Problems with CGMiner 2.4.1 and Singles?

[xDGDZEx]

Is anyone else having this problem? Just downloaded version 2.4.1, and saw this



Couldn't find any reference to this problem in the cgminer thread, so I thought I would ask here.

[Scared]

You need a version of CGMINER compiled with --enable-bitforce

[rjk]

You need a version of CGMINER compiled with --enable-bitforce

All the official binaries are.

[Scared]

You need a version of CGMINER compiled with --enable-bitforce

[EDIT]

You can run CGMINER as is but you can only use COM1-COM9 and you'll still get the Test Failed messages but it'll run

[Scared]

You need a version of CGMINER compiled with --enable-bitforce

All the official binaries are.

I wish it was but it's not. I'm in the process of compiling them again with --enable-bitforce.

Is it possible that the windows version only supports COM1-COM10 then?

[rjk]

You need a version of CGMINER compiled with --enable-bitforce

All the official binaries are.

I wish it was but it's not. I'm in the process of compiling them again with --enable-bitforce.

Is it possible that the windows version only supports COM1-COM10 then?

For high COM numbers, you need to use "\\.\COM25" for example. Anything over 9 requires \\.\COMxx

[xDGDZEx]

But as you all can see, those things were done. I think it was epoch that told me how to use the ports above 9 originally...

[-ck]

You need a version of CGMINER compiled with --enable-bitforce

All the official binaries are.

I wish it was but it's not. I'm in the process of compiling them again with --enable-bitforce.

Is it possible that the windows version only supports COM1-COM10 then?

The official binary most definitely is compiled with bitforce support.

[xDGDZEx]

And as an aside, 2.3.6 works perfectly.

[kano]

The order of detection has changed since Bitforce detection messed up Icarus detection
Looks like on your setup Icarus detection is also messing up Bitforce detection.

You can force the detection you want by:
 -S bitforce:\\.\COM10
or
 -S icarus:\\.\COM10

[xDGDZEx]

With those changes, now this

[kano]

Then the bitforce code isn't detecting your singles.
Run it with -D --verbose -T and pastebin the output and post a link (or come to IRC Freenode/#cgminer and ask there since this will take forever on the forum)

[fuuka]

Does "-S auto" not work on windows?

[kano]

Does "-S auto" not work on windows?

auto doesn't work on windows (as the README says)

[xDGDZEx]

So...update.

I spent a couple hours on it last night. I could not get any meaningful information other than it saying that it failed to detect bitforce, even when I manually specified ports (ie -S bitforce:\\.\COMx)

Went back to 2.3.6 and am having no problems.

And still couldn't find any logs generated. Oh well, works great now!

[Epoch]

So...update.

I spent a couple hours on it last night. I could not get any meaningful information other than it saying that it failed to detect bitforce, even when I manually specified ports (ie -S bitforce:\\.\COMx)

Went back to 2.3.6 and am having no problems.

Yes, I've stayed with 2.3.6. Its bitforce support/compatibility is solid and I've been running it on multiple hosts with multiple Singles for weeks at a time (Win7/64) without any issues.

IMHO the 2.4.x branch has been getting too many issue reports for me to bother with it; improvements in newer versions are welcome, but it is frustrating/disappointing when things which used to work are broken in the process. 2.3.6 is solid. As the saying goes, a known devil is often better than an unknown angel. And, in the case of 2.3.6, the 'devil' is more than satisfactory.

[xDGDZEx]

It is a bit disappointing, and odd because one of the main changes to the 2.4 branch was better icarus detection.

So maybe it's a bit ironic that it detects icarus better, but BFL not at all anymore 

[Cablez]

That's strange I haven't had any issues with 2.4.1 recognizing my BFLs or running funny.  weird.

[kano]

It is a bit disappointing, and odd because one of the main changes to the 2.4 branch was better icarus detection.

So maybe it's a bit ironic that it detects icarus better, but BFL not at all anymore 

The change was simply to detect Icarus before BFL rather than BFL before Icarus.
(since it was reported that BFL detection was messing up Icarus and that was the ... ... that someone else supplied ...)

Anyway, with the names added to the front, there should be no difference at all between the detection in 2.3.6 and 2.4.1 since the names (another ... ... that someone else did quite a while back) would force the devices to never even be looked at by the other driver
i.e. with a name like bitforce:\\.\COM10 - the Icarus detection would never touch the \\.\COM10 port before the Bitforce detection gets a look at it.
Hmm - maybe I should actually force that rather than let the code work the way it was implemented.

i.e. there is likely be some other problem ... and ckolivas suggestion to try this with a single BFL on your rig is probably the best chance at sorting it out
Remember you can run your rig with 2.3.6 using all but one BFL and then run testing on the last BFL with 2.4.1 at the same time (and switch that one between 2.3.6 and 2.4.1 as needed)

Anyway, if you get a chance come back and visit #cgminer again and if I'm around and have time then I'll see if we can resolve it with a single BFL

(P.S. the translation of '... ...' is 'hack' )

[kano]

... and I've put up a pull request change earlier today to get rid of that 2nd hack I mentioned above.
There are 3 source code file changes to tidy that up (and help with the windows problem)
See the pull request here:
https://github.com/ckolivas/cgminer/pull/215

The problem seems to be when you have more than 6 BFL's on xDGDZEx's version of Windows7
(it will only handle 6)
This change should fix that.