Topic: [ANN] jminer v0.5.3 - GPU assisted PoC-Miner for Burstcoin (BURST) - PoC2 - page 3. (Read 23320 times)

If it's 64-bits it will say so, otherwise it's 32-bits.
source: http://stackoverflow.com/questions/4574090/installed-jvm-is-64-bit-or-32-bit

mine:
java version "1.8.0_60"
Java(TM) SE Runtime Environment (build 1.8.0_60-b27)
Java HotSpot(TM) 64-Bit Server VM (build 25.60-b23, mixed mode)

You can try on the command line:
If it's not a 64-bit version, you'll get a message that looks like:
If not maybe add -d64 as argument to the *.bat, to force 64bit. (if 32bit and 64bit are supported by your java vm, and jminer why ever takes the 32bit one)
source: http://stackoverflow.com/questions/2062020/how-can-i-tell-if-im-running-in-64-bit-jvm-or-32-bit-jvm-from-within-a-program

EDIT:
Thats because 1G means the Java VM can only use 1GB of memory, thats to small ... we should focus on running with 64-bit and 6GB :-)
Just to explain, a 32-bit is limited in addressing memory, you noticed that, guess even -Xmx2G did not work.

C:\>cd Burst

C:\Burst>java -version
java version "1.8.0_66"
Java(TM) SE Runtime Environment (build 1.8.0_66-b18)
Java HotSpot(TM) Client VM (build 25.66-b18, mixed mode, sharing)

C:\Burst>



Never installed 32bit version. Clean fresh 64bit install only.
Again problems, jminer with option  java -jar -Xmx1G -XX:+UseG1GC burstcoin-jminer-0.3.6-RELEASE.jar
it just freezes at 100% plot reading done, not telling how long it takes to read, and not getting next block info.

Well, thats strange, please make sure you use 64bit Java ... test it in command line with:

To download Java 64bit JRE (uninstall the 32bit java before installing the 64bit one)
http://www.oracle.com/technetwork/java/javase/downloads/jre8-downloads-2133155.html


Btw. if that was the problem, you will not need that -Xmx6G anymore i guess ...

Memory usage is below 1,5Gb according to task manager.
Going to try right now your suggestions. Thanks.


c:\Burst>"run 0.3.6-RELEASE.bat"

c:\Burst>java -jar -Xmx6G -XX:+UseG1GC burstcoin-jminer-0.3.6-RELEASE.jar
Invalid maximum heap size: -Xmx6G
The specified size exceeds the maximum representable size.
Error: Could not create the Java Virtual Machine.
Error: A fatal exception has occurred. Program will exit.

c:\Burst>"run 0.3.6-RELEASE.bat"

c:\Burst>java -jar -Xmx4G -XX:+UseG1GC burstcoin-jminer-0.3.6-RELEASE.jar
Invalid maximum heap size: -Xmx4G
The specified size exceeds the maximum representable size.
Error: Could not create the Java Virtual Machine.
Error: A fatal exception has occurred. Program will exit.

In device manager i 6.95Gb of ram available (of 8Gb maximum).


And chunkPartNonces=160000 is already from start, I changed it.

Will start with parameter  java -jar -Xmx1G -XX:+UseG1GC burstcoin-jminer-0.3.6-RELEASE.jar

Java VM is running out of memory (sometimes this happens even if not all system memory is used) ...
you should monitor, if really all available memory of your system is used ...
You can force Java VM to use more memory if not. Your machines have 8GB momory, right?

On memory issues:
1. allow java to allocate more memory e.g.: -Xmx6G will allow java to use up to 6GB of memory, for example:
2. reduce 'chunkPartNonces' in jminer.properties (will only help, if GPU is fast enough, will increase CPU usage)
Maybe try 160000 instead of default 320000.


If there is a fast block ... so the previous round is not finished, the memory usage may have a little peak.
Maybe you didn't have a fast block while testing ... but I'm quite sure that -Xmx6G will fix your problem.

Thank you for post luxe,

I have a problem with win10 and jminer.
Installed everything from scratch and got this error on all machines:

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "Timer-1"
Exception in thread "readerPool-52" java.lang.OutOfMemoryError: Java heap space
Exception in thread "readerPool-51" java.lang.reflect.UndeclaredThrowableException: Failed to invoke event listener method
HandlerMethod details:
Bean [burstcoin.jminer.core.checker.Checker]
Method [public void burstcoin.jminer.core.checker.Checker.handleMessage(burstcoin.jminer.core.reader.event.ReaderLoadedPartEvent)]
Resolved arguments:

• [type=burstcoin.jminer.core.reader.event.ReaderLoadedPartEvent] [value=burstcoin.jminer.core.reader.event.ReaderLoadedPartEvent@c6b981]

        at org.springframework.context.event.ApplicationListenerMethodAdapter.doInvoke(ApplicationListenerMethodAdapter.java:240)
        at org.springframework.context.event.ApplicationListenerMethodAdapter.processEvent(ApplicationListenerMethodAdapter.java:113)
        at org.springframework.context.event.ApplicationListenerMethodAdapter.onApplicatio nEvent(ApplicationListenerMethodAdapter.java:103)
        at org.springframework.context.event.SimpleApplicationEventMulticaster.invokeListe ner(SimpleApplicationEventMulticaster.java:163)
        at org.springframework.context.event.SimpleApplicationEventMulticaster.multicastEv ent(SimpleApplicationEventMulticaster.java:136)
        at org.springframework.context.support.AbstractApplicationContext.publishEvent(AbstractApplicationContext.java:380)
        at org.springframework.context.support.AbstractApplicationContext.publishEvent(AbstractApplicationContext.java:347)
        at burstcoin.jminer.core.reader.task.ReaderLoadDriveTask.load(ReaderLoadDriveTask.java:91)
        at burstcoin.jminer.core.reader.task.ReaderLoadDriveTask.run(ReaderLoadDriveTask.java:59)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
        at java.lang.Thread.run(Unknown Source)
Caused by: java.lang.OutOfMemoryError: Java heap space


Win 10 Pro 64bit, java latest. Getting same error on every machine.
This occurs really in matter of minutes, sometime after hour or so.


Regarding speed, this is a situation:
Plots of 26Tb are read in less than a minute, 50-60 seconds.


2016-01-13 05:13:29.007  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '550384831' > '60000' skipped
2016-01-13 05:13:29.022  INFO 4856 --- [  readerPool-26] burstcoin.jminer.CommandLineRunner       : 1% done (0TB 2GB), eff.read '8 MB/s'
2016-01-13 05:13:29.022  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '300150332' > '60000' skipped
2016-01-13 05:13:29.044  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '75916006' > '60000' skipped
2016-01-13 05:13:29.091  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '39550334' > '60000' skipped
2016-01-13 05:13:29.107  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '29188557' > '60000' skipped
2016-01-13 05:13:29.528  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '21378323' > '60000' skipped
2016-01-13 05:13:29.634  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '14026369' > '60000' skipped
2016-01-13 05:13:29.796  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '8353362' > '60000' skipped
2016-01-13 05:13:30.095  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '2465560' > '60000' skipped
2016-01-13 05:13:30.387  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '2067519' > '60000' skipped
2016-01-13 05:13:32.017  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '1476051' > '60000' skipped
2016-01-13 05:13:32.095  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '108588' > '60000' skipped
2016-01-13 05:13:34.569  INFO 4856 --- [  readerPool-25] burstcoin.jminer.CommandLineRunner       : 12% done (2TB 846GB), eff.read '128 MB/s'
2016-01-13 05:13:40.236  INFO 4856 --- [  readerPool-24] burstcoin.jminer.CommandLineRunner       : 23% done (5TB 668GB), eff.read '128 MB/s'
2016-01-13 05:13:45.591  INFO 4856 --- [  readerPool-26] burstcoin.jminer.CommandLineRunner       : 34% done (8TB 490GB), eff.read '130 MB/s'
2016-01-13 05:13:49.630  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : dl '43168' send (solo)
2016-01-13 05:13:49.894  INFO 4856 --- [  networkPool-1] burstcoin.jminer.CommandLineRunner       : dl '43168' confirmed!  [ 0d 11h 59m 28s ]
2016-01-13 05:13:50.980  INFO 4856 --- [  readerPool-31] burstcoin.jminer.CommandLineRunner       : 45% done (11TB 310GB), eff.read '131 MB/s'
2016-01-13 05:13:56.462  INFO 4856 --- [  readerPool-26] burstcoin.jminer.CommandLineRunner       : 56% done (14TB 132GB), eff.read '131 MB/s'
2016-01-13 05:14:03.833  INFO 4856 --- [  readerPool-29] burstcoin.jminer.CommandLineRunner       : 67% done (16TB 978GB), eff.read '124 MB/s'
2016-01-13 05:14:09.519  INFO 4856 --- [  readerPool-29] burstcoin.jminer.CommandLineRunner       : 78% done (19TB 798GB), eff.read '124 MB/s'
2016-01-13 05:14:15.283  INFO 4856 --- [  readerPool-29] burstcoin.jminer.CommandLineRunner       : 89% done (22TB 620GB), eff.read '124 MB/s'
2016-01-13 05:14:25.380  INFO 4856 --- [  readerPool-24] burstcoin.jminer.CommandLineRunner       : 100% done (25TB 440GB), eff.read '115 MB/s'
2016-01-13 05:14:25.395  INFO 4856 --- [    roundPool-1] burstcoin.jminer.CommandLineRunner       : FINISH block '185941', best deadline '43168', round time '56s 420ms'


Considering very low-end GPU integrated in APU it is ok for me, GPU usage is on average of 60% (I thought that GPU will be bottleneck - but it is not).
Here are screenshots of GPU usage while mining:

     



I dont know what could cause problem, while I tested it had no issues... never expected this situation.

I am very pleased to hear that ...
You say you need 52sec. thats still a little slow.
Data of a 4TB drive should be read in below 20sec. (my experience, but can depend on sata controller etc.)
If you GPU is not 100% while checking deadlines, the number of drives should not have influence on the 52sec round time.
So basically, if you add another 6x4TB under same conditions, your read speed would be 240mb/s and round would still be finished in 52sec.

Anyway, 120mb/s divided by number of drives is 20mb/s per drive (not cool) ...
If you drives read speed is not limited by hardware, i suggest:
- Use every drive as single drive (no raid or extended partitions over multiple drives or something)
- Use a high staggersize for your plotfiles to reduce seak operations of your harddrive, https://bchain.info/BURST/tools/overlap will provide info about that! Optimize your plotfiles if your staggersize is 8096 or something!

Plot Optimizer v1.6:
https://bitcointalksearch.org/topic/m.9440236
Plot Optimizer v1.6 with GUI for Win:
http://forum.burst-team.us:4567/topic/26/plot-optimizer-v1-6-with-gui-1-0-3-j6jq-win

I remember that discussion, my point is still, that i will not implement a Win7 specail case in a java application ...
I hoped for a way, to edit this FILE_FLAG_NO_BUFFERING  in the OS itself, in registry or something ... or via parameter for java vm ...
 

Hello,

Today I upgraded win7 to win10, results are good.
JMINER works well, using only few Gb of RAM.
No stress on CPU. I have a feeling that even 4Gb of ram would be ok.

I plotted 6x4Tb drives, read time with GPU and jminer is 52 seconds.
GPU load is high 99%.
So far, so good. Before read time was 1m53s. So, very good result.

Considering adding more HDD drives, jminer is a must have tool.
Great job luxe.

GPU load is really low, from 30%-70%, even lower on hdd read, speed is 120mb/s shown in jminer.
Thinking like there can be better result.

I'm with 16 GB RAM. Blago's miner had the same problem with win7 and then he fixed it, so it's doable.

If you know a way to switch off caching for Win7, feel free to tell us.
In your case of 25TB ... what about adding more memory to your PC?
16-32GB should be able to handle the cache needs?! (not sure if the benefit will be big enough to legitimate such a investion)

I'm impressed, guess you could save a lot of power, if you manage it to run jminer and attach more drives to less PCs!

Well, i have this case http://geizhals.de/lian-li-pc-d8000b-schwarz-a834927.html (20 drives)
and i added http://geizhals.de/raidsonic-icy-box-ib-545ssk-54105-a817979.html (5 drives) so 25x4TB ...
my motherboard allows 14xSATA and i added two of this controllers: http://geizhals.de/highpoint-rocketraid-2680-a380035.html
I also have some additional none-burst drives attached via USB3+USB3 Hubs ...
You could also use external cases like http://geizhals.de/fantec-qb-x8us3-1461-a669223.html ...
There are many ways to attach a lot of drives to one PC ...

Cooling is no big issue, GPU only runs a few seconds every 4min. ... I just used the case internal 'slots' for air cooling.

But back to you issue ... i asked about you configuration for a reason ... if you have max. 32TB per PC, you should be fine with another OS and one of your mentined GPUs above. Maybe even your embedded GPU will do the job.

Add external usb3 HDDs. You also need some usb3 ports + hubs. You could use your internal HDDs at the same time. I need the win7 installation on my pc, so I'm mining 25 TB on CPU. Jminer won't work on win7 (it'll work, but you can't use the PC for anything else).

Edit: Mining on HDD doesn't need the caching at all, so if this is switched off on all OSs jminer may consume even less power, no?

I have 9 pcs with 8x4Tb each, plus 2 pcs with 4x4Tb each.





Will you please tell me how did you managed 100Tb in single comp, and some more info on cooling system.

Just use the most powerful GPU ... to ensure there is no bottleneck ...

Ubuntu should be fine, sry no idea what version is best.
You can also use Win8 or Win10. It is only Win7 with that caching issue.
(I use Win8 and a 280x, my GPU is used 50% while mining 100TB (4TB Drives) @ 15sec per round with a memory usage of 2-4GB)


Nice to hear that, we will booost your 300TB :-)

Are this 300TB all attached to same PC? Or do you have multiple PCs running?

Hello luxe,

I will add additional GPU to system. I have several cards here laying around. Please advice best for me:

ATI 7790 1Gb ram (bonaire chip)
ATI 5850 1Gb ram
ATI 5870 1Gb ram

Also will change system to linux distribution, is Ubuntu OK and do you suggest some version of it?

Thanks for help.
BTW I support burst, and like it a lot. I'm a believer. Mining with 300Tb.

Hi,
You use Win7! There are known caching issues on mining BURST with Win7, that can solved by using another OS! (I also had to use another OS, as i started mining BURST)

About memory usage (beyond Win7 caching issues ...)
The part of the miner that reads the data/plot-files will just read as fast as possible all data into memory.
If your GPU is not fast enough to handle that data, the memory usage will grow (max. 0,025% of your plotfiles + base memory for application ...)

Additional info: https://github.com/de-luxe/burstcoin-jminer/wiki/burstforum.com-thread-recovery (search for 'memory usage')

I have inspected this situation, and noticed that jminer runs well until memory is filled, after no physical memory left it just stall whole thing.

Anyone having problem with running it on 980 ti? It keeps crashing on me with that GPU, but is stable on, eg, 750 ti and 970.

Hello luxe

I have an issue running jminer on APU AMD A4-7300.
Java is version 8, win7 64bit, 8 gigs of RAM, latest drivers.

Jminer runs and freeze a computer, after 10-20 seconds, everything looking normal until it blocks it totally.

Last 3 messages in jminer are WARN 4000 unable to get mining info from wallet: null