Blago's Burst-Miner (Win)

BeholdMiNuggets

@sevencardz said in Blago's Burst-Miner (Win):

Depends on your hardware specs and which miner you're using. Also depends on how many external drives are hooked to a single USB3.0 controller.

All the HDDs are SATA-III /internal. May add a PCIe SATA controller for more capacity (later). No external (USB3) drives yet!

My FX-8350, 16 GB RAM, GTX 960 system scans 88TB (optimized) in 25 seconds with the latest jminer and 60 seconds with blago's (due to no AVX2 support on the AMD chip). With more drives connected, it slows down to 32 seconds due to lack of USB 3.0 controllers on that system.

Your Blago is faster, but need to run some more tests here. Is ~60 seconds too long for scans, do you miss many short cycles? I also wonder how those with hundreds of Tb handle this.

If you use jminer with more than 50TB or so, make sure you're using a gaming-level GPU (not iGPU or budget workstation GPU) and a CPU with at least four threads, otherwise your system will bottleneck the scan times. If you have unoptimized plots, make sure you have a lot of RAM to back up jminer. If you have a newer proc that supports AVX2, use blago's miner and use your GPU to mine on Nicehash.

It's a good question - Is it better to use your shinny new GPU to mine ETH, ZEC, ETC - or to speed-up Burst-mining?

vmantilla

So, y'all are saying that if I have an older CPU (i5 3450, 4 cores, 4 threads) and more than 50 TBs, and my CPU can use AVX2, I should use Blago's miner because it is faster than Jminer? I have a NVIDIA GTX 1060 6GB that I think is a gaming GPU. I'm currently optimizing all my plots (snooooore <smile>).

This is quite interesting.

Thanks for any info.

sevencardz

@vmantilla I'm afraid that i5 3450 is too old. Looks like AVX2 was introduced into Intel procs with the Haswell generation. Ryzen chips have AVX2 as well, but rumor on the internet is that it's not as good as Intel's implementation.

BeholdMiNuggets

@sevencardz said in Blago's Burst-Miner (Win):

@vmantilla I'm afraid that i5 3450 is too old. Looks like AVX2 was introduced into Intel procs with the Haswell generation. Ryzen chips have AVX2 as well, but rumor on the internet is that it's not as good as Intel's implementation.

Don't know how much AVX2 is affected by the Bios (+ chip drivers & firmware), but the AMD Ryzen /Threadripper series optimizations are catching fast. So any such deficiencies may be short-lived.

sevencardz

@BeholdMiNuggets Instruction sets are baked into the chip architecture, so I suspect we will have to wait for AM4+ chips to see AVX2 on par with Intel, but who knows what sorts of wizards AMD are...

gumbi17

I just finally got my java working but now I am getting an error offset 88...would this be my path?

{
"Mode" : "pool",
"Server" : "burstcoin.ml",
"Port":8080,

"UpdaterAddr" : burstcoin.ml",
"UpdaterPort": "8080",

"InfoAddr" : "burstcoin.ml",
"InfoPort": "8080",

"EnableProxy": false,
"ProxyPort": 8126,

"Paths":["F:\plots"],
"CacheSize" : 40000,
"ShowMsg" : false,
"ShowUpdates" : false,

"Debug": true,
"UseHDDWakeUp": true,

"SendBestOnly": true,
"TargetDeadline": 80000000,

"UseFastRcv" : false,
"SendInterval": 100,
"UpdateInterval": 950,

"UseLog" : true,
"ShowWinner" : true,
"UseBoost" : true,

"WinSizeX": 76,
"WinSizeY": 60
}

gumbi17

And just an fyi I do have F:\ \ but it appears the forums removes it if you put the \ next to each other.

haitch

@gumbi17

"UpdaterAddr" : burstcoin.ml", - you're missing a " preceding the burstcoin.ml

gumbi17

Thanks...I was staring at every little detail for an hour trying to figure out what I was missing!

haitch

@gumbi17 been there, done that. Sometimes it just takes a second set of eyes to see what's been staring you in the face

BeholdMiNuggets

@haitch said in Blago's Burst-Miner (Win):

.. Sometimes it just takes a second set of eyes to see what's been staring you in the face.

What's the name of that site that checks (simple) code for users, and shows up any syntax errors?

gumbi17

I just use notepad ++, its free and will show you coding errors

haitch

@BeholdMiNuggets https://jsonlint.com/

BeholdMiNuggets

@haitch said in Blago's Burst-Miner (Win):

https://jsonlint.com/

Yeah, that's it. - Thanks Mr Haitch.

Lexicon

hi guys,

ive been having my miner crash over night for the past few days. with the following in the logs. just wondering if theres a bug that causes this to happen.

this is the line just before it bombed out to desktop

12:29:18 * GMI: Sent: POST /burst?requestType=getMiningInfo HTTP/1.0\r\nHost: burst.lexitoshi.uk:8124\r\nContent-Length: 0\r\nConnection: close\r\n\r\n
12:29:18 * GMI: Received: HTTP/1.1 200 OK\r\nDate: Wed, 21 Jun 2017 11:29:20 GMT\r\nConnection: close\r\n\r\n{"requestProcessingTime":0,"baseTarget":"404996","height":"373452","generationSignature":"a4dadfda2d8924685e25cf44f39c5468953077029281271705da3ae91251cc9b","targetDeadline":2592000}
12:29:20 * GMI: Sent: POST /burst?requestType=getMiningInfo HTTP/1.0\r\nHost: burst.lexitoshi.uk:8124\r\nContent-Length: 0\r\nConnection: close\r\n\r\n
12:29:20 * GMI: Received: HTTP/1.1 200 OK\r\nDate: Wed, 21 Jun 2017 11:29:21 GMT\r\nConnection: close\r\n\r\n{"generationSignature":"b7615ccc7868c42ae64f16232719828cb0848249cbc22d562f4617dfcb117df9","baseTarget":"434512","requestProcessingTime":0,"height":"373453","targetDeadline":2592000}
12:29:21 realloc: 1023
12:29:21 realloc: 1352
12:29:21 realloc: 1420

it could be from the show winner part of the application. which makes me think why do we have this feature in.
700 miners. trying to get the winner all from the same wallet servers each block is a bit excessive. and will probably get worse as time goes on.

BeholdMiNuggets

@Lexicon said in Blago's Burst-Miner (Win):

Hi guys, - I've been having my miner crash over night for the past few days. with the following in the logs. just wondering if theres a bug that causes this to happen. It could be from the show winner part of the application. which makes me think why do we have this feature in. > 700 miners. trying to get the winner all from the same wallet servers each block is a bit excessive. and will probably get worse as time goes on.

When the Block Winner (quite frequently) can't be located, I just get an error message, no crash.
WINNER: Connect function failed with error: 10060

Also, can you not just comment-out the winner-polling command lines from your config file?
Eg. (For the Blago Miner), remove: InfoAddr & InfoPort.

Lexicon

@BeholdMiNuggets i get the same however at a random point during the night the miner will crash to desktop. ive removed the lines from my config for now but its better to fix issues than to leave a bug that could potentially wipe out someones miner for a night.

when i say crash. the screen greys out and it says this process has stopped responding. every time its after those 2-3 lines in the logs realloc: ####

also this isnt mentioning that every single miner will be querying the same web wallets on block start.

now lets say we have 13000 miners. each doing 2 api calls to a wallet when the round starts. that 26,000 api calls. in just a few seconds. at the rate at which burst is getting noticed. this will get bigger and bigger.

haitch

@Lexicon Is the realloc statement coming from the miner, or Java itself - possible memory leak issue ?

Lexicon

@haitch its coming from the miner from the looks of it. ive got a few others reporting their miner crashing as well with same issues im facing. lol so i turned on logging and i cant see any other instances where realloc appears except when it crashs

haitch

@Lexicon Yeah, it looks like it's from the miner - but I've never seen it. Possibly you need tweaks to java memory allocation ? -xmx and -xms ?