Still have memory issues wit Jminer, Help Chuck



  • I'm still getting memory errors with Jminer. Saying i ran out of resources.

    I get this message on a lot of drives and then it kills the miner and stops mining.

    IOException in: C:\MountPoints\2-28\burst\plots\12753280106257581144_2800000000_30400000_10000 -> Insufficient system resources exist to complete the requested service

    I have 32gb of ram now. Adjusted chuck nounces.
    Even running O&O clever cache. Which keeps the ram usuage under control but still errors out within a few hours



  • @ChuckNorris Do you use win7? When I ran this, I always had memory leaks with java. Since I am on win8.1 I have no problems anymore.



  • @Daforce Windows 10 home 64 bit



  • @ChuckNorris I found out recently that jminer uses all of my friend's 16GB of RAM on his un-optimized plots. My optimized ones only use about 4GB on top of the 4GB used by the system/browser/wallet.

    Did you say you upgraded to i7s? Try the latest version of blago's miner and run the .exe for AVX2. I found the scan times to be impressive on my friend's i5 system. You might not need the GPUs at all anymore.

    Notes for blago's miner solo:

    Add passphrases.txt to your blago miner install with just your passphrase in it.
    https://forums.burst-team.us/topic/412/solo-mining/4



  • @sevencardz What's his read times.
    So far i'm reading 250tb in 80 seconds with unoptimized plots. (Yes i'll be optimizing them)



  • @ChuckNorris Stats: 84TB UNoptimized plots, i5 6600K, 16 GB RAM, GTX 970

    jminer 0.4.10 -> 60% CPU usage, 100% RAM usage, minimal GPU usage, 32 second scan time
    blago w/ AVX2 -> 100% CPU usage, minimal RAM usage, zero GPU usage, 34 second scan time



  • @sevencardz Hmm. i'll give it a try and report back. So blago must be a parallel scanning miner too than.



  • @ChuckNorris @blago 's miner will use the threads of your CPU to read drives concurrently, but that's typically slower than using jminer with a GPU. However, this newest version utilizes the AVX2 instruction set and that changes everything. 🙂 My friend can now dedicate the 970 to mining another currency and we don't have to ask the system nicely for some extra RAM just to run a browser.



  • @ChuckNorris CRAZY!. And THANK YOU. I actually get better speeds. lmao. 1100mb/s instead of 900mb/s. I'm sure that will solve my memory problems as well. Let me tip you. What's your burst address. I'll send you 1000 burst. My read speed went from 100 seconds to 64 seconds. I didn't think blago would be that fast. Now i need to setup for solo mining


  • admin

    @ChuckNorris

    On jminer memory issues, you have multiple options ...

    1. The miner will read data into memory as fast as possible, not waiting for GPU to compute the data ... so if you have slow GPU it could cause memory issues.

    2. You can adjust readerThreads ... the default setting is 'readerThreads=0' what means, the miner uses one thread for every of your drives / plotFilePaths. Try using 'amountOfDrives/2' for reader threads ... will cause longer round time but use less memory.

    3. On optimized plotfiles you can reduce chunkPartNonces, to use less memory.
      https://github.com/de-luxe/burstcoin-jminer#miner-memory-usage

    4. Check with task manager if java uses all available memory. If still resources left, you could manually give java more memory e.g. 26 GB with -Xmx26g ...

    java -jar -d64 -Xmx26g -XX:+UseG1GC burstcoin-jminer-0.4.10-SNAPSHOT.jar
    


  • @luxe Thanks Luxe, Your the man around here. Forgot about java command for more memory. I think i'm going to stick with Blago for right now. I get better speeds with it. I'm also using gtx 1070 per machine for jminer. so def shouldn't be to weak. Tried the render thread option but reduces my speed too much.



  • @ChuckNorris - Wow, so you're saying an i7 with blago's miner using AVX2 will currently beat a GPU with jminer on large buffered plots... I will have to see how well my Ryzen 1700 does on optimized plots with AVX2. Oh, and no tip necessary. Send some to blago if you're feeling generous. 🙂 Thanks to him, I can now use my GPUs for other things.

    @luxe - Any way to incorporate AVX2 instructions into jminer? It'd be great if jminer could use the iGPU and the AVX2 part of the CPU in tandem somehow, but of course I don't really know what I'm talking about. 🙂



  • @sevencardz Yeah. Totally man. I Abandoned jminer as i was getting out of resource errors with 32gb or ram. Blago is just faster anyways. Looks a little better too. I'm able to scan 250tb in 60 seconds with OPTIMIZED plots. I'm sure that will be reduced to 40 seconds when i'm done optimizing. I don't think it's worth too much effort to improve it much farther. I get an average read of 1000mb/s with 7 drives per hub. Not too bad if you ask me.



  • @ChuckNorris I feel jminer is better, as my cpu is shit though, AVX2 carries Blago out of the water, and you don't need to use your GPU anymore, setting my chunkPartNonces at 960000, is my sweet spot