Show Posts
|
Pages: [1] 2 3 4 »
|
Ya you guys deserve a heads up and some insight into how things work in China.
A lot of the price changes you see are driven by the trading market in Shenzhen.
Let's take an example: We list an old model ( say 850) at $250 on the website ( just an example ). A customer will come in and order 4000 and pay immediately. The next minute they are all over alibaba, wechat, etc, being adverstised for $275 in Shenzhen. Traders basically. there is no stopping them. What it means basically is that we got it "wrong" when we thought the market price was $250.
We try to calculate a market price based on BTC prices and difficulty, but the market has different expectations for the future. the market may think MOON! and snap up all the product, reselling to make a quick buck. It's rather vicious and defies everything you might think from a western perspective. Of course things go the other way, when traders think MOON and the opposite happens. think 2018.
So we try to pick a price where the stuff wont sell too slowly and not too quickly. We also have to decide how much to pre sell and how much to save for future spot market selling.
Lets say I have X machines projected to be built in sept. do I want to pre sell all of them now? or do I want to sell some now at a discount and save some portion in case BTC continues to rocket to the moon?
The best way to look at this is that the miner manufacturers are the longest betters in bitcoin. When we bet the cash to build a machine that bet wont be sellable for 90 days ( 60- 90 depends) if BTC moons, then its all good; if it crashes there is no way we can exit the position as a trader could with a stop loss. The cash to build a machine gets laid out in advance. we are commited deep and hard.
As BTC continues up expect the same kind of craziness we saw in the end of 2017 when supplies were really tight.
This is just bitcoin. When it moons the hardware manufactures have no way to respond rapidly; we will always lag the moon because it takes time to make machines. The shortage of machines then works to make it moon harder and if difficulty does not keep up to moderate profitability then the feedback for even more demand is set in place. and it moons harder.! If the hardware manufacturers overbuild (2018), then eventually difficulty will kick in to moderate profits and then reduce profits... and then the hardware costs come down... just in time for the slow movers to jump on the train and make it even more difficult.. or it comes down so hard that miners exit..
Long ago I used to work in modelling.
I tried to model this.
still trying.
|
|
|
Bottom line. It is selling out pretty quickly so that now most large customers are pre purchasing Sept/Oct.
capacity at the fabs ( TSMC and Samsung) is pretty tight which means NO ONE will be able to build a huge wack of machines ( unlike Q1 and Q2 of 2018 )
With Upstream supply tight and BTC mooning even old shitty stuff is flying out the door. We just started selling 850! which is basically 851 that didnt met the spec. Moving like crazy.
it looks to me like we will have a repeat of end of 2017 when demand was 10X the supply. If BTC holds or goes higher then you can expect shortages.
life in bitcoin, never dull.
I will try to reserve some quantity for smaller customers and put some aside for Nick at blokforge. he stuck with bitcoin and us through thick and thin.
|
|
|
I'm also waiting for a full review, complete with: unboxing, web ui settings, power consumption, noise levels, tweaks for maximum efficiency and its respective parts disassembly ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) This is Cannan's first asicboost chip eh? My first immediate negatives would be: price, and the 30°C max ambient temperature. Its a good positive that they would (finally) allow simple LAN cables to be used, greatly simplifying setup (but is this still controlled from elsewhere as usual?) haha no not our first. It was in previous versions, just not exposed. but now that the patent issues are resolved... machine operates up to 40C, been testing from -20 to 40C. Controller is in the miner, on chip. Our AI chip runs cgminer
|
|
|
hi guys I am back. we have completed our enviromental testing up to 40C, for all the modes of the miner and yes they can take the heat. More samples should be going out shortly , we had some software tweaks and wanted to test the first few thousand machines in all the modes, especially turbo. We will open up the ordering. Demand is running pretty high, and manufacturers are light on inventory so first come first serve https://canaan.io/shop/
|
|
|
Ok, we only had an hour or two in the lab and the customer wanted to point the machines at their pool after we did our initial tests. We were in such a rush across beijing in traffic to get to the lab that I totally forgot about alerting you.
We were also surprised by getting 38TH as we predicted 37 based on the voltage setting of the PSU. There is even more headroom in frequency, ( say at least 33%) but we may hit cooling limits and we hit the limits of the power strip when we pushed it beyond 2500.
I need to do some more structured tests and longer times at temperature, Needs some prep time and we will give you a heads up when we head to the lab again.
|
|
|
It was just a link to a youtube vid of the test at the enviromental lab. I emailed it to Matti. its on my youtube channel.
Finalizing pricing now. It's rather difficult because most of the pricing is done on a RMB per TH basis in china. Irrespective of efficiency.
In short, the domestic market doesnt care about efficiency that much.
Lets compare:
38 TH @ 2500 watts .065J/GH @ 3 cent power = $5.79 per day 2 years = 4226 31 TH @ 1800 watts .058J/GH @ 3 cent power = $4.89 per day. 2 years = 3569
Same machine: same cost. and everyone sells on RMB per TH.. well which performance do you pick to price it?
Now compare M10
33 TH @ 1850 Watts .056 J/GH @ 3 cent power = $5.26 Per day 2 years = 3839
Now compare
33 TH @ 1850 Watts .056 J/GH @ 6 cent power = $3.93 Per day 2 years = 2868 38 TH @ 2500 watts .065J/GH @ 6 cent power = $3.99 per day 2 years = 2912
Of course guys here know how to do this math. The math on having a reliable system is harder. The math on how many machines actually deliver the promised spec ( or higher) is also harder to estimate. The math on reducing your number of network connections because of daisy changing is probably easier.
So, its trivally true that the lower your power price the more likely you are to trade efficiecny for hash rate. The vast major of huge customers are below 6 cents, all in. So they have to look at all the various modes the machine can run in. After seeing what Phillip was doing, I kinda pressed on engineering to make on the fly frequency changes a lot easier. There is STILL a long way to go on what I would call proper specifications for systems: in particular I would like to see independent lab testing of machines and like to finally publish some hard data on how performance changes as a function of temperature. Who knows with a built in neural net asic maybe somebody will discover that the machine can learn to control itself. The clock cores already adaptively adjust their clock rate as a function of DH.
Any way, a few other updates. machines will probably be built in a mix of with PSU and W/O PSU. You can run the machine with an integrated PSU ( it has voltage control over I2C) or just attach your own PSU. And if you get it with the integrated PSU you can always switch it.
Price on the website is just a stand in price. It will be below that once we start taking orders. While we have a bunch of machines in test, Lily and I want to get test results from the first Trial run. You should see those system hit the pool any day now.
|
|
|
Thanks for sharing these details with us Steve!
I will email you some other stuff that the mod deleted.
|
|
|
Have we worked out that these are standalone devices? i.e. if you just buy one A10 then you don't need anything else? Just http to the management interface on the miner itself?
Does that mean that if you have the choice of running them individually, each unit plugged into a switch ala most other miners out there, or you can just have one plugged into the network and all the others daisy chained off that one miner via rj45 cables?
A miner consists of 2 hash boards and a MMU. the PMU is gone. The MMU ( miner management unit ) has two versions, and this will quickly transition to 1 version. The first version uses the old style rasb pi to AUC. My best estimate is that only a few machines will be built in that style. In parallel we are finalizing test on a new style MMU that has RJ45 instead of the old 5 pin connector. with this version, the PI will be gone, the AUC will be gone and machines will daisy chain over ethernet. no real limit on how many you can chain.. This MMU is up and working, final gerber released and so mass pro will switch over ASAP. If you get a AUC version, we will probably have a replacement MMU ( with RJ45) to upgrade. I ordered the parts to do this.
I know they might just want to sound "advanced" and all, but that description is kinda scary. AI System On Chip? What else does that thing do besides handling the Asics?
The AI chip is here https://kendryte.com/https://kendryte.com/downloads/it's not scary.
... and I checked one mining on the pool, it was getting 31.89TH/s for an hour of shares. (advertised rate is 31+TH/s)
I was running at 38TH earlier today at 40C in the enviromental test lab. fun times.
Hmmmm, that would be interesting to see them go back. Maybe they noticed MicroBT's results. I think it's more likely whoever made the product information page slipped up a little. In the past it's generally a copy past of older hardware then they edit it with the new specs. We'll just have to wait on Steven to come back and enlighten us a bit.
Biffa-
I think there will still be need for the controller, not positive. For the networking I believe that you will be using the rj45 cables in place of the AUC cables. So only 1 device whether that be the PI/Controller is connected to the router/switch, then the rest are connected in the same fashion as previous models just a different cable.
I could be wrong but that's my understanding of the info given. I'm interested to hear more from Steven on this, it's all pretty cool. Especially the option to have a built in PSU or not.
Nope its 16nm because the wafer cost per TH on yeilded die is lower on 16nm than 7nm. No controller required in the RJ45 version, cgminer runs on the AI chip, dual core risc V. I think the option with the non integrated PSU will probably rule the day, especially when folks figure out that if you have the right PSU, you can run it above 31TH.. (31 TH is running about 1/2 the max clock) you'll take an effciiency hit that more than pays for itself at lower power prices. The purchasing decision and operating decisions become a bit more complex when you look at running the machine in max effciency mode ( say 24TH ) versus running it in max hash rate mode ( 38+ today) And how do I just put one number on the spec sheet? and then you have to factor in how efficiency changes with ambient temp. Its almost like an AI problem. just kidding. But I do foresee customers who will now have fine grain control of the frequency that is tied to the actual operating conditions of their farm.
|
|
|
Well Lily and I are back in charge of setting Specs and firmware requirements! Trying to get super finegrain control of the power comsumption of machines so that freq can be set and reset on the fly. In testing on the 841/851 should migrate easily. On the 10 series there will be TWO options: 1. Option with "intgrated" PSU. basically this means a PSu with I2C control. There will be no PMU. Upside, miner controls the PSU and you dont have to worry about your PSU burning up the PMU ( if the PSU supplies too low of a voltage for ex) 2. Option for regular PSU. The hash rate will take a little hit. Also the AUC is gone. Machines daisy chain via internet connection. more details follow and now for some really exciting news.... Just open sourced the 851 http://ehash.org/
|
|
|
Thanks Steve makes sense Because in a 95 f room with 39/92 set to stock clock
That 92 is still well under 105 shut down guard.
I am happy that slight downclocks are working so well NJ temps got to 90f today so the down clock worked well .
I know temps will moderate in fall and I will have a solid 7 months of above spec numbers.
So a few months of just short of spec are not a big deal at all.
I've sent you a mail with an introduction to the local guru. The voltage setting are just one way to change the clock rates. if you have the cooling and know what you are doing 14-16+ TH is achievable. Basically setting zero gives you a clock rate of around 750, but the max is 1G, so theoretical max is 16.25. Do you have the instructions on bypassing the standard voltage settings? Finally, I should probably invite a luck guy to come to beijing to write a overclocking/underclocking guide. All expenses paid + free miner. If anyone on the forum is interested write me at steve @ canaan.io
|
|
|
If the numbers at 30C as printed are only for the benefit of showcasing the optimal performance temperature ranges, then I would have no problem with the Miner.
That's the intent. the miner will operate up to 40C, but above 30C the performance (efficiency) decreases for SOME machines. So, we would have speced, 40C and then increased the +- on the power consumption, There is no issue about the warranty. you wont void the warranty. Moderator's note: This post was edited by frodocooper to correct erroneous quote formatting.
|
|
|
Until the firmware is improved, I feel that it is kind of false advertising to say that the T1 is the most power efficient machine available.
At the moment there are three different machines that can achieve about the same power efficiency level (~0.10 J per GH/s) --> Antminer S9, AvalonMiner 841 and DragonMint T1.
Ya, One of the things I am working toward is a manufacturing test report that shows the ACTUAL distribution of machines power efficiency during our burn in test. For the 841 75% of the machines are over 13.5TH, but spec wise we call that "13". specksmanship kinda sucks so what do I call the 841 Average 13.5? or Minimum 13? and how do you get other suppliers to do the right thing? Any way, I hope to have the same kind of real measurement actual distribution of power efficiency performace at final test. haha. we should do a blockchain for this with machines writing their ID and peformace into a blockchain
|
|
|
Verdict is in. Samsung. Explains the yeild issue and the variance in power efficiency.
Moderator's note: This post was edited by frodocooper to remove an unnecessary quote.
|
|
|
WTF? it's an S9 for fck sake...exact clone..with overt ASIC boost.... Big fck deal..rip off
It looks like an S9 but remember that you can find all the design files for an S9 in shenzhen if you know where to look. And you dont even need the design files to reverse engineer it. you just need cash and connections
|
|
|
It's simple enough.
10nm
which fab?
how many masks for that 10nm?
Answers to those two will explain a lot
I am really interested in hearing the answer to this question. 10nm with the ASICBoost? I don't see the benefit. If you had to guess, which fab did they use or do you think the whole 10nm is fabrication? 10nm. The first thing is most customers do not know the right questions to ask a mining company to ferret out if they are bullshit or not. So, I encourage everyone to ask the right questions about chips, fabrication, etc. If the guy doesnt know what he is talking about it will be clear right away. Speculations: In the beginning a bunch of folks were saying it was a 16nm chip. And some speculated that it was just and overclocked S9. Hmm, dunno. The problem with it being a 16nm chip from a new company is that TSMC would not give a new customer any wafer capicity. How about TSMC 10nm? Well if they answer TSMC 10nm, then that fuels another speculation that buzzed around on twitter that Halong is just a Bitmain subsidary and it's Bitmain's method for selling chips. It also helps them clean up their act and diffuse concerns about the concentration in mining. Information vacuums are never good as they fuel speculation. Maybe TSMC 12? performance is somewhat consistent with that. Intel? Nope. Samsung 10nm. From a performance and shipping standpoint samsung 10nm makes sense because there is an apparent yeild issue ( there always is ) and because there is what looks to be a voltage leak problem which could lead to excessive power consumption. You'll get chips that run at sub .09J/GH and devices that run above this. If you can just run the thing through a variety of frequencies and see what happens to the power consumption that will be a good clue. Maybe they can tweak the the process to reduce the portion of chips that might have this issue. Ideally, they would do different SKUs, those that run between 15 and 16 and those that run faster. Hard to do unless you have big volume. In any case if it is samsung 10nm, then the best thing to do is just to address the performace variations and explain what you are doing to reduce/improve the situation. If they are tweaking the process to run at a fast corner (for example), just tell folks. We are tweaking the process. If you build chips it is what you do. In the end someone somewhere will decap chip or take it to their dentist for an xray and post the dimensions. I should do a bounty on that. If you know the dimensions of the bare die that will tell alot. Has anyone decaped a Bitmain chip? Finally, the other thing is some guys use chip specs to derive machine specs. That means you might spec 16TH .07 from the single chip performance ( averages of lots) but when you actually build a machine its 15-16TH and .1 at the wall. what we do is a little different. We spec machines after they are built and tested. So 13TH, actually means a tested MINIMUM 13, and empirically 75% of the machines run at 13.5 or higher. Ideally there would be some kind of standard benchmarking but I dont think the industry is ready for that. Anyway, more information from everyone, open information from everyone is always a good thing. Moderator's note: This post was edited by frodocooper to trim the quote from NODEhaven.
|
|
|
It's hashing! All I did this morning was to change the network to DCHP from static. I guess I missed this because my other Avalons (5-741s and 821) must have been already defaulted to DCHP as I never had to change this before ![Tongue](https://bitcointalk.org/Smileys/default/tongue.gif) I feel like a fool - oh well, at least it's working! >>>> One other quick question - if I run another 841 on a 110v line, will the reduced voltage hurt the machine at all besides reducing the TH/s? THANKS for putting up with me - much appreciated. Congrats !
|
|
|
[...]
Competing in this space is incredibly difficult for many reasons. Firstly it's incredibly costly to tape out your own ASICs, especially for new process node (we're using 10nm), the NRE is already ludicrous - not to mention all the cascade of expenses that follow, and yield loss at every stage of the process. Then you have to deal with highly questionable, anti competitive practices perpetrated by monopolies - like buying out all the components needed to make the miner or making arrangements that they be unavailable to competitors; and of course price wars where they can afford to sell at cost or below cost for long enough to kill competition. It would seem many of the posters in this thread still think of mining back in the days when you only needed a million or two at most from start to finish and there wasn't any strong monopoly.
[...]
It's simple enough. 10nm which fab? how many masks for that 10nm? Answers to those two will explain a lot Moderator's note: This post was edited by frodocooper to tidy up quote formatting.
|
|
|
I hope the Canaan folks are still reading this thread because I want to reiterate what the others have said. This genuine concern for the customer is almost non-existent in *any* industry, not just miners. This cements my decision to buy Avalons the next time I am ready to purchase!
I'm old school, bro. Customer comes first. I wish I could do more. At some point once I have enough guys from this forum trained. I will set them lose to train others. You guys have already done that for me, so thank you for that! My last order of 1000 Avalon 741s came with 20 spares, of which we have only had to use 13 in the last 9 months.
Canaan has no equal in this industry as far as the reliability of the units go. The issue for me has always been efficiency and price. While I dont mind paying for quality, its hard to take a hit on the efficiency side when you are talking thousands of units like I work with.
In my experience, the reliability has not made up for the power usage. It looks like with the 8 series you guys are on track to be at least competitive in that regard and I cant wait to get my test 841 and put it through the gauntlet.
Thanks. I have tested this out with a few people and no news is good news. Once the repair guide and troubleshooting guide is done. Then I will feel better about shipping free spare parts with every order.
|
|
|
It is a great idea. My repair needs were minimal, but it was nice to have the fan shroud shipped out so I could replace it myself rather than 1) leave it and see if it caused issues, or 2) ship it out for such a minimal repair.
Love seeing Canaans business and service network continue to grow.
Once I have accurate stats on failure rates and modes ( stats iwas one of my side gigs-- field failure on autoparts) I will start to include free spares on larger orders.. small sample stats is hard..LOL
|
|
|
[...]
The options would be. 1. I ship customers the spares they will need In advance. basically the in field failure rate is 1%. so I can just ship units with the order. 2. Stock parts at Distributors. 3. Stock parts at Certified technicians. ( hey I stole your idea!) The Key of course was explaining to people that its not optimal to send a whole unit back IF you design a machine to be in field repairable. Bitmain is less forthcoming about it, and at the time did not want it known publicly as they had already decided among their customer base who would benefit most from the training. When I send RMAs to bitmain, its usually at least 250-300 units and I am one of their top large scale customers so im pretty sure thats why I was targeted. They even took us out on a yacht!
I know this isnt a really good description of what im trying to say, but from the pictures posted Avalon looks alot more "first world" in their manufacturing and build process than what I saw at bitmain. Much cleaner, nicer facility, much more attention to detail.
The key is designing for reliability. If you design for reliability then the hardest problem is solved. Marketwise we ship about 25% of the hash rate last year. Every Bitmain customer that comes to me complains about the same thing: reliability. They report failure rates from 5-30%, so i dont know about the "attention to detail". They key is making a device that is so robust that you cant screw up the manufacturing. They also complained about the downtime from shipping back products. I guess the kind of untold story is that our product is designed to be easily assembled and disassambled with no specialized training. That means I can ramp production 10X with minimal investment and time. The other untold story is that its easy to repair. When matti and Sky are done with the guide, I hope folks will understand. If not I can always send matti to your farms to train you. No yacht, sorry, all business here.
|
|
|
|