Topic: [Guide] Dogie's Comprehensive ASICMiner Blade Setup - page 66. (Read 580775 times)

have you tried to run proxy with --help ?

hi guys,

can anyone tell me what the -rt command does exactly in stratrum? I'm mining on slush but still a bit confused if I should use it or not.

Thanks

Pool software may say you anything according to statistical variation and whatever else it want to count in. If you want measure the raw performance check your blade's admin web-page for MH/s.

there is someone claiming to run a blade at 1.24v and achieving 16gh/s based on pool's share metrics

but probably closer to 15gh/s in the real console


is this feasible? or is this luck based on board quality variance, how would this affect longevity with proper cooling

It's normal (on High clock settings). It will fluctuate between 130xx and 12900 over time should you have good and stable connection with your pool/server/proxy.

my block erupter blade increases hash rate as time goes on, why is this?


2 days ago I was getting around 12200 MHS, yesterday 12800 MHS, and today 13000 MHS , no tampering on my part.

I have 4 120mm fans on them, two in front, two in back, and due to weather it is hotter in my apartment, but can someone explain this?

Look at this great thread and how much time I've saved you answering questions and not having to go through "omg my blade is broke" PMs

Cool, thanks everyone for your help! Turning voltages up to 1 brought 29/32 chips working (9.6GH/s!), so I'll raise them to 1.03 (or 1.05) and then leave them at low clock for the time being. Next stop, another yellow wire from the PSU, and a more efficient fan.

1.05V is the recommended undervolting figure.

You could try 1.03 but some of the chips couldn't really work on such low voltage due to variety in wafer production. If you are lucky 1.03 should be fine.

Yes that's because you undervolted too much and now one bank isn't receiving enough power. That can destroy your hash rate. I'm not sure why you messed around with your blade so much, they ARE plug and play if you do it properly.

I just know I wouldn't trust more than 5 amps down a single PSU wire, better safe than sorry! And at the moment it won't really be drawing any power due to sitting at idle; ie the chips aren't at their TDP and neither is the power draw, just strip another cable coming out of the PSU and add that to the other input. I really don't see the need for more grounds than positives two of each should be fine, maybe three if you want to overclock. And get some Arctic F12s they're only a few quid.

As for what he's done, yeah it is odd to do it like that. Although the smaller the run of low gauge wire the less problems there should be.

Thanks Caesium! s/101/100/ was exactly the fix.

The blade is now hashing merrily. Still only one yellow wire, but it's not even warm (keeping a close eye/finger on it though). Noisy but powerful tabletop fan blasting across the blades and behind (it's biiiig).

Five minutes in, hashrate appears to be 5.8 GH/s, power draw ~61W. Are these figures reasonable? I should point out that in the interest of safety I twiddled all those voltage screws down to their lowest setting which appears to be 970mV (varies by a few mV for each screw).

ETA: just noticed the row of chips is about a third 'x'ed. Is that just cause I've undervolted?

Nancarrow,

The blade restarting at 2:20 is simply because it's not hashing. It's a built in failsafe, reboot after a set period of time of no hashing, whatever the reason is, to attempt to clear the problem.

Your IP setup looks broken. You have 192.168.0.101 mentioned in several places, some of which can't be right.

The top IP box on the blade config is the IP of the blade itself. Pick one that your local network can see, 192.168.0.101 is probably fine.

The gateway box on the blade config should be your local router, although technically it's irrelevant since you're not asking the blade to route to the Internet; only another box on your local network. Probably it's 192.168.0.1, but check your PC's network settings and you can probably find it (edit: oh you said its 192.168.0.1, so you can fill that in correctly).

The mining proxy command line is also incorrectly using 192.168.0.101, and you're getting that error because 192.168.0.101 isn't the IP of the box you're running it on.
The easiest way around this for now is just to leave out the -oh parameter and it will listen on all available IPs. If your blade config is correct the actual IP of that box with the proxy on is 192.168.0.100, so you could use -oh 192.168.0.100

tom_o, I appreciate your concern but I'm not being QUITE as foolhardy as you imply.


I suppose they could. But I think they'd first have to get hot enough for me to actually notice their warmth when I touch them.


The restarts happen at ~2:20 exactly. Some sensible design decision from friedcat I suppose.


Well again they have to get hot before they can burn, which they are not doing. In fact, downgrade 'very warm' in my original post to 'feebly tepid'. However for the time being I have brought a house fan down from the attic, just in case.


I have done so, which is why I post, since in the world of Bitcoins, *nothing* ever works straight out of the box.


I have thought some more about these bloody power cables. I do not have a molex splitter, but my PSU does have two standard 4-pin molex connectors. I cut up one of them as per the setup guide. If I cut up the other and plug those into the remaining holes in the blade's green adapter thingy, that should do the trick, yes?

In fact, there's something I don't get about dogie's setup. If you use the molex splitter, then yes, you'll be splitting the current along (at least) two yellow wires from the molex junction up to the green adapter... but won't there still be just one yellow coming from the PSU? So there's still a point at which ~10A is flowing along a single wire? And I presume my proposed setup just above would resolve this?

so the plastic connector thingie is included with the blade?



- this

Those wires could melt or burst into flames, they have almost 10 amps going down them! This may also be the cause of the random restarts, either that or overheating; running without a fan is just asking to burn out your chips.

Stop being daft and read the bloody setup guide! Unless you want to kill your new 50BTC worth of hardware before you can even mine.


What would be better, getting it improperly setup today and running for a few hours till it dies. Or actually researching and setting up properly and getting years of use out of it.

Got my blade yesterday, holy fuckballs that was fast!

Now I have a plethora of questions, mostly related to the fact that I just want to get this blade up and running for a few seconds in a quick and dirty fashion before doing it all properly.

1) Power: I seem to have it working. But I don't have a molex splitter, so I have just used the one yellow+two black from a SINGLE molex on a cheapo old PSU I've got. Is this okay on a short term basis? As in, I only intend to mine for a minute or two at a time, for testing purposes?

2) Cooling: at the moment I have NO FANS AT ALL. Again, is this safe just for a minute or two of mining? I haven't actually been able to mine yet. Are the power/heating problems lower if it's idling? I'd have thought so... The heatsink and the back of the board (where all the ASIC chips are) feel very warm to the touch but not uncomfortably so (I can leave my hand there indefinitely).

3) Mounting: the blade is resting against an upright cardboard box. It's maybe 10-20 degrees from vertical, so there's some space behind the blade, and plenty in front. Any comments?

4) Configuration: this I don't get. Good news is, the config screen HAS come up, and I've set the blade's address to 192.168.0.101. I've also installed the slush stratum mining proxy on my laptop (just for now). I want to connect to my BTCguild account. Tell me what's not right yet. Laptop is at 192.168.0.100. Wireless router is connected to the internet, its local ip is 192.168.0.1, and I have an ethernet cable from the router to the blade.

Config screen: all numerical performance values are zeroed. Chips are all 'OOOO...' (that's good?). Low clock for the mo.
- IP   : 192.168.0.101
Mask: 255.255.254.0   
Gateway: 192.168.0.101
WEB Port   : 8000
Primary DNS   : 212.23.6.100
Secondary DNS   : 212.23.3.100
Ports   : 8332,8332
Server addresses: 192.168.0.100,192.168.0.100   
user:pass   : Nancarrow_Asicminer:123,Nancarrow_Asicminer:123

Mining proxy command: ./mining_proxy.py -o eu-stratum.btcguild.com -p 3333 -oh 192.168.0.101 -gp 8332

Complains thus:
twisted.internet.error.CannotListenError: Couldn't listen on 192.168.0.101:8332: [Errno 99] Cannot assign requested address.


Halp.

ETA: also noticed that the blade seems to restart about every 2:20. Good or bad?

That's the way I have it... both the blade and the fan controller are attached to the PSU (Molex). I don't know why I think I'll have to change the fan controller. I'm using this one: http://www.aerocool.us/peripheral/x1000/Strike-X%20X1000-manual-out.jpg I bought it cos it was the fastest to ship to my city.

Now is beeping randomly... so annoying...  

dont attach it to the blade, attach the cooler to the PSU, and u wont have any problems

I really don't know if this will affect, since we're not using getwork protocol right?. I extracted it from: https://github.com/slush0/stratum-mining-proxy

Compiling midstate C extension

For some really big operations using getwork interface of this proxy, you'll find useful "midstatec" C extension, which significantly speeds up midstate calculations (yes, plain python implementation is so slow). For enabling this extension, just type "make" in midstatec directory. Proxy will auto-detect compiled extension on next startup.


So I wake up today to find that the hash rate of my blade has slowed down to ~5000 Mhash/s. Any idea? 

Edit: Ok, I think it is because of the fan controller. It shuts down the setup when it detects high temperature. Fuck, I should have bought a simpler one...