Topic: (OLD) BFGMiner: modular FPGA/GPU, GBT, Stratum, RPC, Avalon/Lnx/OpnWrt/PPA/W64 - page 2. (Read 260061 times)

Wee!

So,is that a BFL unit

Are you still at BFL helping to sort out the miner software

Thanks Luke  

Nice to see the SC hashing with BFGMiner, Luke.

Nice try removing the rest of kano's troll.
The pool balancing stuff was written by Con Kolivas, including all the bugs which CrazyBlane is concerned about.


I proposed a truce to Kano a month or two ago, that if he stopped trolling and FUD all over BFGMiner, I'd just ignore cgminer. He flatly refused.

For not running as superuser, see Here. Worked for me.

I primarily use cgminer but I switch back and forth from time to time. I switched to bfgminer when I found out you removed BFL serial-usb support, which broke my instance based load balancing script. Not that I see any problem with libusb, I just needed to change the syntax of my start-up command. This led me to my next issue, cgminer cannot access my bfl devices unless I run as super user. I'll spend some time figuring it out later as I think it's important to have both forks of cpuminer ready to go in the event that one doesn't work properly with BFL ASICs.

Load-balancing options in both forks are uneven and leave much to be desired. I brought it up in September and November in the cgminer thread but I seem to be the only person concerned about even load distribution across pools with ASIC devices. This is what led me to write my script to provide device based load balancing, and I'm just suggesting this would be a good thing to have in the application. I'm not upset, these are open source projects and I appreciate the work you guys have put into them with little to no monetary compensation. You've both advanced mining tremendously over the python based miners I used in the beginning.

On that note, I think it's detrimental to the community that you guys are arguing in each others release threads.  Personally, I'd like to see you guys working on the same branch. If that's not going to happen, why don't you just leave each other alone?

Were the pool balancing options written by you, or by the CGMiner team? If the latter, then he has a point, and he's not trolling.

Ignore the troll

You are asking in the wrong thread.
The pool balancing options were written in cgminer - they were just copied to the clone here.
If they aren't working correctly in the clone, then consider that yet another reason to not use a clone, but instead use the original cgminer.

Probably not too difficult, but it'll be a lot easier when I get to rewriting the work-fetching interfaces to be cleaner.

Hmm, you may be right about older pools suffering more from ASICs. Already Deepbit is having trouble keeping up with some faster FPGA farms.
I would recommend picking a couple of pools known to manage with ASIC-like loads to use - pretty much anything with vardiff and (GBT or stratum) should do the job.
Personally, I prefer and recommend Eligius. Besides having originally started the pool, I also maintain the software that keeps it going, and have seen it tested with plenty of hashrate.

How difficult would it be to implement the ability to tie a pool to a specific mining device?(Using backup pools as failiver only for each device)My current method is to run multiple instances with each device on its own instance. It appears I'll be receiving more devices from BFL than I had originally planned, and I'm not sure I want to be running that many instances at once.

My concern is the unreliable load distribution using the various load balancing switches. An unproportionate amount of hashing power given to a pool that's having issues or underperforming could result in a huge loss in mining profit for the day. With ASICs rolling out in mass quantities soon, I predict pool issues are going to climb dramatically.  

Spawning another process in app would prob be the easiest way to go about it, but I'm not sure that would be any better than running multiple instances. I'm really looking for an even distribution of hashrate across pools, and I'm thinking the best way to do that with multiple devices is tie each device to its own pool.

Anyways, am I the only person concerned about even hashrate distribution and pool issues from ASIC rollout?

Xiangfu wrote the cgminer code changes in the Avalon and he's rewriting it to bring it up to speed with the current improved cgminer code ckolivas has written since then and the much better driver USB code that I've written to replace the rubbish that is still in the clone.
Go discuss with Xiangfu if you want to get somewhere useful with it ...

Joke fail ... there is no port of that rubbish.
Lulz, no one in their right mind would port that piece of rubbish to anything.
There's a reason we don't take code back to cgminer from him, but he attempts to copy copious amounts of the cgminer code to the clone ...
The last biggest section of code from him that went into cgminer was the MMQ driver that didn't even work.
I've rewritten it completely in every aspect of how it works and it of course works better and hashes faster than the MMQ code in the clone ...

Interesting ... to use your words: Luke's best effort attempt ...
... is to brick the Avalon

You all are ruthless...

Chill ,the avalons a prototype, and so is anything it's running. It could be as simple and one fat finger somewhere in the code or build config.

I'm just glad I didn't have to get out a EEPROM burner or hot swap flash chips or something weird.

Luke's the only one actually trying to help getting it doing things I'd like to do.

Everyone else just bitches about how they got left out.

No one is willing to test or brick their units. I don't mind. I'm in it for more than mega BTC. In trying to tinker and make it better. I'd really like to get it working well on p2pool.

Yeah, it's impossible to get access to a WR703N and see if the image boots properly.

Luke makes a best effort attempt to support third party hardware which he has no access to, posts the code with appropriate warnings and initial recovery instructions.

How Luke's unrelated efforts which are in cooperation with another hardware vendor are related in anyone's guess.

Good thing Kano's port of BFGMiner to avalon runs perfectly...

No network at all. Nothing. Put a network tester on it. Nothing. No dhcp/bootp not activity at all. No response to any of the common ip addresses.

Recovery doesn't need a serial port.

To recover an Avalon.

Shut unit off. Setup a pc. Get a http server running on it. Put known good firmware in root directory of webpage. Setup PC on ip 192.168.1.5. Start a ping to 192.168.1.1 plug cable between pc and Avalon. Power up Avalon. Watch blue led on wrt. When it blinks after 10 seconds or so hold down reset button for 1-2 seconds. PC should be able to telnet and ping to 192.168.1.1

Telnet to device. It's in limp mode.
cd /tmp
wget http://192.168.1.5/firmware.bin
mtd -r write firmware.bin firmware
It will reboot Avalon when done.
Setup as clean Avalon. Avalon up will be back to 192.168.0.100

Connect to default webpage and restore backup settings or reconfigure.

WR703N ∈ Avalon