Got my two s2's today. Both where fucked up inside, Cards everywhere, bent pci slots. Plastic thingy holding the cards had about 50% survival rate.
After assembling them back together one did not start and one worked at 650Ghs. After some hours of trying different things I got two new PSU's, XFX Black Edition 1250W and RECOM 1600W(Overkill but had it laying around) One are stable at 1Ths the other is working at 920ghs with row 10 showing only lines.
I feel the s2 is a step down from s1 when it comes to build quality and overall design.
Next time pack each cards individually with bubblewrap and stick them inside the miner. Shipping the miner with the cards already in their slot is borderline retarded.
mine arrived fine, and had rubber strips in the lid designed to keep the cards tight in the slots. The damaged units must be going through some rough handling - maybe a 'this side up' is necessary? |
|
|
|
What's the actual process of applying a coupon to an order? I have four available to me, but I don't ever remember seeing anywhere to enter coupon details at check out.
When you get to checkout you are offered a coupon there - be aware to change the quantity of coupons to match the number of units you are buying - the default is only to apply one coupon. So if you buy three then change the number of applied coupons to three etc... ^This. 1 coupon per antminer. I just got 16 of the 0.05BTC ones and am quite pleasantly surprised  just pulled the trigger on 3 more units, almost entirely because the extra ~8% discount puts the unit at a nice price point. At current BTC prices, the unit looks poised to breakeven but bring in ~0.9BTC or more in its lifetime so any increase in USD/BTC would offset the power costs nicely |
|
|
|
|
in the last ~6hrs my mining speed at eligius for some units has been a little sporadic. even having 3 stable antminer S1s ont he same power supply, only one of them shows 60-80GH for a data point and then bounces back. seems like when these data points happen about 1/3-1/2 of my machines will show slightly low hashrate at the pool then jump up again to normal.
is this a pool issue? (high difficulty to the miner, stats page issues, etc)? Its not enough to severely impact my 3hr hashrate, but my 22.5min hashrate shows about 20% lower whenever it occurs. Its not the first time, these little chunks of instability or low hashing on select units seems to occur during small 1-4hr periods of time and then absolutely problem-free for hours or days after
|
|
|
|
going from 55nm to 28nm would improve efficiency by about 4x at chip level (perhaps more if they revise the design at all). The overall efficiency might increase by 3.5x to account for the power use of everything else in the BOM and the control modules
and you have to get over the idea that the S2 is 'too big'. The S1 cost 4.75BTC at first, now its about 15% of that price. The S2 is due for a pricedrop to a more realistic ~5BTC very soon if it want to stay relevant. This will drop to 3.5BTC after a month.
The biggest advantage in the S2 is some clear design improvements. these include: space for ATX psu, rack-mounting, denser chip arrangement, and 140mm fans (these are VERY good fans, more airflow than from the S1 fan but less RPM and noise)
If they come out with a new chip, I expect they will maintain the S2 profile, but pack in 4-5TH of chips.
re OP on "too big". S2 is overpriced for its power and hashing at ~$3.2/GH with S1 currently at $1.4/GH. S2 has problems of boards falling off the rails during transport (or was it fixed already?). I don't want an expensive unit where reliability might be a problem (with flash cards on restarts, etc.). Modularity is very important, I agree with OP. the price/GH is definitely a bit off in the S2. (keep in mind though that it uses 10x the chips as an S1, which is *almost* logical if the chips are the biggest cost). IMO it should be 5BTC right now at most, if not 4.5BTC. my unit is a batch 2 and i had no issues in shipment or in usage. (~980-990GH stock was a smidge low of the 1TH). the rails that hold the modules are fine on mine and do the job well enough that i really wonder what went wrong with the damaged units some people received. they are small plastic slides on a metal brace, all of which seems firmly put together in my box. The PSU was new 1000W enermax GOLD, or looked like new. I have not had issues with the SD card on reboots or power cycles, but my bitfury unit used to so i am not surprised that the some of the S2 sd cards have corrupted when using cheap cards. The S2 is rack-mountable. the hashing boards are removable (if swapping them is ever needed). I would say that is important compared to the havoc of dealing with over a dozen S1 units and their seperate power supplies and difficulty stacking. 1!1!1!1 would testify to this - all you need to do is see the difference in his racks for S1 units and S2 units. one rack is clean and gorgeous, the other is a hodgepodge of zipties, server PSUs, and power/ethernet cables. |
|
|
|
sorry now i did my research - they are real but IMO not a very good method of cicuit overload protection when the key safety component can be adjusted to higher/lower sensitivity. What happening is that under load the voltage was going down to 115 which caused the overload. Our electrician came over on Sunday morning and adjusted the voltage up to 130 which under load went down to 122 and solved the problem. makes sense. however, why are you not using 208V power? almost every 12V power supply is 3-6% more efficient when drawing 200+ volts, and you can virtually double the amount of equipment without hitting the same amperage off the circuit panel |
|
|
|
how to understand the column HW in web of antminer S2?
it is too high
HW Error Calculation Formula: HW / (diffA + diffR + HW) * 100 it is in % of accepted? so with 212.5 Mhz i have 0.001% or 0.1% or...? Yes thats a % calculation should be less than 1% normally at 180Mhz If you run higher than 180Mhz then you will get higher HW errors and have just invalidated your warranty. The stock speed is 196Mhz for ~990GH. I am running at 218.75MHz for the past 49hours: 1085GH at pool, 1115GH at webUI, 2.02% hw errors (probably drawing ~1250w from the PSU) temps range from 48C-55C, fans range from 2400-2520 |
|
|
|
Does anybody think that S2 is an intermediate product before Bitmain releases S3=S1 form factor with a different chip (40nm or 28nm)?
Can S1 form factor even work with 40nm or 28nm chip? I imagine that it could, perhaps with two fans (push-pull)?
I would be thrilled to have the S1 equivalent form factor, hashing at 400-500GH/s.
It would be a huge blockbuster if it will be as reliable as S1 has been (at least in most cases).
The thing is, a simple die shrink wouldn't help efficiency THAT much I don't think, they would nee to do a bit of redesign as well in order to get better efficiency, though a die shrink would help. They would need to go from 55nm to 20nm to stay competitive or give miners a huge incentive to buy within the next 6 months. That said, I firmly believe that the S1 form factor is extremely important if BTC is going to succeed. The problem with having all these GIANT units is that they will only be held by small amounts of folks. I can convince friends and family and other people interested in mining to pick up an Antminer S1 at these current prices very easily, so in that sense, its awesome. However, the problem with these more efficient big boys is that they cost thousands of dollars. If Bitmaintech could release a 20nm Antminer S10, or something like that, that was basically a 500 GH.S 20/28nm version with some efficiency and UI upgrades, and sell it for say, 2 btc, I think a lot of people would hop on board, and they could churn these out for at least 6-8 months. going from 55nm to 28nm would improve efficiency by about 4x at chip level (perhaps more if they revise the design at all). The overall efficiency might increase by 3.5x to account for the power use of everything else in the BOM and the control modules and you have to get over the idea that the S2 is 'too big'. The S1 cost 4.75BTC at first, now its about 15% of that price. The S2 is due for a pricedrop to a more realistic ~5BTC very soon if it want to stay relevant. This will drop to 3.5BTC after a month. The biggest advantage in the S2 is some clear design improvements. these include: space for ATX psu, rack-mounting, denser chip arrangement, and 140mm fans (these are VERY good fans, more airflow than from the S1 fan but less RPM and noise) If they come out with a new chip, I expect they will maintain the S2 profile, but pack in 4-5TH of chips. |
|
|
|
sorry now i did my research - they are real but IMO not a very good method of cicuit overload protection when the key safety component can be adjusted to higher/lower sensitivity. |
|
|
|
We kicked the main 200 amp breaker.
Edit: We kicked one of our main 200 amp breakers also took down some of our switches. When these switches went down then some of our other racks could not communicate with the internet. We will be correcting this problem.
how do you trip a 200A breaker? that's not an accident if you know anything about the rule of thumb: 'don't load a circuit past 80% with 24/7 draw'. Looking at eligius it seems like the breaker tripped at least twice and the system is now/still offline. (eligius stats are working today - so the absence of 128/256 stats means no hashrate) also: 200A*80% load = 160A. at 110V this is 17.6kW at 220V this is 35kW. This means that the single breaker is about 30-60% of total load |
|
|
|
Im pretty sure most current 28nm vendors could hit those efficiencies if they wanted to
I'm pretty sure it's not that simple or they would all be advertising/doing it. This is exactly what bitmain did, and when the need arises, so will Cointerra, HF, KnC, Bitmine, BFL, and all the other 28nm vendors. 0.35W/GH at the chip level is nothing special.
If knc could simply lower their voltage why would they spend 10 million on 20nm nre just to get 0.4w/gh at the chip level? What makes you think the next gen of asics won't be similarly underclockable?every chip is underclockable. Bitmain can achieve anything from 1w/GH to 2.2w/GH depending on the frequency and voltage. I am sure 0.8w/GH could be achieved if they wanted to go even lower |
|
|
|
I don't quite understand why 16 chips won't reach 50GH/s. How is this not scalable?
If you double the powersupply and use active cooler too get rid of the heat you could reach 50GH/s right?
My experience with The rev1 chips on v1.2 H-boards was that with a good board you could get up to 25-30GH/s stock. With a small pencil mod AND either decent airflow OR heatsinks 30-34GH/s was achievable. With a moderate pencil mod AND decent airflow AND some heatsinks on the backside 35-38GH was achievable. With all the above plus a bit of luck, or some very dedicated tuning, 38-42GH was achievable. At this stage the VRM is about 150% its maximum amperage specs but could handle it. The rev2 chips are about 15% more powerful at the same wattage presumably. I recently had a 38GH board (3rd slot) stop hashing and fail to show up in .stat.log. my assumption is that either the TPS53355 burnt out or the inductor burnt out. Haven't made much headway on the diagnostic process |
|
|
|
got mine clocked at 212.5Mhz which yields about 1080gh/s consistently. tried 225Mhz and while it shows 1200 gh/s the pool side shows under 1000 gh/s, so there must be a lot of hardware errors or stales or something. without some kind of voltage mod or increase i dont know how much higher than 212.5 you can go. i dont have the settings to try any other ranges between 212.5 and 225 though.
if you do the following the config will save even through reboots:
1. have 4/10/14 firmware
2. ssh into miner (user root, password admin)
3. cd /config
4. vi asic-freq.config
5. comment out current stats (place a # in front of lines that dont have it)
6. add these:
option 'freq_value' '0801' #212.5M
option 'chip_freq' '212.5'
option 'timeout' '40'
7. save
8. you can either type reboot or you can go to the web interface and click save on the miner configuration page.
ADD'T INFO: - you can combine steps 3&4 as: vi /config/asic-freq/config I finally got around to doing the calculations myself for the antminer S2 - and so far it looks good (testbench was a U1, which on success menat the code could be applied to the S2 safely) 5002 = 206.25MHz 5082 = 212.50MHz (different from the one above, but both should work fine) 5102 = 218.75MHz all three meet the requirements stipulated by the above rules and pattern. 5082 is currently giving a hashrate of 1095GH and error rate of about 0.95% after 15minutes. (212.5 is 8.5% higher than the stock speed) |
|
|
|
 It comes from the Antminer U1 overclock manual. But I think this should be compatible to Antminer S1, too. Let's see the example of why 0780 and 4F81 are 400MHz. HEX BINARY 15/BS M N OD 0780 0000 0111 1000 0000 00 00 0111 1 = 15 000 00 = 0 00 = 0 which NO=1 4F81 0100 1111 1000 0001 01 00 1111 1 = 31 000 00 = 0 01 = 1 which NO=2 5F82 0101 1111 1000 0010 01 01 1111 1 = 63 000 00 = 0 10 = 2 which NO=4 Fout = 25 * (M+1) / ((N+1)*NO) Fout(0780) = 25 * (15+1) / (1*1) = (25*16)/1 = 400 MHz Fout(4F81) = 25 * (31+1) / (1*2) = (25*32)/2 = 400 MHz Fout(5F82) = 25 * (63+1) / (1*4) = (25*64)/4 = 400 MHz Actually, 5F82 is also 400 MHz, but due to the instruction below 500 <= Fout * NO <=1000 Fout(5F82) = 400MHz, therefore, Fout*4 = 1600 and over 1000. So I think I should not config 5F82 for 400MHz, according to this manual. I don't say that it does not work (I tried it before understanding the HEX, it worked) but you have also another 2 freq_values which give you the same MHz and get along with the instruction in the manual. A little bit too long, anyway, I hope you enjoy my lecture, lol. Any questions are welcome. If this is useful, any donations/tips are welcome => 12QAQhbmTzV7sJ9sg8xT96JAneE4S89sS6  I finally got around to doing the calculations myself for the antminer S2 - and so far it looks good (testbench was a U1, which on success menat the code could be applied to the S2 safely) 5002 = 206.25MHz 5082 = 212.50MHz 5102 = 218.75MHz all three meet the requirements stipulated by the above rules and pattern. 5082 is currently giving a hashrate of 1095GH and error rate of about 0.95% after 15minutes. (212.5 is 8.5% higher than the stock speed) tips accepted at 1LQZnkJuUbdeVdxGjR27zkNJWSW4FSLBhh |
|
|
|
Here they are ... It comes from the Antminer U1 overclock manual. But I think this should be compatible to Antminer S1, too. Let's see the example of why 0780 and 4F81 are 400MHz. HEX BINARY 15/BS M N OD 0780 0000 0111 1000 0000 00 00 0111 1 = 15 000 00 = 0 00 = 0 which NO=1 4F81 0100 1111 1000 0001 01 00 1111 1 = 31 000 00 = 0 01 = 1 which NO=2 5F82 0101 1111 1000 0010 01 01 1111 1 = 63 000 00 = 0 10 = 2 which NO=4 Fout = 25 * (M+1) / ((N+1)*NO) Fout(0780) = 25 * (15+1) / (1*1) = (25*16)/1 = 400 MHz Fout(4F81) = 25 * (31+1) / (1*2) = (25*32)/2 = 400 MHz Fout(5F82) = 25 * (63+1) / (1*4) = (25*64)/4 = 400 MHz Actually, 5F82 is also 400 MHz, but due to the instruction below 500 <= Fout * NO <=1000 Fout(5F82) = 400MHz, therefore, Fout*4 = 1600 and over 1000. So I think I should not config 5F82 for 400MHz, according to this manual. I don't say that it does not work (I tried it before understanding the HEX, it worked) but you have also another 2 freq_values which give you the same MHz and get along with the instruction in the manual. A little bit too long, anyway, I hope you enjoy my lecture, lol. Any questions are welcome. If this is useful, any donations/tips are welcome => 12QAQhbmTzV7sJ9sg8xT96JAneE4S89sS6 I finally got around to doing the calculations myself for the antminer S2 - and so far it looks good (testbench was a U1, which on success menat the code could be applied to the S2 safely) 5002 = 206.25MHz 5082 = 212.50MHz 5102 = 218.75MHz all three meet the requirements stipulated by the above rules and pattern. tips accepted at 1LQZnkJuUbdeVdxGjR27zkNJWSW4FSLBhh |
|
|
|
perhaps Ken and his entire staff do not yet realise that going to http://virtualminingcorp.com/ shows a ton of different products and none of these are actually for sale? The only product for sale is almost impossible to find and differentiate from all the other cancelled gear unless you have the direct link Ken - its no surprise VMC cant sell more than a few dozen hashfast boards when the homepage of the site showcases a cancelled project that became obsolete months ago. (768GH/800W/$9000) |
|
|
|
The better method would be to make 16-chip boards by the official spec, and using the newer revision of chip that should allow each 16-chip board to push ~50GH.
The 8-chip board just results in more slots on the m-board being filled which is an issue if you don't sell m-boards to go with it. just my 2c
Hash rate is very dependent on amperage an voltage and with positive correlations to both. A rev.2 16 chip board wouldn't reach 50 easily with out some serious heat extraction and over clocking It would be no problem at all if another 30A power phase was added to the board. But then you have 70+ watts to pull out of a board That's some of the rationale I used when deciding 8 vs 16 I'm wanting to get the max out of these chips and and doubling the amp ration they each get is one way to do so. If I knew that 16 could crunch along at 50 It would be a no brainer decision. But in most real life situations a rev2 16 ASIC card with heatsinking on the vrm and each ASIC OV'ed to 0.833 would probably pull ~35-40 maybe 43 TL;DR: a 16 chip card won't run at the same per chip hashrate as an 8 chip card using the same 30A power supply.
This is because of thermal and electrical constraintsI was talking about this rev2 chip: http://www.bitfurystrikesback.com/product/bitfury-55nm-rev2-samples/864 cores instead of 756 (14% more) and presumably the same or slightly lower power usage. All my 1.2 h-boards can be overclocked to ~32GH without heatsinks and up to 38GH with heatsinks. 14% more means you could get 36-45GH per card if overclocked. (which at current pricing and considering a 3week production, will only be able to sell for ~$75 each or possibly less) |
|
|
|
|
I hope we can get some better info for overclocking. A hex for clock speed around 215MHz would be great for a 5-7% hashrate boost if you swap in a 1200W PSU. anything higher (such as 225MHz) does not work well without knowing which resistors to modify in order to overvolt the system.
who calculated all the clocks for the S1 (393,387,381, etc)? We really need the same done for the S2 in the 210-220MHz range. (225MHz gave me 7% errors - 200MHZ gives 0.4%)
|
|
|
|
so 3 1/2 hours and i got it running a few guys pm'd .... thanx but wound up using some phone support.... pi wasn't connected and a fan male prong was bent...got it back straight with a wine key, i pour a lot of wine at my job... so its running now,, what should i expect? is all the sudden its gonna stop if i don't flash something or replace an sd card??? i cant believe i made it this far.. the interface looks so crappy like calico vision, do i up date it??
pool recommendations??
if its running leave it be as for pools, i suggest eligius. you get newly minted coin, no fees, and nmc merged mining. also something like 9% of the network so its an ideally sized pool IMO |
|
|
|
|
I actually looked at my S2 unit (Batch 2) for the first time yesterday after almost 2 weeks of remote monitoring only - and I was impressed.
Those with shipping issues - your UPS guy is terrible. My unit had rubber strips on the lid to prevent the cards unslotting, and all the plastic slide-rails were still in place and firmly mounted to the metal cross-rails of the case. The boards swap in/out easily and I like the idea of leaving the control board (which has space for a LOT of extra features!/?) as its own swappable card.
thoughts:
1) I got a 1000W GOLD PSU. looks pretty new, still a clear sticker over the brandname on the fan grill, no 'months of dust pileup' like reported from Batch 1
1a) The load at my PDU registered just under 13A - or over 1300W. I don't know if this means the PSU is overtaxed (>1000W draw) or the gold-rated efficiency drops off severely at 100% load. If anyone else can confirm this drastic efficiency difference it might mean that a 1000W gold is not the best option. 1100W-1200W might provide significant (100W+) power savings due to the efficiency at 80% load compared to 100% load.
1b) I would love to see a simple revision if it can be done easily: include 1-2 external PCIe connection ports. Connect these to a relay so if they receive power, 2-3 of the hashing blades switch to draw power from them, lightening the load on the included enermax PSU. relay clips off if the secondary PSU is detached, and current to the blades resumes from the primary PSU.
1c) Alternatively, an option to disclude the included PSU in favor of 6 external PCI-e hookups would be great for those with lots of power equipment already, such as server supplies. (similar to what KNC did)
2) the hashing boards and LCD all look great. From a rack-mounting POV this machine is probably one of the best. (1!1!1!1 would probably back that up)
3) I hope we can get some better info for overclocking. A hex for clock speed around 215MHz would be great for a 5-7% hashrate boost if you swap in a 1200W PSU. anything higher (such as 225MHz) does not work well without knowing which resistors to modify in order to overvolt the system.
|
|
|
|
|
Show Posts
