BTCMiner - Open Source Bitcoin Miner for ZTEX FPGA Boards, 215 MH/s on LX150

[gr0bi42]

(I'd like to do this on my Windows box for now)

Oh, sorry. No idea how it works on Windows. Maybe you can run a Linux with VirtualBox. The resulting jar-file of course will run on Windows.

[antirack]

Thanks a lot gr0bi42, that's very helpful. I can give it a try tomorrow morning on my Ubuntu box. I can then still give it a try on my Windows PC to see how make etc works there. I'll let you know how it goes.

[gr0bi42]

Anyone interested in this?

Code:

001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 208.00MHz to 212.00MHz
001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 212.00MHz to 216.00MHz
001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 216.00MHz to 208.00MHz

Here is the patch agains BTCMiner.java (+ 2 removed warnings due to garbage a character in the comment):

Code:

--- bitminer-src/BTCMiner.java	2012-04-17 22:55:56.000000000 +0200
+++ bitminer-new/BTCMiner.java	2012-06-10 09:08:51.429426206 +0200
@@ -126,7 +126,7 @@
 	start();
     }
 
-// ******* checkNew ***********************************************************
+// ******* checkNew ************************************************************
     synchronized public boolean checkNew ( byte[] data ) throws NumberFormatException {
 	if ( data.length < 36 )
 	    throw new NumberFormatException("Invalid length of data");
@@ -1510,10 +1510,13 @@
 		bestR = r;
 	    }
 	}
-	
 	if ( bestM != freqM ) {
+	    if ( freqM > 0 ) {
+		msg ( "Change frequency from " + String.format("%.2f",(freqM+1)*(freqM1)) + "MHz to " + String.format("%.2f",(bestM+1)*(freqM1)) + "MHz" );
+	    } else {
+		msg ( "Set frequency to " + String.format("%.2f",(bestM+1)*(freqM1)) + "MHz" );
+	    }
 	    freqM = bestM;
-	    msg ( "Set frequency to " + String.format("%.2f",(freqM+1)*(freqM1)) +"MHz" );
 	    setFreq( freqM );
 	}
 
@@ -1725,7 +1728,7 @@
 	return "bitfile=" + bitFileName + "   f_default=" + String.format("%.2f",freqM1 * (freqMDefault+1)) + "MHz  f_max=" + String.format("%.2f",freqM1 * (freqMaxM+1))+ "MHz  HpC="+hashesPerClock+"H";
     }
 
-// ******* resetCounters ******************************************************
+// ******* resetCounters *******************************************************
     public void resetCounters () {
 	while ( freqMDefault<freqM && errorWeight[freqMDefault+1]>100 )
 	    freqMDefault++;

Compile it by yourself or feel free to trust me and just download it from: http://www.file-upload.net/download-4432190/ZtexBTCMiner.jar.html

[antirack]

Anyone interested in this?

Code:

001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 208.00MHz to 212.00MHz
001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 212.00MHz to 216.00MHz
001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 216.00MHz to 208.00MHz

Good job gr0bi42! Are you using Ubuntu and what JDK are you using?

[antirack]

I have been using the test mode with "-m t" a couple of times. I noticed that only FPGA1 gets warm. Is this supposed to test all 4 FPGAs or only FPGA1?

[gr0bi42]

Anyone interested in this?

Code:

001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 208.00MHz to 212.00MHz
001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 212.00MHz to 216.00MHz
001-1: ztex_ufm1_15y1-04A346BE51-2: Change frequency from 216.00MHz to 208.00MHz

Good job gr0bi42! Are you using Ubuntu and what JDK are you using?

Yes, I'm using ubuntu. It's version 11.04 (Natty). Further: openjdk-6 and java-common (ubuntu 11.04 default packages via Synaptic Package Manager).

[ztex]

Is Ztex the only one developing the Java BTCMiner?

Could it be run from source code or does one need to package it into the .jar file?

Is there anyone here who could give me a quick rundown what's required to get this to work after I made some changes to the source code (mostly to the logging functions for now)? I can donate some BTC in return!  (I'd like to do this on my Windows box for now)

Compiling tutorials (including windows) can be found at http://wiki.ztex.de/doku.php?id=en:software:tutorial_compiling. The only thing that need to be considered additionally is the adjustment of variable ''ZTEXPREFIX'' in the Makefile of the BTCMiner source package. It has to point to the directory where the ZTEX SDK is installed, see gr0bi42 postings.

In which features you are interested. A new release which contains new logging functions and minor changes (typos and so) is scheduled for June.

[Nachtwind]

In which features you are interested. A new release which contains new logging functions and minor changes (typos and so) is scheduled for June.

one very very simple request: make it run at 300MH
No wait.. just that: Could you include a parameter (option) to enable it to blink when a share was found? I know this is utterly useless but i would love to see this *G*

[BR0KK]

Make fmax configurable to underclock the fpgas;)

[Turbor]

And I would like to see a "die hard" software forcing the limping FPGA to mine at 200 MHz

[ztex]

Make fmax configurable to underclock the fpgas;)

I wrote it several times before: If the frequency is limited of overheat detection mechanism does not work anymore. This feature is unsafe and will not be implemented.

[ztex]

one very very simple request: make it run at 300MH
No wait.. just that: Could you include a parameter (option) to enable it to blink when a share was found? I know this is utterly useless but i would love to see this *G*

Can be implemented easily for 1.15x boards :-),  but not for 1.15y board because these have not general purpose LED.

[Nachtwind]

one very very simple request: make it run at 300MH
No wait.. just that: Could you include a parameter (option) to enable it to blink when a share was found? I know this is utterly useless but i would love to see this *G*

Can be implemented easily for 1.15x boards :-),  but not for 1.15y board because these have not general purpose LED.

Well since i have a 15x i owuld be happy for it to get implemented
...but i suppose you didnt mean the 300mh idea?

[rupy]

In which features you are interested. A new release which contains new logging functions and minor changes (typos and so) is scheduled for June.

one very very simple request: make it run at 300MH
No wait.. just that: Could you include a parameter (option) to enable it to blink when a share was found? I know this is utterly useless but i would love to see this *G*

I like this too!

[gr0bi42]

Some other modification:

* new option -nolog to not write a logfile (I have only a super small and slow ssd)
* new option -ac <seconds> to auto reset error and performance counter after given seconds (I'm using a 1 hour interval)
* and the modified frequency change logging

Here's the link to the patch file:
http://www.file-upload.net/download-4438741/BTCMiner.patch.html

To apply the patch:

Code:

> cd bitminer-src
> patch -p1 < BTCMiner.patch

[vv01f]

just ideas .. nothing worth, but probably sth others would also like to have

*edit due to the explanation of r0bi42
correct piping of stdout/stderr Help-output for usage with pipes
I tried -h command line switch either on Windows 7 and Ubuntu 12.04.
Not surprisingly I got the help-output for commandline switches.
As I am lazy I like to print such things into a textfile for better access.. but that doesnt work the pipe-way I am used to.

command at mine was ending:
 -h >BTCMiner-help.txt

That file still is empty afterwards, but why?
- IMHO errors should be pointed to stderr instead of mixing them with 'normal' output-msg, so we could pipe just errors in some log
- e.g. help instead belongs to stdout in my mind
- help could be printed when the jar / BTCMiner is called with unexpected switches or none (just starting the jar right now results in error msg)
/edit

Error-handling
I'd like to have some possibility to actively handle errors during runtime automatically.
One could repeatingly scan the log-file for errors.. but it appears to me a simple reporting as (documented) error-numbers and exiting BTCMiner would (optionally?) be a far better solution than running nevertheless there are problems that could be dealt with.
Even better would be to communicate with another (repeating temporary) instance of BTCMiner and the running one, so that the cluster has not to stop anytime.


Changing -oh at runtime
As the runtime-commands are waiting for string-input and release by hitting [Enter]:
can you implement some runtime-command like oh=#.##
where users can (re)set overheat protection for testing without stopping the BTCMiner?


Changing pool (using -host and -b) at runtime
Same thing like the other runtime-option:
perhaps via: pool=2
could change to 1st -b and: pool=1
back to the one given by -host


Thanks for now..

[gr0bi42]

just ideas .. nothing worth, but probably sth others would also like to have

Help-output for usage with pipes
I tried -h command line switch either on Windows 7 and Ubuntu 12.04.
Not surprisingly I got the help-output for commandline switches.
As I am lazy I like to print such things into a textfile for better access.. but that doesnt work the pipe-way I am used to.

command at mine was ending:
 -h >BTCMiner-help.txt

That file still is empty afterwards, but why?

Because it goes to stderr and not to stdout.

Code:

> java -cp ZtexBTCMiner.jar BTCMiner -h 2> BTCMiner-help.txt

will do the job.

[vv01f]

Because it goes to stderr and not to stdout.

Thanks for clarification - but now it sounds even more like a bug to me

[gr0bi42]

Because it goes to stderr and not to stdout.

Thanks for clarification - but now it sounds even more like a bug to me

OK, if you request the usage, then it should not come out on stderr. But it's easy to fix:

Code:

@@ -1923,7 +1940,7 @@
 		    forceEP0Config = true;
 		} 
 		else if ( args[i].equals("-h") ) {
-		        System.err.println(ParameterException.helpMsg);
+		        System.out.println(ParameterException.helpMsg);
 	    	        System.exit(0);
 		}
 	        else if ( args[i].equals("-n") ) {

[vv01f]

K, so I am not failing with my understanding of pipes. 

I mean thanks to this forum and of course ztex and ppl like you it is all fine, we can read and ask .. also I could read the code myself - but for now I prefer just to document my silly ideas.

just edited my suggestions-post .. perhaps it inspires someone