One K-32 plot in 20 minutes?

Error I get: Cannot run because “VCRUNTIME140_1.dll” is missing

Search for and install " Microsoft Visual C++ 2015 Redistributable " That will install the missing dll files.

I did. Rebooted. Still the same.

Did you use the X64 version, I believe that is the one I downloaded. https://aka.ms/vs/16/release/vc_redist.x64.exe

OK, thanks.
So the thread count set does not matter, is that so? It will max out anyway?

I haven’t tried anything else but max threads since that’s all I’m using my plotter pc for. Others reported the thread function not working and it just maxes out the system. Mine was running all threads and stable overnight.

1 Like

Or fix it yourself, it is open source

1 Like

Test results:

Test done on an i7-8700K at 4.3 GHz (100% core and thread utilisation), phase 1 finished in 55 minutes, on 2 Evo Plus drives as temp1 and temp2.

You seem to not understand how open source software is done. Also, you are being impolite. You want a Windows version, port it yourself instead of nagging here.

1 Like

RAM usage:
adding -b 12000 or -b 3408, all these change nothing.
It only uses between 3 and 4 GB of RAM.

May I ask you what temp drive configuration do you use? just 2 nvmes? Any raid? Full specs of your system would be very helpful also, as I also have the 5900x and plan to try this.

I’m keeping my Core i9-9900K with the usual “parallel” plotting.

If this improves to use better RAM usage as well, I might shift from parallel plotting to this, currently not worth it, unless the only good thing about the plotting machine is the CPU.

I’m running the 5900x, with temp A = “2tb P5 Crucial”, temp B = “1tb Samsung 970 pro”. I have 64 GB ram. One additional minor speed up I added is to leave the destination default to temp A, then use a batch file to copy the plot file to hard drive as needed.

In the long run I might gain 2-3 plots per day vs what I was getting on SWAR, but the best improvement will be if I need to update the computer or do other tasks there will be only one plot to stop vs 12-14 plots in progress.

1 Like

Yeah, I have the same problem. Since I do not want to waste the TBW used for running plots, I usually have to wait 6 hours for all running plots to finish once I stop creating new ones. And that is really annoying. I hope I can stay above 40/day with madmax.

How much RAM does it use when you set it to 24 threads?

It uses very little Memory, could easily get away with 32GB stick, Looking like 6Gb during phase 1.

2 Likes

Reduce the bucket-size to use more memory. Im playing around with this thing on my Phenom ii - Miner and it is ridicoulous fast!
-r 4 -u 128 (==DEFAULT) use ~3.5gb
-r 4 -u 64 use ~7gb
-r 4 -u 32 use ~14GB
-r 8 -u 128 SHOULD use ~7GB
-r 8 -u 64 SHOULD use ~14GB
Thsis is explained in the madmax-repo!

3 Likes

As Madmax and Stotik suggest using a RAM-Disk for Temp2 … im wondering if it is nessesary at all to write temp2 to a (ram)disk at all and if they just could keep the tables in memory all time. Should be 10-100 times Faster without the file-io-overhead…ramdisk or not.

Based on post history we can only conclude that ChiaMiner is confused by the simplest of computing concepts. I don’t think it would behoove us to have this impatient individual contribute anything back to the project :slight_smile:

1 Like

Yes, would be a good optimization, and removed some overhead of the filesystem read/write, I suspect that will come in the future. For now the RAM disk setup is left as an exercise for the reader. Reducing buckets and increasing RAM allocation should help, but you won’t ever hit “zero” at this time.

I just looked at the thread allocation code in stotik’s solution and I think I can see the bug in the take/push_back function of ThreadPool. Need to compare with the madmax repo as it might already be fixed there as stotik’s repo is several commits behind madmax’s repo. If not, I’ll make a fix later today and submit a PR for the threading issue.

4 Likes