Author

Topic: OLD: BFGMiner 3.10.0: modular ASIC+FPGA, GBT+Strtm, RPC, Mac/Lnx/W64, AntU1, DRB - page 129. (Read 1193368 times)

hero member
Activity: 1246
Merit: 501
Avalons use the 703N?  Wow, that didn't occur to me. 
hero member
Activity: 826
Merit: 1000
°^°
working fine on a MR3020 with AA
Avalon's WR703N Firm is very limited
legendary
Activity: 2576
Merit: 1186
I would be very keen to use this if you did.  I'm planning on using the TL-WR703N, which I think has 4MB flash and 32MB RAM. 
4 MB flash is too small for even the normal BFGMiner packages (or really any packages).
The only way I expect you'll be able to make this work is to build a custom firmware with BFGMiner included in the main image.
Whether libmicrohttpd could be squeezed into this or not, I'm not sure of.
nah, its much easier: "opkg -d ram"  Wink
Did you get this to work?
I tried for Avalon's TP-Link, and found this feature is pretty broken. Sad
hero member
Activity: 826
Merit: 1000
°^°
I would be very keen to use this if you did.  I'm planning on using the TL-WR703N, which I think has 4MB flash and 32MB RAM. 
4 MB flash is too small for even the normal BFGMiner packages (or really any packages).
The only way I expect you'll be able to make this work is to build a custom firmware with BFGMiner included in the main image.
Whether libmicrohttpd could be squeezed into this or not, I'm not sure of.

nah, its much easier: "opkg -d ram"  Wink
hero member
Activity: 1246
Merit: 501
Never mind, did a apt-get upgrade and all is working. Smiley
hero member
Activity: 1246
Merit: 501
Just installed bfgminer on Ubuntu 13.04 Server.  I'm a linux n00b, so I think I've installed as per the README (I cloned off git, compiled, etc).

When I run bfgminer I get:

"bfgminer: error while loading shared libraries: libblmaker_jansson-0.1.so.0: cannot open shared object file: no such file or directory"

Err...help?  Sad
legendary
Activity: 2576
Merit: 1186
If Windows is reporting ERROR_ACCESS_DENIED maybe Windows is reporting that error because of a problem that is different than the one you anticipated.
It's not a matter of anticipation.
ERROR_ACCESS_DENIED is defined as a permissions issue.
If Windows is returning it for a non-permissions problem, Windows (or a driver instructing Windows to do it) is at fault.
member
Activity: 74
Merit: 10
On Windows it seems bfgminer doesn't know the difference between a problem with user access privileges and a device not responding/device sending back bad data due to a low power condition.

The error doesn't make sense on Windows since there's no user privileges restriction on non-storage USB devices.

I tested this by using a USB2.0 powered hub that doesn't provide enough power to operate Block Erupters and a USB3.0 hub that provides adequate power.

With the USB2.0 hub that did not provide enough power, when more than 2 block erupters were plugged in an error was reported by bfgminer on startup:

"Do not have user privileges required to open \\.\COMxx"

...where "xx" is a port number.

When I plugged the devices into a USB3.0 hub that provided adequate power for all the devices the error went away.

I was able to replicate this consistently by switching between the powered hubs.
This is an OS/driver issue. The message about user privs is shown when Windows reports ERROR_ACCESS_DENIED.

This is provably not a driver/OS issue.

If the same device is plugged into one hub it does not work and the error is presented. However, if it's unplugged and plugged into another hub it does work and the error is not presented.

If Windows is reporting ERROR_ACCESS_DENIED maybe Windows is reporting that error because of a problem that is different than the one you anticipated.

If you could anticipate every possible scenario there would never be bugs in your software and you'd be omnipotent.
legendary
Activity: 2576
Merit: 1186
I was wondering how stratum pool distributes work to miners.  Does it split extranonce1,2 ranges among miners?
On the miner side, if you have 10 devices attached to one PC, is bfgminer splitting work and distributes ranges to individual devices.
And finally do devices split work to individual chips and engines?
Extranonce1 is assigned by the pool, unique to each connection.
Extranonce2 is what the miner is free to do whatever they want with.
BFGMiner currently just increments extranonce2 to create unique block headers for the drivers.

Someday ASICs might get fast enough that they need to do the their own header production, in which case drivers will be able to get a stratum-like job for them. Some devices are already in development to work this way, but it's a risky thing to do because it can negatively impact Bitcoin scalability if they have unreasonable limits on how quickly they can produce headers internally.

Is it better to have many miners or few more powerful miners.  Say in case of mini rig, would it be better to run 24 miners each 60GH/s or one 1440 GH/s miner?
A single 1.4 Th/s miner (eg, 3 minirigs) would make more sense.


Hmm.  If Extranonce1 is assigned by the pool, having more miners would distribute the work better as each miner would get its own version of Extranonce1, no?
Not sure I understand the question. Distributing the work "better" is useless - all that matters is that nobody overlaps.

So miner can create new merkleroots/work jobs as it iterates through Extranonce2.  Each new/generated work would need 1 sec@4GH/s to do a full nonce scan?
Correct.
16TH to do a full Extranonce2 scan?  And 256 PH to have any block solved in 1 second, by hashing all Extranonce1,2 and nonce all at once (assuming the miner can iterate through Extranonce1)?  Is that how it works?
Extranonces do not have fixed sizes. Either could be any length (including 0, although I suspect no miner actually supports this for extranonce2).
legendary
Activity: 2576
Merit: 1186
On Windows it seems bfgminer doesn't know the difference between a problem with user access privileges and a device not responding/device sending back bad data due to a low power condition.

The error doesn't make sense on Windows since there's no user privileges restriction on non-storage USB devices.

I tested this by using a USB2.0 powered hub that doesn't provide enough power to operate Block Erupters and a USB3.0 hub that provides adequate power.

With the USB2.0 hub that did not provide enough power, when more than 2 block erupters were plugged in an error was reported by bfgminer on startup:

"Do not have user privileges required to open \\.\COMxx"

...where "xx" is a port number.

When I plugged the devices into a USB3.0 hub that provided adequate power for all the devices the error went away.

I was able to replicate this consistently by switching between the powered hubs.
This is an OS/driver issue. The message about user privs is shown when Windows reports ERROR_ACCESS_DENIED.
member
Activity: 74
Merit: 10
On Windows it seems bfgminer doesn't know the difference between a problem with user access privileges and a device not responding/device sending back bad data due to a low power condition.

The error doesn't make sense on Windows since there's no user privileges restriction on non-storage USB devices.

I tested this by using a USB2.0 powered hub that doesn't provide enough power to operate Block Erupters and a USB3.0 hub that provides adequate power.

With the USB2.0 hub that did not provide enough power, when more than 2 block erupters were plugged in an error was reported by bfgminer on startup:

"Do not have user privileges required to open \\.\COMxx"

...where "xx" is a port number.

When I plugged the devices into a USB3.0 hub that provided adequate power for all the devices the error went away.

I was able to replicate this consistently by switching between the powered hubs.
legendary
Activity: 2576
Merit: 1186
I did read ASIC readme but I can't figure out from it. If I install Linux and put BFGMiner on. Can I change pools for a blade as simple as for other devices? And if a pool has a different username is not a problem since it is not the blade username and password but BFGMiner right?
When blades are configured to use BFGMiner, they work pretty much like any other device.
So yes, you can change pools, etc just fine.
hero member
Activity: 826
Merit: 1000
I did read ASIC readme but I can't figure out from it. If I install Linux and put BFGMiner on. Can I change pools for a blade as simple as for other devices? And if a pool has a different username is not a problem since it is not the blade username and password but BFGMiner right?
legendary
Activity: 2576
Merit: 1186
I was wondering how stratum pool distributes work to miners.  Does it split extranonce1,2 ranges among miners?
On the miner side, if you have 10 devices attached to one PC, is bfgminer splitting work and distributes ranges to individual devices.
And finally do devices split work to individual chips and engines?
Extranonce1 is assigned by the pool, unique to each connection.
Extranonce2 is what the miner is free to do whatever they want with.
BFGMiner currently just increments extranonce2 to create unique block headers for the drivers.

Someday ASICs might get fast enough that they need to do the their own header production, in which case drivers will be able to get a stratum-like job for them. Some devices are already in development to work this way, but it's a risky thing to do because it can negatively impact Bitcoin scalability if they have unreasonable limits on how quickly they can produce headers internally.

Is it better to have many miners or few more powerful miners.  Say in case of mini rig, would it be better to run 24 miners each 60GH/s or one 1440 GH/s miner?
A single 1.4 Th/s miner (eg, 3 minirigs) would make more sense.
JLM
full member
Activity: 164
Merit: 100
Hi Luke!!!!

Could i mine SHA and scrypt AT SAME TIME?
SHA with Block Erupters and FPGA´s; Scrypt With GPU.
If possible?
What should do?

Thanks!!!

Run multiple instances....
I had that idea, but i don´t know how.
Could you give a hand?Huh?
legendary
Activity: 2576
Merit: 1186
Yeah, I think running the system off a 4GB USB stick is probably a better idea.  Cheesy  Assuming there's no limitation in storage space, is there any other issues you can foresee?
OpenWrt doesn't seem to handle external storage for applications too well Sad
hero member
Activity: 1246
Merit: 501

I would be very keen to use this if you did.  I'm planning on using the TL-WR703N, which I think has 4MB flash and 32MB RAM. 
4 MB flash is too small for even the normal BFGMiner packages (or really any packages).
The only way I expect you'll be able to make this work is to build a custom firmware with BFGMiner included in the main image.
Whether libmicrohttpd could be squeezed into this or not, I'm not sure of.

Yeah, I think running the system off a 4GB USB stick is probably a better idea.  Cheesy  Assuming there's no limitation in storage space, is there any other issues you can foresee?

I'm guessing most people running OpenWRT will be running it on bigger routers that have more flash.
legendary
Activity: 2576
Merit: 1186
OK I made a mistake with GH and you made a mistake with 355...
Right, sorry, missed that distinction somehow. Fixed my original post.
hero member
Activity: 826
Merit: 1000
After a few hours, those last 12 seconds becomes less and less relevant, and your average should begin to settle on 355 Mh/s.
Is this error or is USB ASICminer really doing 355GH with your software? I thought 336 is theoretical max...
It's really doing 335 Mh/s for me at least (5½ day average):
Code:
BES 0:       | 336.0/335.7/335.7Mh/s | A: 37464 R: 72+0(.19%) HW:202/.54%
 BES 1:       | 336.0/335.7/336.5Mh/s | A: 37564 R: 66+0(.18%) HW:191/.51%
 BEE 0:       | 327.5/335.8/333.1Mh/s | A: 37177 R: 60+0(.16%) HW:153/.41%
 BES 2:       | 336.0/335.7/330.5Mh/s | A: 36887 R: 71+0(.19%) HW:202/.54%
OK I made a mistake with GH and you made a mistake with 355...
legendary
Activity: 2576
Merit: 1186
Is this the same issue with the OpenWRT version?  I just picked up a little TP-Link pocket router to run my Block Erupters, but it'd be nice if it'd run the Blades as well.
No, OpenWrt has a native port of libmicrohttpd already, so it should be possible.
The problem there is that it's all or nothing: if I build with libmicrohttpd support, it won't be usable without it, and may increase the flash requirements.
I haven't taken the time to measure just how much yet... Maybe I can offer multiple alternative BFGMiner packages to get around this.
I would be very keen to use this if you did.  I'm planning on using the TL-WR703N, which I think has 4MB flash and 32MB RAM. 
4 MB flash is too small for even the normal BFGMiner packages (or really any packages).
The only way I expect you'll be able to make this work is to build a custom firmware with BFGMiner included in the main image.
Whether libmicrohttpd could be squeezed into this or not, I'm not sure of.
Jump to: