Which is kind of dumb since why bother using 0402 parts if using as much space as 0603.
Are you talking about the C1005X7S1A474K? (The one with 192 pieces on the BOM.) Because that would be pretty awesome, considering stock of the 0402 version is starting to run low at most places. Could we use a 0603 version instead, then? If it's a standard 0402 cap, if one specific manufacturer part on one vendor runs low or out, there are 5-10x other vendors with the same part, and probably another 10x manufactuer parts (each listed on the vendor sites) that will work The only part i saw that was tricky to source was the PIC, i had to order it directly from the manufacturer and it had a ~2 week leadtime. |
|
|
|
Let's see how the race between TerraHash, KnC & Bitfury will end...
My vote is on bitfury, hands down. Terrahash is off avalon chips right? they should be first.. |
|
|
|
Since when is a heatsink not part of a PCB assembly design?
Sure but a heatsink is kinda useless without a working circuit to generate the heat, this thread is focused on the first step, comparatively the heatsink is easy. You seem to expect more than R&D in an R&D thread, there is nothing for sale yet here. Most assemblers are offering heatsinks, if you want to assemble yourself try to buy from one of the assemblers. Heatsink : https://bitcointalk.org/index.php?topic=208381.0It's not the design, it's the sourcing. Of course getting all the circuity going is critical. But if you find out you can't get a heatsink for 2 months, you're screwed I'll work on it this weekend, maybe even setup a group buy or something this is supposed to be a DIY project, (saying a critical component has to be bought from an assembler?) |
|
|
|
What are the options/alternatives for the k16 heatsinks?
There are many options covered in this thread, I suggest you look at some rather than having people repeat them over and over. Type the word heatsink into the search and you should get 6 pages of hits. People shouldn't need to search through multiple pages of mostly irrelevant information to find 1 comment which makes a passing mention of the info you're looking for. Any kit that is currently ready should be linked to in the initial post. There are no kits in this thread, it's a design discussion thread. Go look at one of the board buy threads, there are many, like Burnin's or Terrahash threads. Since when is a heatsink not part of a PCB assembly design? |
|
|
|
What are the options/alternatives for the k16 heatsinks?
There are many options covered in this thread, I suggest you look at some rather than having people repeat them over and over. Type the word heatsink into the search and you should get 6 pages of hits. Yeah about that, pretty much looks to be all all custom solutions with questionable sourcing availability. Time to design and quote one or find one that can be easily modified to work i guess. There doesn't appear to be anything off the shelf that works, but that's probably unavoidable |
|
|
|
Modern software tools from companies like Cadence can predict very precisely how silicon will behave once manufactured in a particular foundry's process, so as long as the design has been done properly there are NO excuses for not pinning down the specification. Out of curiosity / devil's advocate, how does shit like 1.7% yields (Nvidia's famous blunder) happen then? |
|
|
|
Quit? Let the riot begin: The Firmware is not yet production ready cgminer throws a shitton of errors on starting. AND the chip currently runs at 128Mhz because its not properly cooled, i have tested it at 256mhz too and it works but its too hot for long tests. Now the bad news: I will very likely not get the boards tomorow so we have to wait until Monday. What kind of hashrates do you see at 256mhz? |
|
|
|
|
What are the options/alternatives for the k16 heatsinks?
|
|
|
|
|
(I haven't gotten caught up ITT)
I'm getting at least 1 sample chip (maybe a second) from one of my group buys. I'll be having a local PCB assembly house build up a K16 with one of the asics on it (one chip per pcb) and then doing some burnin and Overclocking tests, as well as thermal analysis, and then publish the results here.
Not sure when the chips will be in, but i also have the microcontrollers due to arrive at the end of june so it would be after that.
I'll have to check/talk with people about the k16 PCB assembly revision, i can build up one without asics (test it with a fixed resistive load or similar) and if it looks good, build the assembly with the asic. Especially if i have only one, i'll have to make sure it's good to go.
I think BkkCoins will be getting quite a few chips from people though?
|
|
|
|
Also curious to know the plan with the chips
My understanding is there are 30 sample chips for every 10k order, so every 333.33.. chips would get 1 sample chip, if divided evenly.
[ I think we don't want to go old-testament style "the chip shall be cut in two and each person shall receive half" ]
I've purchased 572 chips from batch #1 (252 from atcsecure, 320 from dserrano5, both transactions posted publically, confirmed by original owner, and raginazn himself)
I am an EE with access to a lot of nice test equipment and plan on having my Klondike-16 boards assembled by a local professional PCB Assembly (P&P + Reflow with Pb63Sn37 solder) company I've worked with and can get boards made very fast. I'd like to get one or two of the sample chips to test and publish for everyone's knowledge the results of these tests before the full batches ship in July:
* The Klondike-16 design as-is (P&P files, Solder Paste locations -- especially the QFN pad, Solder paste stencil thickness, etc) as populated by an outsourced assembly house
* The standard functionality of the boards (with a single chip, my understanding is the design will work with any number of chips, might need to tweak the firmware). Hash rate, power draw. Power draw of ASIC at various clock and hash rates.
* Thermal analysis of the board. With various heatsink, without a heatsink, with various heatsinks with poor mechanical connection (I'll watch the temp closely so it doesn't overheat and damage the chip), with and without forced airflow.
* Overclocking. I can modify the design, change out clock sources, modify the settings on the PLL on the chip(s). Am very curious to see what these chips are capable of with proper cooling. I would not be surprised to see 20, maybe 30%+ better performance. I do not think the current 282MH/sec is a thermal limit, but I can't know for sure until i can play with it. Similar to thermal analysis, I'll keep the chip within acceptable operating range. I don't want to blow anything up until we have a lot more information. Especially with only the one chip.
* Affects of the above on the power requirements (increased current on the 1.2V line, perhaps above 2A/chip. Possibly need for additional decoupling capacitance on the chips. Oscope measurements, Spectrum analyzer (depending on if there is harmonics or noise coupling onto the Vcc plane(s)).
* After the above, some modifications to the design to support overclocking. -- While at this time I believe there is a good amount of headroom on these chips, the current design, both Avalon's official reference and Klondikes, provides exactly the max current spec for the chips (i.e. 32A @ 1.2V for 16 chips, each rated for 2A max.) I expect that OC'ing will push this beyond, something i'll confirm with power/current measurements when OC'ing. (Will report Clock freq, Hash rate, Current draw, Temperature, Heatsinking methods used, etc.)
If i happen to get more than 1 (i have some chips in other orders), I will probably have multiple PCB assemblies built up, each with 1 chip. This will help to verify a tiny bit better the yields of the assembly, which in theory should be very high. But there would be nothing worse than suddenly realizing that 50% of the boards are scrap. If i get say, 4, i will have 2x boards with 1 chip and 1x board with 2 chips made.
Of course we want to check to make sure that a board with say, 16x chips works of course. But i expect that BkkCoins will be getting several dozen chips from various people, so he should be able to take care of that.
Quoting my old post ITT. Would i be able to get one (or dare i say two, if people don't mind? I'd be more than willing to buy out the remainder or whatever for a few engineering prototype samples.) of the sample chips? (I've got 572 chips from batch #1, which would be ~1.71 sample chips)). I would like to be able to test these things on the K16 assembly (specifically the OC'ability and thermal analysis and limits), and of course publish and share the results. I may be able to get another sample chip through a different group buy, but with more than one i can dedicate a PCB assembly to a long-duration burn-in test (say +10/20% OC) and i can feel free to quite literally kill a chip or two with some absurd OC'ing/undervolting/etc -- since i'll only be putting 1 of 16 chips on the PCB assemblies for these tests. Definitely. I can add your 1 to the 8 and send 9. Awesome, thanks! |
|
|
|
|
Also not sure but if testnet is normally say, maybe 100 Gh/sec -- if you mine with terahash the difficullty skyrockets, when it goes away, nobody can mine blocks and it takes hours if not days to find them, to get to 2016 and readjust takes forever
some of the shittier scamcoins (altcoins) had this, hashrate jumps from 50 MH/sec to 2500 MH/sec, difficulty goes up orders of magnitude, prices drop, everyone stops mining, nobody can get blocks and the coin is fucked until enough blocks are mined that difficulty readjusts.
|
|
|
|
So 1200 units in Batch #2 and 3# combined?
65Gh/sec * 1200 = 78,000 Gh/sec = 78 Th/sec
Current difficulty is 19,339,258. Their mining was prior to this. It was as low as 10 thousand Again, ueber conservative, use this 19 thousand difficulty
78 Th/sec generates 2028 BTC per day.
If they test each unit for 24 hours, at this difficulty they would have made 2028 BTC.
We see ~780 which is something like 9.23 hours-per-unit average.
Take into account the difficulty is lower, and you have more like 7 hours per unit of testing.
take into account they probably are only testing say, half of their units, that's 14 hours per unit of testing.
Now, if there's a second pool with the same amount of coins, that's 28 hours per unit of testing.
Doesn't seem outrageous. People are forgetting how much fucking hashing power they have. 78 Terahash.
Wait to see what their response is. But for shipping 6000$ hardware, you want to run some basic tests, it's just good engineering.
Yup, they are driving up difficulty while testing, so you get less BTC when you receive the device and start mining. Testing should be done on testnet. How does the equation for difficulty adjust if they 'spike' say, 30 units at at time or 1.95 Th/sec for 1 day of a 7 day week? Does it just get calculated as an integral/average hash rate/block time over 2016 blocks or something? I thought people sometimes back off mining prior to diff adjust to artificially lower it? iirc it doesn't look like they're running these even close to full time. Also, does testnet have acceptable difficulty for this much hash power? or would it just find blocks every few seconds.. not familar there's also the chance they're the nicest people ever and donate their BTC from testing to everyone who ordered.. seems like that would be a fair thing to do if this is the case. |
|
|
|
|
So 1200 units in Batch #2 and 3# combined?
65Gh/sec * 1200 = 78,000 Gh/sec = 78 Th/sec
Current difficulty is 19,339,258. Their mining was prior to this. It was as low as 10 thousand Again, ueber conservative, use this 19 thousand difficulty
78 Th/sec generates 2028 BTC per day.
If they test each unit for 24 hours, at this difficulty they would have made 2028 BTC.
We see ~780 which is something like 9.23 hours-per-unit average.
Take into account the difficulty is lower, and you have more like 7 hours per unit of testing.
take into account they probably are only testing say, half of their units, that's 14 hours per unit of testing.
Now, if there's a second pool with the same amount of coins, that's 28 hours per unit of testing.
Doesn't seem outrageous. People are forgetting how much fucking hashing power they have. 78 Terahash.
Wait to see what their response is. But for shipping 6000$ hardware, you want to run some basic tests, it's just good engineering.
|
|
|
|
Also curious to know the plan with the chips
My understanding is there are 30 sample chips for every 10k order, so every 333.33.. chips would get 1 sample chip, if divided evenly.
[ I think we don't want to go old-testament style "the chip shall be cut in two and each person shall receive half" ]
I've purchased 572 chips from batch #1 (252 from atcsecure, 320 from dserrano5, both transactions posted publically, confirmed by original owner, and raginazn himself)
I am an EE with access to a lot of nice test equipment and plan on having my Klondike-16 boards assembled by a local professional PCB Assembly (P&P + Reflow with Pb63Sn37 solder) company I've worked with and can get boards made very fast. I'd like to get one or two of the sample chips to test and publish for everyone's knowledge the results of these tests before the full batches ship in July:
* The Klondike-16 design as-is (P&P files, Solder Paste locations -- especially the QFN pad, Solder paste stencil thickness, etc) as populated by an outsourced assembly house
* The standard functionality of the boards (with a single chip, my understanding is the design will work with any number of chips, might need to tweak the firmware). Hash rate, power draw. Power draw of ASIC at various clock and hash rates.
* Thermal analysis of the board. With various heatsink, without a heatsink, with various heatsinks with poor mechanical connection (I'll watch the temp closely so it doesn't overheat and damage the chip), with and without forced airflow.
* Overclocking. I can modify the design, change out clock sources, modify the settings on the PLL on the chip(s). Am very curious to see what these chips are capable of with proper cooling. I would not be surprised to see 20, maybe 30%+ better performance. I do not think the current 282MH/sec is a thermal limit, but I can't know for sure until i can play with it. Similar to thermal analysis, I'll keep the chip within acceptable operating range. I don't want to blow anything up until we have a lot more information. Especially with only the one chip.
* Affects of the above on the power requirements (increased current on the 1.2V line, perhaps above 2A/chip. Possibly need for additional decoupling capacitance on the chips. Oscope measurements, Spectrum analyzer (depending on if there is harmonics or noise coupling onto the Vcc plane(s)).
* After the above, some modifications to the design to support overclocking. -- While at this time I believe there is a good amount of headroom on these chips, the current design, both Avalon's official reference and Klondikes, provides exactly the max current spec for the chips (i.e. 32A @ 1.2V for 16 chips, each rated for 2A max.) I expect that OC'ing will push this beyond, something i'll confirm with power/current measurements when OC'ing. (Will report Clock freq, Hash rate, Current draw, Temperature, Heatsinking methods used, etc.)
If i happen to get more than 1 (i have some chips in other orders), I will probably have multiple PCB assemblies built up, each with 1 chip. This will help to verify a tiny bit better the yields of the assembly, which in theory should be very high. But there would be nothing worse than suddenly realizing that 50% of the boards are scrap. If i get say, 4, i will have 2x boards with 1 chip and 1x board with 2 chips made.
Of course we want to check to make sure that a board with say, 16x chips works of course. But i expect that BkkCoins will be getting several dozen chips from various people, so he should be able to take care of that.
Quoting my old post ITT. Would i be able to get one (or dare i say two, if people don't mind? I'd be more than willing to buy out the remainder or whatever for a few engineering prototype samples.) of the sample chips? (I've got 572 chips from batch #1, which would be ~1.71 sample chips)). I would like to be able to test these things on the K16 assembly (specifically the OC'ability and thermal analysis and limits), and of course publish and share the results. I may be able to get another sample chip through a different group buy, but with more than one i can dedicate a PCB assembly to a long-duration burn-in test (say +10/20% OC) and i can feel free to quite literally kill a chip or two with some absurd OC'ing/undervolting/etc -- since i'll only be putting 1 of 16 chips on the PCB assemblies for these tests. |
|
|
|
|
ITT : Disinfo shills trying to make money via the expense of others
Correction
ITF(orum) : Disinfo shills trying to make money via the expense of others
Correction
ITC(ommunity) : Disinfo shills trying to make money via the expense of others
|
|
|
|
|
Terahashs anonymously solomining their own full blocks mining via a Tor exit node in Germany, a shit ton of the exit nodes are there
(Solo block bitcoins on its own unique unused wallet are worth more than non-solo block coins because they are much easier to make anonymous and hide. Tor + solo mined block is theoretically untraceable to anyone)
You'd have to be an idiot to associate significant hashing power with any IP that can be tracked to you without a massive amount of work, for many reasons.
|
|
|
|
|
Avalon is a monopoly like the GPL is a monopoly
|
|
|
|
Design type: 100% Hand routed for performance density
Bullshit. They should change the supplier. They must have found the shittiest design and verification engineers on the planet to get such low performance from a full custom 65nm asic. Further, for a chip with such little logic on it their yields are crap. My guess would be they used shitty engineers and global foundaries' standard cell library, the engineers didn't setup their simulations correctly and as such the cores fail under even the slightest real world heat. This guess is additionally backed up by the high hardware error to accepted rates posted in screenshots by various Jalapeno recipients. I would guess that the estimated hash rate on the pool side is much lower than the promised 5 gh/s. Many people have speculated (Avalon themselves as well) that it's not actually a 65nm chip |
|
|
|
|
Interesting.
Still reading through but I'm very interested in getting some chips
I'm an Electrical Engineer in the USA and have access to spectrum analyzers, oscilloscopes, power supplies, clock generators, FLIR camera/fiber optic thermometers, etc that would be needed to properly test these chips and quantify their performance in an objective way. I can design PCBs and then mill or order them from a board house as well
I haven't seen anything about the hardware, if there's a reference PCB design or we should create something, but I'd be able to do that myself. Soldering I could do by hand or have them pick & place + reflowed to build the PCB Assemblies.
You say it's QFN, do you have a thermal pad on the bottom similar to Avalon's design or is your heatsinking provided with an external heatsink on the top?
In the spirit of disclosure - If you look through my posts I've spent my time working on the Avalon chips and have purchased a number of them and am doing similar work on the PCB assemblies for those. I was aware of the bitfury project but being unable to read russian (which it seems most of the material is in) and the difficulty to get the chips outside of Russian i was wary. I would put together quantitative, objective measurements of these chips to allay fears and concerns of the community and myself
And reading through your posts in this thread I can tell you really know your stuff. Very much the opposite of BFL. The power supply decoupling (and buck converter supplies themselves) and associated PCB routing for the chips is critical. Deadbug would be fastest but the parasitic ESL may very well limit performance.
Ok, i see you say they're designed to dissipate heat into the PCB (e.g. thermal pad like Avalon, i'm assuming you have a lot of insulator so putting a heatsink on top is pointless, like avalon design). Have you tested these chips deadbug'd? sounds like it may require some careful heatsink mounting to the pad to run at full power dissipation
|
|
|
|
funny whois zerohedge.us Billing Contact ID: EDD09205F595F517 Billing Contact Name: news manso Billing Contact Organization: sdgsdgsdg Billing Contact Address1: dfhfdh Billing Contact Address2: fdhhdfh Billing Contact City: new york Billing Contact State/Province: NY Billing Contact Postal Code: 10001 Billing Contact Country: United States Billing Contact Country Code: US Billing Contact Phone Number: +1.987654321 Billing Contact Email: blackapples@yahoo.comthe root page is just an index of 'blackapples' is also in the .jar exploit code however v-panel.info seems to be down |
|
|
|
|
Show Posts
