tuned for that.

I'll have a chat and see what's up.