Topic: HashFast BabyJet users thread - page 40. (Read 69008 times)

can't we be a little more professional and give out the MD5? Asking too much? (yea, I'm getting frustrated, can you tell.)

For all the people here that are running on something other than the RPI's, Here's the source code for cgminer 3.9.0h2 (same version we're currently using on the RPI).  This version has somewhat correct hashrate reporting, and has a few other things, such as the per-die voltage and temperature readout.  It seems to be the most stable during our testing.  One of our top Engineers is currently working with Con Kolivas right now to get some of these things improved.  Thanks Con!

cgminer 3.9.0h2 source (can be complied on most Linux distros): http://setup.hashfast.com/cgminer-3.9.0h2.tar.gz

I'd love reports back to see if this improves your mining performance/stability.  Again; please ignore the strings of HW errors, this should not be a problem.

-Phil

Au Contraire.

I downloaded your minepeon raspi image as soon as I learnt of it:

MinePeon Version
0.2.4.3hf8
Miner Version
cgminer3.9.0h

This has without a doubt, the "1969" date/time upon bootup problem and the daemon in part "ntpd -g" is not correcting the date/time as it should. I said to heck with it and wrote a small perl script that manually setup the time properly upon boot (I am too lazy to go treking thru arch linux hunting for whatevers, howevers.)

ntpd -g should correct things, but it doesn't. and because it doesn't, cgminer won't run. and because cgminer won't run then of course baby jet won't make any "noise". Initially this all puzzled me like I said in my post and I spent money and RMA requests in my initial attempt to "fix" it.

I downloaded the image about two days ago and maybe something changed in the meantime? If you could download your image yourself, load it on a mem card and then boot up off it, you will experience the year "1969" error, and not the more common/expected year "1970" reset. And how come ntpd -g doesn't correct a thing? Dunno why, you tell me?

Maybe I missed it, but what is the MD5 for your image? Remote outside chance that it got corrupted on its way to me here in Hawaii...

I will say windows is working great for me as well. I got an old pc running my hashfast and BFL hardware.

yes, thats right I was lucky enough to pick the 2 best companies

@HF Engineer, Can the unit itself get software updates that make it faster. I am simply asking as right now like i said im running windows but if new builds that improve the speed come out id much rather switch to the RPI to update the unit.
Also will you eventually provide a way for windows users to update their unit?

Direct USB cable from PC to the BabyJet. Also only a little litecoin mining on the GPU. I also tried without anything else on the PC.

How do I get the cores warmer without overclocking? Space heater?

I dunno what to say.  I've got two pretty different BJs, arrived at different times, built in different places, different "rev" stickers on the boards, different hash rates, and 3.11 works for me.

Cables or USB controllers?  Are you doing other stuff on the PC?  (Mine only controls BJs, plus a little Litecoin mining on the GPU.)

My experiments suggest you want the cores warmer, whether or not you overclock.  Mine seems to mine a little faster around 70c.

We just brought several new support people on-board, and set up a new queue system.  I'm CC'd on all the support emails and will look at each and every one.

Know that one bad core out of the 384 that are in your BJ can spew a LOT of errors.  New firmware that's being tested now will hopefully fix some of these issues.  We will push it out as soon as verification & regression testing is complete.

Don't let a spew of errors scare you off.  Check the pool's reported work rate and go by that.  Almost all of my test board here spew lots of errors, but it's usually just one or two "blabbermouth" cores doing this, and the overall hashrate is still good.

If you are attempting overclocking, (not recommending you do) but ignore these errors and watch for the "Work Watchdog Timer Reset Imminent" messages.  If you see these, then you've gone too high.  Back off in 5mhz increments until most these go away (one on occasion is ok) and then check your hashrate at the pool.

-Phil

My PC cgminer crashes all the time on 3.11. I've never tried 3.12 on the PC. The longest I've ever had 3.11 working successfully is 7 hours - then it just quits (no error to diagnose).

The only stable platform so far is 3.09.0h2 on the RPI.

Is there a way to overclock the cgminer on the RPI? My cores are all now running at between 56C and 66C. Seems like a waste.

I left both 3.11 and 3.12 on there for you guys.  There is a symlink pointing to 3.9, just change it to 3.11 and you're rolling on 3.11.

SSH into your RPI, the login is "minepeon" and the password is "peon" unless you've changed it, then execute these commands:

If you decide that 3.9 was better: (I think you will)

I'm running 3 RPI's here with different configs of some our reject hardware, some very unstable, and 3.9 does seem to be great.  Uptime is 2 day 3 min 2 sec on one.

-Phil

Guess it's not so Easy2Mine after all.

You got two options (edit: see also dbbit below, who disagrees):

A) You can wrassle with the raspberry pi and follow HF engineer's advice, and I have no reason to think it isn't good, or;

B) You can getchyerself an old PC, the stock version of cgminer 3.11, switch the driver with Zadig (easy), and with a tad of patience hotplugging it a couple of times, unless you have a bad machine, you will be up and running, STABLY.  You can also probably overclock it, depending on your huevos and chips.

To me, it's a no-brainer.  No offense, Engineer, but I just don't like the whole rpi thing...

Also, I've had zero downtime since setting the machine up - except when I flirted with 3.12, and a few voluntary minutes of downtime when I was trying out things in the .conf file.  

The PC might not be cool, but dood, your BJ is presently capable of 'printing' maybe USD $65 a day.  Do you want to leave that money on the table for a number of days while you hassle with the $39 Raspberry Pi controller?

I am glad I just only ordered 1 BJ.
What a FUCKING piece of crap is this.
I don't have hope this will ROI anymore, even when they start to ship the MPP.
You don't know when this shit reboots.

I've tried underclock but I have yet to find a frequency that doesn't produce a ridiculous amount of hardware errors. I think I just have defective hardware.

Underclock it until you get a hold of someone.

--hfa-hash-clock 530 (find a rate below 550 that doesnt produce hw errors)

From Hashfast that just responded to my week old RMA request:

Thank you for your message.

We are currently receiving a very high volume of customer support cases, and as a result, we might be slower to respond than usual. However, we will get back to you as soon as we can.

You can check the status of your message at the following URL:

https://hashfast.fogbugz.com/default.asp?blah-blah-blah

You may want to save your case's tracking ticket: blah-blah-blah

Please reply to this message if there's anything else we can do for you.


--
HashFast Support
[email protected]


Okay, HashFast is swapped. Canned replies, trouble tickets that go nowhere and all the while asking if there is anything more they can do for me.

I'll cut them a week or two break... (And KNCMiner was just like this in the beginning. But only a month late and their second batch Jupiters were under $5K and are now doing 670GH/s. People that ordered the second batch got their Jupiters under three weeks from the time of payments. As far as USD flow, Jupiters are pure profit miners now.)

I have two BabyJets now. One works great. It's been running at 430 GH/s pretty consistently for over a week now (using the average stats from eligius).

The other one I just got today and is broken, right out of the box.

According to the cgminer its running at 380 GH/s but in reality (according to eligius) its running at about half that. The reason is the ridiculous error rate that I'm getting.

Here is the cgminer stats for the working bj:

 cgminer version 3.9.0 - Started: [2014-02-01 15:02:13]
--------------------------------------------------------------------------------
 (5s):452.8G (avg):461.7Gh/s | A:19290234  R:92767  HW:64312  WU:11909.7/m
 ST: 2  SS: 0  NB: 417  LW: 20691920  GF: 0  RF: 0
 Connected to stratum.mining.eligius.st diff 256 with stratum as user xxxx_BabyJet1
 Block: 1e5dc7ba...  Diff:2.19G  Started: [20:27:33]  Best share: 6.18M
--------------------------------------------------------------------------------
 [P]ool management Settings [D]isplay options [Q]uit
 HFA 2:  71C/.79V 79C/.79V 71C/.79V 70C/.78V 498.9G/458.7Gh/s | A: 4331326 R:20480 HW:14351 WU:11964.2/m
--------------------------------------------------------------------------------

As you can see, its been running for 2 days now at 460 GH/s (430GH/s in reality according to the pool). The HW errors are just 64,312 and aren't increasing.

Now, the broken babyjet (running stock, no overclocking or modifications):

--------------------------------------------------------------------------------
(5s):407.5G (avg):378.3Gh/s | A:54179 R:512 HW:611323 WU:6780.2/m
ST: 2 SS: 0 NB: 4 LW: 89015 GF: 0 RF: 0
Connected to stratum.mining.eligius.st diff 128 with stratum as user xxxx_BabyJet2
Block: 1391fb73... Diff:2.19G Started: [15:29:41] Best share: 47.1K
--------------------------------------------------------------------------------
[P]ool management Settings [D]isplay options [Q]uit
HFA 1: 75C/.79V 77C/.79V 68C/.79V 59C/.82V 393.2G/383.3Gh/s | A:33279 R:512 HW:369376 WU:6803.7/m
--------------------------------------------------------------------------------

Its only been running for half an hour or so and already the HW errors are 611,323. Ten times the amount of the first BJ!

The temperatures are the same for both bjs. In fact, the broken one seems a bit cooler.

It seems to me that I've received a broken bj.

Has anyone had any luck returning or exchanging them??

Okay I have my BJs set up. 3.11.0.  So far, so good.

I have to go to work (they're already pissed at me for staying home to receive this fedex shipment), but these are my thoughts so far:
- Risers and screws were jingling in both units when I received them, the problem is the extra set of risers on the bottom seem to fall off.  HASHFAST, if you're reading this, tell CIARA to tape these to the inside of the case in a bag instead of extra screwing them in.
- Hashrate for first unit is 398 GH/s and the second 405 GH/s, the one is pulling 460 W while the other is pulling 476 W (Blue Planet energy meter, usually more accurate than a Kill-A-Watt).  Temps are 76 C and 79 C.  Not exactly the efficiency we were promised, and not a lot of room for OCing, oh well.  Pool confirms these hash rates.
- Would be nice to get a voltage control in cgminer, I want to try undervolting.
- Comes with a nice Seasonic 1000 W Platinum PSU, unfortunately there is no 24+4 pin and 8-pin 12V cables for a motherboard in case you want to recycle it.  Also, shows that even at 90% efficiency these are actually less efficient than my Bitfury rig running at 85% efficiency and overclocked.

But, well, at least they're up and running!  Seem to work nicely as space heaters.

Load drivers with zadig

http://zadig.akeo.ie/

Okay I may seem stupid but im new to this and I cant get connected using the raspberry pi at all. So does anybody know the drivers for the BJ I switched it to winUSB but I still cant get Cgminer to read it. If someone could please help me setup cgminer as I cannot get the pi working after many attempts and I am new to mining any help would be great.

I feel like the Rodney Dangerfield of this thread.  Don't get no respect.

Fellas, I've been saying for days that on a PC you need to use 3.11 because 3.12 is subject to crashing (apparently related to the firmware "watchdog" and cgminer not getting along).

But if you guys wanna keep ignoring me and keep the network hashrate a little lower with your offline machines, you can be my guests...