burstcoin-jminer v0.4.10 - GPU assisted PoC-Miner (All Platforms)
-
@eneloop 950MB/s is a great read speed, and also most likely your bottleneck. If a faster GPU isn't processing faster than the previous one, the limit is likely to be how fast you can read from disk.
-
in cases of disk read bottleneck i would recommend a pcie sata III add on pc card that in conjunction with high end graphics card would really slice nonce reading times like butter but not essential. USB3 drives have in my case read faster with an addon card for that too simply that it isn't integrated into the motherboard which isn't essential either :). Like Haitch said, your read time is on par with proper read speeds. My cpu is simular to yours @eneloop i recommend not to use its gpu cores. Components don't have to be brand new. I've had my nvidia GT730 low profile passive cooled GPU for several years and been mining with it for over 6 months it has kept the pace for mining as i use xplotter for plotting. There's the power profile management in control panel with CPU adjustments that the minimum can be set to 100% which is healthier on desktops that they have no battery. That could set your gpu cores to be less latent. BUT being integrated into cpu it might not do anything but cpu will maintain top frequency even when idle. jminer can be manually configured to device id of your gpu cores they can be listed with the gpu ploting tools command switches to determine correct gpu address so as not to use cpu. EDIT: AMD built in graphics require monitor plugged in to be accessed or CPU only access!
-
Hello everyone, I want to try out Jminer to see how much I might be able to increase my read times but have a couple of questions before i can try it.
Setup:
i7-4790k, ASRock H81 BTC Pro R2, W10, 8GB RAM, 6x rx470s (mining ETH + DCR), 6x Seagate 5TB HDDsIssue:
Since i am mining ETH + DCR 24/7 and this fully utilized (stresses) those GPUs, I cannot direct any Burst mining to any of those 6 GPUs.
However, I understand that Jminer can also use onboard graphics (GPU cores) available in some CPUs. Lee from I-Mine-Blocks youtube channel is doing this with his setup, but he is using an AMD CPU.Question:
Even though I have an Intel CPU (with onboard graphics), can I still do the same thing (use GPU cores in CPU) to effectively run Jminer? Or, will this only work with on AMD CPU?Also, if this is indeed possible, I have 2 questions about how to set it up and make it work:
1- for the deviceID in the setup file... I obviously cannot use devices 0-5 because those are all assigned to my 6 MSI 470s. I looked up the specs of my 4790k and found this under the graphics section:
Device ID: 0x412
Would this "0x412" be the value I enter in the deviceID field in setup then?
2- I have just installed Java8 (W10) update 131 so I'm good there. My question is about the openCL/sdk drivers. I googled this for Intl W10 and this is what I found:
https://software.intel.com/en-us/articles/opencl-drivers
Would this be the correct package to install or is there something better? Also, can I assume that installing this will not impact my Radeon 470 mining operation?
Thanks in advance for any advice!
-
@GabryRox If your internal GPU supports OpenCL you can use it. To get it's ID, down load gpuplotgenerator, and run it in setup mode - you can get it to list the devices.
However you can dual mine ETH/DCR and Burst on the same card, there's just some GPU contention during the process - setting the Burst miner to a higher priority may help, and even with the contention, a standalone GPU is like to be way more effective than an intel CPU based GPU.
-
@GabryRox to utilize built in graphics on AMD chip you must have monitor plugged into the motherboard and recognized by windows for it to function otherwise you will either get an error message or CPU will be in its place.
-
Thank both for the tips... so, based on what Haitch said about being better to use a stand-alone GPU, i attempted this on a 2nd PC with 4 HDDs hooked up to it. I thought this would be easier since this is my newly built main personal PC that I am not using for mining (other than Burt). I have 1 msi 470 on the mobo with my monitor plugged directly into it.
I also update this PC to Java8-131 but then attempted to install the AMD-APP-SDKInstaller for 64-bit Windows and it was not able to install. I rebooted and tried again but still didnt work. Some kind of "cant connect to http protocol" error or something of that nature. I will post the exact error later after i am done GPU-writing a plot but do i even need this package to run Jminer?
Also to note, both before and after attempting this install, I try to launch Jminer on this PC and it simply doesnt launch. I double click the JAR.exe and nothing happens. I try launching through the .bat and I see a brief command window appear but then it's automatically closed immediately. Same thing happens when I try to launch OpenCL option through AIO client... it asks me if I want to use 2nd openCL devices but then only appears briefly before closing whether i click yes or no.
Is this behavior likely due to me not being able to install that AMD SDK package? It's the 3rd item on this page:
http://developer.amd.com/tools-and-sdks/opencl-zone/amd-accelerated-parallel-processing-app-sdk/
Is this the correct thing to install? Do I need it even if I've already got Crimson 16.10 drivers installed for my 470s on both PCs? Not sure why this won't install on my personal PC but would like to double check that it's the correct thing before putting onto my mining rig.
-
@GabryRox I don't have the SDK installed, and I multi mine just fine
-
@GabryRox Verify that you have 64 bit Java installed, and add a "pause" command after the java launch command in the bat file - it'll keep the window open and show you the error.
-
@luxe hi there, quick questions. the numbers of computing units is the number of threads used for mining ?
Because comparing those numbers with a friend of mine, his intel integrated gpu has 20 'true' units vs 14 of my RX 460, it seems really weird that an integrated gpu is more efficient for this compared to a dedicated amd card.
-
I have a new problem for 2 days now. I have a lot of strange DL warnings. like pic below:
i found an old message from @luxe which had the same problem, but you only said it was from time to time every 2 or 3 days. I really changed nothing. I never had that Problem before.
-
@Ther0byte I'm quite sure the GPU is not your bottleneck ... but you did not provide any additional info on the compared systems ... so i can not tell.
You also mix up threads and GPU parallel computing, there is only one cpu thread providing data to gpu, there the data is computed parallel.
The miner also uses one thread (software) per drive, means if you have 20 drives there will be a threadpool with 20 threads in it. Switching from one to the other if one has to wait, to ensure all cores of your cpu are used as good as possible.
-
@Zaziki your strange dl result is -1 ... due to previous error ... the pool did not respond to your committed deadline in time, leading miner taking -1 as response from pool. Compared with own calculated deadline it does not match. What results in that debug logs ... the only thing you should worry about is, why the pool you use does not respond. Your miner and plots are fine. Maybe the pool accepted the committed result, but was just not able to respond in time, check the pool webpage.
Maybe pool is attacked or it has heavy load, you could increase connectionTimeout, but from my point of view pool should be able to respond within 20sec.Disable debug mode to no longer see that strange deadline messages.
-
@luxe i'm not assuming i'm bottlenecked :) that's why i didn't provide additional data beside the 2 "gpu"s
I was just curious if it's possible that an integrated gpu of a laptop has more computing units and hence is "better" for mining compared to a desktop grade rx 460.
Or if it's a comparison between oranges and apples and even if a number is smaller or bigger it doesn't really says anything about the reading speed or whatever.
-
@Ther0byte burstcoin is not gpu mining ... the gpu is just used to assist in checking data read from hard drives... if your hard drives are slow, no whatever gpu could speed that up.
-
@Ther0byte It comes down to terminology used by the different companies. Your RX460 has 36 Compute Units, but each of those compute units comprises of 64 shaders, which are the part doing the math, giving a total of 2,304 shaders, or Computing Threads. Intel uses different meanings to similar terminology, but the integrated GPU has way less computing threads than a discrete GPU.
-
@haitch Thanks! after adding that pause, it showed a 64 bit version error. Then i realized it seems the only way to download 64 bit Java is to do it through a 64 bit browser, so once i did that i was able to install and verify 64 bit.
From there I was successfully able to launch Jminer for the first time! It started by showing all of my burst plots, but then after that i keep getting the following error:
"unable to get mining info from wallet, caused by connection timeout, currently 12 seconds, try increasing it"
then it just keep trying to start but only shows "unable to get mining info from wallet"
I originally left the walletServer= at the default burst-coin.us address. After that error, i tried both erasing the address and hash-tagging out the row, but neither worked and resulted in the same error upon start-up.
Any idea what I need to do to fix this?
-
@GabryRox the wallet server needs to be: pool.burst-team.us and verify port is 8124.
-
@haitch thanks again for your help! I tried this and looked like it was going to work as it actually started reading my plots... but then that dreaded error popped up again.
Here are the details in my setup file:
plotPaths=G:/Burst/plots,I:/Burst/plots,N:/Burst/plots,O:/Burst/plots
scanPathsEveryRound=false
listPlotFiles=truepoolMining=true
numericAccountId=1199xxxxxxxxxxxxx615
poolServer=http://pool.burstcoin.biz:8124Winner
walletServer=https://pool.burst-team.us:8124
winnerRetriesOnAsync=4
winnerRetryIntervalInMs=500soloServer=http://localhost:8125
passPhrase=xxxxxxxxxxxxxx
targetDeadline=
triggerServer=
recommitDeadlines=platformId=0
deviceId=0refreshInterval=2000
connectionTimeout=12000
-
I tried to post a screenshot but it didnt work. here is the link to that jpeg on my flickr account:
https://www.flickr.com/photos/97476559@N08/35495393976/in/dateposted-public/
-
@GabryRox Not sure why you're getting the message - if you look just before it occurred you had a DL confirmed, so you're communicating with the pool normally.
