Topic: Bitmain AntMiner U1 Tips & Tricks - page 41. (Read 106330 times)

http://imageshack.us/photo/my-images/36/vp5a.JPG/


okay I have been playing around with them and got 15 to work on 1 pc but 2 went zombie.

I have found some type of hash rate bug..  lets say you are running 6 at 1.6 or 9.6 gh  with windows 7 and the custom cg 3.8.5 build

both cex.io and  bitminter seem to under report    the rate by a lot.  the 6.9 gh on bitminter reading is for 8 sticks



this shows a big difference in hash rate for the antminers using a custom build and the ice fury using bfgminer 3.9

I know that bitminter reporting rate varies but I had the same under reporting with cex.io .

If someone can get  more then 6 on a hub to work please see if you have this error in rate reporting

I resurrected my >10 year old Pentium 4 and installed Ubuntu 12.04 Server on it.

Installed any dependencies needed, most of which are listed in the regular cgminer README document in the 'Basic *nix build instructions:' section.

Combined that with parts of iluvpcs instructions and made the modification to driver-bmsc.c as fractalbc outlined. (I did run cgminer without the update and cgminer did crash.)

My gratitude to the both of you. 

Two things.

First of all, the "--bmsc-options 115200:20 " is MANDATORY.  I was not using it and having it on the command line made all the difference in the world.

Next, what processor / OS are you running?  i.e. does "uname -p" say  "i386" or "x86_64"?

The same exact sequence on an intel i3 running ubuntu 12.04 x86_64 gives a compiler warning that it will crash because of a memory overflow and really does.  Fixing the memcpy from trying to copy 5 bytes into a 4 byte uint32 stops the crash and lets the program run.  Your compiler / OS / processor may be more forgiving of the driver trying to stuff 5 bytes into a 4 bytes entity.

FWIW, what I did to make it work on an i3 running ubuntu 12.04 was

--- a/driver-bmsc.c
+++ b/driver-bmsc.c
@@ -1276,7 +1276,7 @@ static int64_t bmsc_scanwork(struct thr_info *thr)
                goto out;
        }

-       memcpy((char *)&nonce, nonce_bin, sizeof(nonce_bin));
+       memcpy((char *)&nonce, nonce_bin, sizeof(nonce));
        nonce = htobe32(nonce);
        curr_hw_errors = bmsc->hw_errors;
        submit_nonce(thr, work, nonce);

which is essentially to only copy 4 of the 5 bytes returned by the driver before the byte swap.  Only 4 bytes are used by the code so this is a safe change.

After spending hours in the lab...  


Larger
http://i1273.photobucket.com/albums/y416/stumbling-orangutan/17977-5395-28979_zps62770a10.jpg
http://i1273.photobucket.com/albums/y416/stumbling-orangutan/32526-1913-8642_zpsa49d98b0.jpg
http://i1273.photobucket.com/albums/y416/stumbling-orangutan/18575-19537-19672_zps0ed14d60.jpg

Screws: M3 .5 pitch 5mm length

The grey thing is the thermal conductive pad.
The grey flecked 3mm allen key is the BitMain Antminer U1 collector's edition model.

Edit: Pretty much all of the screws on the U1s I had could use a little turn. I doubt it REALLY matters but it makes me feel better.

20 minutes and I have it running on my BAMT GPU linux box.  So I decided to try this on one of my BAMT devices with 4 GPU's running mining Scrypt (NO I'm not mining Scrypt with these, just using the same Linux device).

I logged in on my BAMT device and did this exactly:

git clone https://github.com/bitmaintech/cgminer
cd cgminer
sudo apt-get install libusb-1.0
sudo apt-get install libudev-dev
./autogen.sh
./configure --enable-bmsc
make -j 6

and then did (as in the example from the .PDF)
./cgminer --bmsc-options 115200:20 -o 50.31.149.57:3333 -u ktzhan_3 -p 123 --bmsc-freq 0781

and viola (I plugged in 11 to a USB 49 port board) - going to go get more and plug them in a minute ,have to go eat dinner.

 cgminer version 3.8.5 - Started: [2014-01-01 01:25:35]
--------------------------------------------------------------------------------
 (5s):23.00G (avg):14.56Gh/s | A:422  R:0  HW:9  WU:198.8/m
 ST: 2  SS: 0  NB: 2  LW: 898  GF: 0  RF: 0
 Connected to stratum.bitcoin.cz diff 8 with stratum as user user.BE
 Block: 2c617726...  Diff:1.18G  Started: [01:25:37]  Best share: 690
--------------------------------------------------------------------------------
 [P]ool management Settings [D]isplay options [Q]uit
 AMU  0:                | 1.250G/1.000Gh/s | A:32 R:0 HW:0 WU:14.0/m
 AMU  1:                | 2.313G/1.074Gh/s | A:40 R:0 HW:1 WU:14.5/m
 AMU  2:                | 2.226G/1.482Gh/s | A:48 R:0 HW:0 WU:20.7/m
 AMU  3:                | 2.696G/1.593Gh/s | A:48 R:0 HW:0 WU:22.3/m
 AMU  4:                | 1.989G/1.407Gh/s | A:24 R:0 HW:0 WU:19.7/m
 AMU  5:                | 3.439G/2.000Gh/s | A:59 R:0 HW:3 WU:26.4/m
 AMU  6:                | 1.701G/1.037Gh/s | A:16 R:0 HW:0 WU:14.5/m
 AMU  7:                | 2.582G/1.296Gh/s | A:40 R:0 HW:1 WU:17.6/m
 AMU  8:                | 2.573G/889.2Mh/s | A:40 R:0 HW:1 WU:11.9/m
 AMU  9:                | 2.547G/1.630Gh/s | A:59 R:0 HW:3 WU:21.2/m
 AMU 10:                | 2.090G/1.185Gh/s | A:32 R:0 HW:0 WU:16.6/m
--------------------------------------------------------------------------------

 [2014-01-01 01:27:09] Accepted 1658c6ef Diff 11/8 AMU 1 pool 0
 [2014-01-01 01:27:10] API bind to port 4028 failed (Address already in use) - A
PI will not be available
 [2014-01-01 01:27:11] Accepted 16b9bf0e Diff 11/8 AMU 3 pool 0
 [2014-01-01 01:27:12] Accepted 12ccba67 Diff 14/8 AMU 9 pool 0
 [2014-01-01 01:27:14] Accepted 1625e5ea Diff 12/8 AMU 8 pool 0
 [2014-01-01 01:27:14] Accepted 135e92f3 Diff 13/8 AMU 9 pool 0

I have wrestled with my share of USB Block Erupters and at the moment trying to get more than one of these Antminer S1 to run at once is just plain ridiculous.  
Even with one running I am getting almost 100% Rejects. Something is amiss.

I am anxious for an update to cgminer or better yet (as mentioned kfactor) bfgminer.

(Edit: Windows 7)

stick a little adhesive-backed heatsink to the U1. you can get them on ebay for pretty cheap, i just got some like this that you could fit 3 or 4 on a U1:

I just can't get more than 7 to run at a time, and I've tried different port configs across multiple hubs. The 8th always comes in error -12.

Also, cgminer is crashing if I prepopulate ports before starting it up

Can't wait for bfgminer support ...

Having to plug in USB's one at a time and reinstall drivers on each of them is a huge pain in the ass.  

May not be an issue for those running 1, but i'm annoyed as hell right now.

just not sure you can use a paste like Arctic Silver that is conductive. I have built a couple fan contraptions already for mine. Debating on buying couple more U-1 or waiting till price drops a bit more.

I haven't even got mine home yet but from looking at it the U1 uses a thermally conductive pad between the PCB and heat sink.

To increase cooling you would increase the surface area of the heat sink or increase air flow.

Much obliged klondike_bar. I look forward to some experimenting (after a Digi-Key order). 

On another note, a summary of Bitmain U1 links:
AntMiner-U1 user guide: https://github.com/AntMiner/AntGen1/blob/master/manual/AntMiner-U1%20user%20guide.pdf
AntMiner AntGen1 cgminer for U1: https://github.com/AntMiner/AntGen1/tree/master/cgminer
Chip datasheet for the BM1380: https://github.com/AntMiner/AntGen1/blob/master/datasheet/BM1380_Datasheet.pdf

has anyone taken their U-1 apart yet to see if a better thermal paste can be used to keep it cooler? I want to see some pics if so.

you can likely increase the frequency by 50MHz without touching the voltage. but to push beyond 275MHz you will need to increase voltage or start seeing some errors and rejects start to submit.

for reference, the antminer systems are 1.1V for operation between 350-400MHz

I understand that aspect of it.
My question is more to the point of whether increasing voltage will allow the higher hash rates to be attained stably.

Installing a 10K ohm at R2 and leaving the 1K at R1 would increase the voltage to .88 volts.
Seems like that would be relativity easy if you have a 10K ohm resistor (Surface Mount 603) laying around.

I am not an electronics guru. I am a merely man with some soldering equipment and a willingness to injure myself and damage sensitive electronic equipment.

So pardon me if this is a silly question...

Am I to take it that by replacing the resistors per this chart we can increase the voltage and thus reach the hash rates beyond the published overclockable speed? (Which would presumably require increasing the ability of the U1 to dissipate heat.)


 

src: https://github.com/AntMiner/AntGen1/blob/master/manual/AntMiner-U1%20user%20guide.pdf

I am having a hard time believing it would be that easy but the ASICMINER upgrades were (after a huge thread of experiments by many talented individuals https://bitcointalksearch.org/topic/block-erupter-usb-overclocking-hacking-241652)

I have two U1's running with my 22 BEs on cgminer 3.9.0, though they are only running about 20% faster than the BEs.  Running a U1 by itself with the bitmain version of cgminer 3.8.5, it runs at close to 2 Gh/s, but the bitmain version does not recognize the BEs.  (I'm running on Win7 and Win8).  I noticed that cgminer lists the U1s as AMU.  Using -n finds all of the usb devices, but fails to distinguish which is which.  Is there any chance that a version of cgminer can be compiled to properly run both BEs and U1s at their 'native' speed?

What are you running?  I run several flavors of linux and originally used the binary supplied (worked on only one of my machines) but then I used the source code for the modified version of cgminer from https://github.com/bitmaintech/cgminer, git cloned from there, then ran ./autogen.sh, then configure --enable-bmsc, then make -j 6, and voila' worked just fine.  It's running right now on a machine that I'm still using bfgminer for proxies, red furies, and chilis.

If you're on a Mac you could still try to build from that source since it looks like a modification on top of a full cgminer (i.e. if cgminer 3.8.5 normal can build on your system, then this has a good chance to as well).

Hope this helps!

Hopefully they'll get their support merged into the mainline BFGMiner/CGMiner soon so those of us NOT running windows to mine can't start using these!