Is there a way to increase CPU usage when mining on Multi Core machine?


#1

I have a duel Intel Xeon CPU machine with 56 threads… ie 56 actual working Cores at my disposal. running at 2200 MHZ. I have 60tb of plotted HHD space on a 5 drive RAID partition… 5 x 7200rpm Exos x12Tb drives. So, as far as HHD goes, its hard to get much better/faster HHDs than that, at least for this purpose… other than SSD drives which is cost prohibitive.

So it seems that my system is still computing even after the deadlines are past. I am operating on CryptoGuru 50/50 presently.

I would like to know the configuration that professional users might prescribe to get better / faster deadlines, and where I am more effectively using my available CPU capability. As of now… I am using only 2 to 3 % of my available CPU capacity. Which, it seems for this kind of operation, we should be able to assign as many cores as we have available to use… or to run multiple wallets, each searching a segment of the plotted space.

Is there a way to more effectively utilize my available computing power than with my current 1 wallet / 1 miner setup, so I get faster results? I had thought of virtual machine setup… each running a portion of the plotted space. But having such limited knowledge of the current setup… I thought it better to consult others experiences on this topic.

I currently use my computer for occasional CAD design projects, otherwise I have it on all the time for storage of my 20 tb movie archive for my own and my families entertainment. Otherwise I use it for hobby experimentation such as this Burst Project. My goal is to get burst to at least cover my electricity costs in earnings capabilities… even if I don’t actually cash/out my burst coins.

That way I can at least support this Burst Crypto project as needed by the developers and community. As my computer is always on anyway. And who knows, maybe BurstCoin will eventually turn into something worthwhile.


#2

First, I’ll start off by saying there is no way to get better or faster deadlines except for adding more space to your plot files, which increases your chances. You can, however, decrease your scan time, which helps with the cg pool algorithm.

The number of threads you have can really run a lot of hard drives for Burst mining.

Since you have this system set up in current configuration for other things as well, I won’t tell you to break down your raid array into individual drives. However, keep in mind in the future, that this is the best way to set up your drives.

You didn’t say how your plot(s) are configured, so perhaps you just have one plot file? If this is the case, you are using 1 cpu thread to scan this. The miner can scan plot files in parallel per cpu thread. So, say you wanted to use 50 threads. I would have 50 plot files. Take the space you want to use and split it up into 50 equal parts. The miner would then scan each file at the same time…using more of your cpu and scanning faster. All of this is depending on if your array throughput can handle all of this at the same time.


#3

keep adding Allan you’ll find if you tripple your current capacity it will use more cpu 56 drives would use 56 threads at once!!!


#4

What about 56 folders, on the same drive? Each folder using a separate path line.
From what I have garnered… I could simply divide the plot files… there are 60 of them, each 1gb, and put each one of them in its own folder. Then import each one of them individually into the miner where it will have a separate path for each. Will this work?

The problem I had with single drives… is that they would go to sleep, no matter what I did… even having that KeepAlive program running didn’t solve the problem with the drives going to sleep. It helped a lot to stop that when I was only dealing with one drive… rather than 5 drives. There must be a solution for that? I have tried the Power properties… still happens.

Anyway… I heard that’s how Jminer works. Except it uses a faster GPU cores. But from what I have learned, I could use the GPU with Jminer, and Blago for the Cpu’s. And run duel miners?

Main thing is… I don’t want my machine sitting around doing nothing. Which is what is happening now.


#5

Ok… I added each line to a separate folder… and it seems to work about twice as fast… still only 350 to 297 MB/s. Now its only using 1 to 3 percent of the CPU capacity during a mining run. Although, all cores are working… but barely a blip on the Task Manager Performance Core Usage screens.

It seems like I am not making as many conformation’s as when it was in a single folder format.

WISH LIST: Wish the Blago Miner window was fully adjustable. Needs to be a lot bigger with scrolling features so its possible to analyze the screen output easier.

I really would like to try out Jminer and see if my GPU will speed things up a bit. If anyone can supply me with a config file they are using with CryptoGuru 50/50 pool.


#6

Just throwing this out there…it could be a bottleneck elsewhere, not the cpu. I have an i5-2400 that does more than that mb/s on a miner. Trying to simultaneously push all the data even through your raid array all at once could be where it’s happening. Just a guess.


#7

Hi Ryan…I did break down my RAID and I am getting half the speed as before… also the other issue of the hard drives going to sleep… at least that is what it says in Blago miner. Which I don’t think it is the hard drives going to sleep at all. As I have KeepAlive running… and it is blinking the hard drives ever couple of seconds. In addition to (for testing purposes) writing a file on each drive each 2nd second.

I think the issue is with Blago software. See the clip attached.
Also, I would like to put it out there for the developers, can you make your software operate in such a way where I can have a window open for each hard drive, where I can assign the specific (processor) or core to run that window. So that would require labeling each window as 1/2/3/4 progressively. That would allow those who have multi-processor (core) machines to operate faster… as it doesn’t seem to be as effective to run them all in the same window. Though… I will be testing out this theory and see if each hard drive finishes running in 16.3 seconds as the first one is. Each hard drive is 12Tb at 7200 rpm. Still they seem kind of slow. I am looking for ways to speed the drives access time up.

Main problem now is blago not starting, and displaying HDD, Wake up.

When I had the RAID configuration, it rarely did this.

Thank you for taking your time to work this problem with me… in advance.

Leslieallen


#8

they aren’t necessarily going to sleep it does that regardless to ensure they don’t go to sleep.
You can mess with cache settings may improve efficiency
Process affinity there is a command line syntax for that in windows I think but I don’t know it.
You can run several copies of Blago on single system just doesn’t show capacity on the pool.