So does anyone have some information to share about underclocking?
I had a scary moment today when my main circuit breaker tripped. I'm running a large number of miners including 3 S1's. I underclocked everything I could to prevent it from happening again.
I changed my S1's from 375 to 325. I dropped from 190GHs to 166GHs which is fine by me. This isn't about the extra pennies I'm losing by not overclocking. This is about power efficiency and waste heat generation.
Does anyone have stats comparing watt usage for 375 to 325? Is there even more significant efficiency to gain by going lower?
I run mine underclocked (for efficiency due to electricity costs). I get 150GH/S per ~200w @ 300, rather than 180GH/S for 380w. I essentially went through the same process described in the link pvnamk19 posted above. So you had to do the pencil mod to get to 200w? I'm not really interested in going to those lengths to undervolt. I'm only wanting to underclock. Sorry, I didn't read the entire thread so maybe my scenario was addressed there. no pencil mod means 2 watts per hash. so a 350 clock gets about 180 gh and around 360 watts with a plat psu a 300 clock gets about 150 gh and around 300 watts a 275 clock gets about 140 gh and around 280 watts but read below: please get a fucking pencil lightly stroke each of the 8 resistors 2 times do not measure the ohms takes 2 minutes. set the clock at 300 and see what the results are. about 150 gh and 200 watts is pretty much what happened for me, it is easy to do this. below one at stock volts and 393 freq gets me 200gh and 406 watts and one at undervolts 275 freq gets me 140gh and 201 watts I am thinking about undervolting, got myself the HB pencil, but the resistor that you supposed to touch is so tiny and there is another object nearby (C10). How you were even able to do it reliably? magnifying glass workstation? i think that you supposed to use the multimeter (ohm meter) as well. I have 3 running on an evga 1300 supernova psu about 420 gh pulling 563 watts at the k-a-watt plug that is 1.34 watts a hash. I can't tell you how easy it is. I have had cataract eye surgery in each eye and I can do it in 5 minutes I used a number 6 pencil on one mod and a number 2 pencil on the second mod. I deliberatively was sloppy on getting perfect ohm measurements after the mods. I have an hi-end meter all resistors read 4.46 or 4.47 k ohm before the mods. I allowed the resistors to range between 2.4 k ohm and 3.3 k ohm after the mod . The key is your ssh. you may want : 250 freq 275 freq 300 freq if you fucked the resistors up a lot you get x's on your asic's just go to a lower freq. I stroked the resistors in 1 direction away from the one next to it and towards the empty space. I used a magnify glass on a stand http://www.amazon.com/Carson-DeskBrite-Illuminated-Magnifier-LM-20/dp/B003EW1ZNC/ref=sr_1_4?ie=UTF8&qid=1401213622&sr=8-4&keywords=magnifying+glass+standthe photo below is with the glass.  Excellent work, Phillip! |
|
|
|
|
Did those of you that upgraded after running the provided PSU for a few days notice a boost in hash rate after the upgraded PSU?
|
|
|
|
Got my S2 unit on the 23th and had it running full speed without a hitch til just now... had to move it to another room and it never woke up again... seems PSU is dead... only worked for like 4 days... shit...
Hope bitmain sends a replacement fast...
The BM RMA process is super-fast now--I had a blade shipped back to them 2-day express and they returned it to me in the US 2-day. Four days, round trip, from the US to China! |
|
|
|
I thought if you rma it back to china they give you their ups account and let ups handle it ? Is this service not for free ?
Nope I just RMA'd a S2 blade back on BITMAIN's UPS account from the US, no problems at all. I've never had a problem with an S1, but I don't see why they would deviate from this practice. Perhaps they haven't provided RMA shipping costs in the past, but they do now. |
|
|
|
Are the duplicate shares not proportionate to the physical numbers of chips on the machine? We've got 10x the chips as an S1, which I presumed was the cause of the high discard rate. Am I incorrect?
Unfortunately yes: s1s made in 24 hours 0 duplicated shares, s2s have made an average of 12k duplicated shares  Thanks for the data. Could the throughput be to blame? Sheer numbers of calculations requested to hash? My S2 spits our discards like it's going out of business, and it doesn't bother me too much as the number of accepted shares versus rejects is reasonable. |
|
|
|
Compared to the other miners I have in my farm (Bitfury, A1's, S1's), it seems the S2 submits a disproportionate amount of duplicate shares.
Has anyone else noticed this, and aware of possible solutions ? Isolated case ?
Are the duplicate shares not proportionate to the physical numbers of chips on the machine? We've got 10x the chips as an S1, which I presumed was the cause of the high discard rate. Am I incorrect? |
|
|
|
New price for AntMiner S1: 0.439 btc each one, shipping cost included.
Is the rumor of a very close s1 EOL true or it will still be sold for other months? Im sure they haven't made any new s1 for awhile now. The ones being sold now are probably the ones being sold off from the farms to upgrade to s2 units. The last five sent to me were completely new items. I received them over the last two weeks. The first 10 /12 I got were all used I received them from feb to early april. Good deal, I may order a few additional units this week, then. Thanks for the info. |
|
|
|
There is no reason the SD card problem would cause a breaker to trip
Of course it wouldn't... I was just reading how others corrupted their SD's when they cut power abruptly. I replaced the PS, then the BBB Board wouldn't get a link light.. Then I replaced the BBB with one of my own from my personal stash.. and it linked up fine. Then I put back the orig BBB (Because I wouldn't have a use for a partially populated BBB like the one in the miner.) Luckily, it linked up once again and decided to work. Anyway.. I'm hashing right along now. I did notice something else while removing the boards however... My boards have larger caps on them attached to the surface mount ones... but the solder is connecting to the cap right next to it on a few of them... maybe someone would know if this is a concern.  Regards, Rich Isn't this the "fix" that bitmain recently released? It appears to be, yes. Have they stated what problem these extra caps address? |
|
|
|
991 is perfectly fine, mine varies between 990 and 1010ish.
Cheers. I'll quit my whingin' then! Ah, I spoke to soon, I updated my miner configuration and the ---xx-- are back! It's dropped to 960 GH/s average as well. If you know its to do with that volt reg though, now you can take it to a local electronics shop and ask them either to resolder the legs or replace it entirely. The reg will cost literally $0.20 and they enjoy soldering. I just now caught this, Dogie--thanks for the advice. I'll see what BM suggests next, but I'm also happy to toss my local shop some cash. I suppose the board warming up could cause it to disconnect again, which would explain why it returned after running for a few hours. |
|
|
|
Sorry, I didn't mean this email to reach "non believers", pool owners, solo pool owners etc, and don't waste my time with your negative attitude. You go pool mine and I will do what I like. So back to the question.
Is there anyone out there who has cracked solo on an S2 or has Bitmain done it again!
It shouldn't be that difficult. Right!
Hey guy, you're not going to get too far here with that attitude--this is Dogie's thread after all, and he's contributed a metric ton of amazingly valuable information and walkthroughs during his tenure here. Did you google "solo bitcoin pool" yet? First hit is " http://bitsolo.net/" which looks to do exactly what you want. There is a thread here regarding it, use the search function to find that. Please spend some time lurking, and less time making demands of people more patient than yourself. Don't be a dick.Cheers. |
|
|
|
Useful but what UPS would be used for an S2?
At what extra cost?
You would only need to provide wattage for about 45 seconds, plenty of time for a monitor to fire off a script to ssh into the box and issue the shutdown command. Again, not elegant, but certainly possible. The extra cost if you don't have the kit laying about is probably not worth the investment, but it you're hashing in an environment that has frequent power issues it might be. A RasPi could do the work, and a consumer rated UPS would buy you the time you needed, although it would need to be large-ish. In my defense, I was just posting the solution to the card corruption, not the power grid. |
|
|
|
|
That said, it's not an elegant solution. I believe a firmware refactor could provide us a solution that will make power events less volatile, but that's (generally) in the hands of the BM devs.
|
|
|
|
I returned home earlier to find the power off - the meter ran out!
My S2 wouldn't restart (I did turn the switch off before putting the power back on!), just the blue screen and the fans spinning.
Stole an 8GB Micro SD card from my kids MP3 player (oops!) and installed the software on it.
Now back up and running :-)
Simple - although it took ages to find a Micro SD card!! I did back up all his songs (there weren't many on it) first and have ordered a replacement for him off ebay.
Now rewriting the image to the original 2GB SD card hopefully as a spare.
So basically it doesn't even need to be a power surge, just a break in the power can corrupt the SD so I guess even turning the machine off from it's switch could as well. Is there a safer way?
Thanks to those on here for the downloadable and the instructions/links, which were on page 86 of this thread.
SSH into the box and issue the "shutdown -h now" command at the prompt before flipping the switch on the PSU. I've power cycled my S2 a millionty times this way with zero corruption. SSH user is root, and the password is admin. You can use PuTTY in windows (for free) or the terminal in OSX or Linux with the "ssh root@your.s2.ip.address" to connect. |
|
|
|
991 is perfectly fine, mine varies between 990 and 1010ish.
Cheers. I'll quit my whingin' then! Ah, I spoke to soon, I updated my miner configuration and the ---xx-- are back! It's dropped to 960 GH/s average as well. Are you sure it's the board and not the pci slot? I've had X's show up from bent pins in the slot. If you swap the board out with another, do the X's still show up for that board in another slot or for the one in slot7? The board was originally in slot #3, and I did examine the PCI slots pretty well when I pulled everything out to reseat them the first time. The x's have returned since I pressed in all the chip and reseated the blade, I'm waiting to hear back from BM regarding the next steps. I presume we're going to RMA this blade, but I'll let everyone know what the outcome looks like. Cheers. |
|
|
|
Am I the only one who's noticed the new PDF on Bitmaintech's website with step-by-step instructions on how to solder capacitors directly to each hashing board....
"Guideline for soldering Capacitors to S2 hash board to fix chip disabling issue"
eek.... happy long weekend to me...
Me too, maybe. I've got a half board of x's. |
|
|
|
What does this mean on a chain? never saw a dash before.
oooooooo oooooooo oooooooo oooooooo oooo-ooo oooooooo oooooooo oooooooo
It's chain 2. Is that from the top, near the PSU, or from the bottom?
M
Chain two is second farthest from the PSU. #10 is closest to the PSU, #1 farthest away. Edit: The dash is probably okay, it'll more than likely clear when you reboot the device or restart cgminer (i.e. when you Save and Apply mining configurations). |
|
|
|
991 is perfectly fine, mine varies between 990 and 1010ish.
Cheers. I'll quit my whingin' then! Ah, I spoke to soon, I updated my miner configuration and the ---xx-- are back! It's dropped to 960 GH/s average as well. |
|
|
|
991 is perfectly fine, mine varies between 990 and 1010ish.
Cheers. I'll quit my whingin' then! |
|
|
|
|
Show Posts
