Just bought MonkerSolver. I'm unable to do anything as I always get the "out of memory" error.

Apparently there are only 0.4448 GB available. That's what that number means, right?



My computer has 16 GB RAM and no other programmes are running. What's going on?

Can anyone tell me whats the best way to reach Monkersolver is? I mean they are not monitoring this thread anymore but still selling their software. They actually don't even respond to emails either.

Assuming you are on Windows, locate the folder that contains your MonkerSolver.exe executable. From within that folder you will want to open and modify the file MonkerSolver.l4j.ini with a text editor. The value '-Xmx' is the setting that you will need to increase. With 16g of installed RAM, you could try setting it to '-Xmx12g'. Also make sure that you have 64-bit Java installed:

https://www.java.com/en/download/manual.jsp

I had done that of course. It seems to be a problem with my OS because I've set up VirtualBox and Monker runs without any problems (using all the available RAM).

The more information you can provide, the easier it is for us to help you. You still haven't mentioned which OS you are using for example. Monker runs on Windows, Mac and Linux. By your screenshot I'm assuming not Mac, and likely Windows 10.

Making sure that the config file is setup correctly, and that the correct version of Java is installed, is the first step to resolving any initial installation problems. Usually this resolves most issues.

I'm using Windows 10 Pro and the latest 64-bit Java version. My config file looks like this:

-Xms3g
-Xmx15g

I have tried various other values but the available RAM in Monker is always 0.4448 GB.

I've sent you a PM with my Skype info.

Has anyone benchmarked Monkersolver in Unix vs Windows? Does it use less memory or execute faster for any given simulation?

I'm running Monker on Ubuntu but I'm considering migrating back to Windows Server because administering a Linux server is a hassle.

No hard benchmarks. Initially Linux seemed slightly faster to me, but I now believe that Linux and Windows servers are approximately equivalent in speed. The main benefit of running Linux is that it's usually quite a bit cheaper.

What is it that you find a hassle with Linux servers? I have a solution that will make it easier to work with Linux servers that will be released soon. This information will be available publicly at no cost.

Almost right out of the gate with Debian 9 I had a critical error with dbus or xfce that prevented me connecting with vnc. I installed Ubuntu 18.04 and it mostly works. The xfce menu icons are broken/missing and the google drive I mounted with google-ocaml-fuse hangs when I call ls.

For everyone using Monkersolver, time is valuable. Even 1 hour/month of time lost to solving issues with Linux can cost more €40/month for the Windows license. In my case, I am able to get a free license through Microsoft Imagine.

I've installed and maintained Debian 9 Linux on dozens of servers. I've never had any trouble with the xfce icons. Also, I've never lost any server time due to Debian crashing, it's very stable for me. Did you follow the guide for installing your Debian 9 server from the solveoptimized blog?

Of course the point of using Linux is to save money. If you don't have to pay for a Windows license, and you think it will be easier to maintain your server with it, then I would also recommend you to use that.

does anybody know how i'd get a ram estimate for a flop solve for ALL boards? the entire tree which includes preflop requires 7799 GB of ram (which I guess isn't solveable), but i'd like just solve from the flop

You can solve all 1755 flops using the scripting function. How much RAM you need will depend on the flop that requires the most RAM out of those 1755 flops.

Solving all flops in one sim is only possible with a preflop solve. The postflop results you obtain from such a solve however are typically not reliable.

ok thanks. the tree was wrong anyway and would've required a lot more ram than i thought. it looks as if i'm gonna have to stick with specific flops, so a better question, going backwards....: how to apply the same tree for each street while still maintaining different bet caps for each street?

The xfce icons are missing in Ubuntu:

https://gyazo.com/b7487ccd67bcde5b0af94cc6f8c79575

I followed the guide on solveoptimized for Debian 9. After I killed the vncserver and restarted it, I got these error messages when I connected with vnc:

https://gyazo.com/740b23869b8a31582aef8dec4c6b558e

https://gyazo.com/be221ba78a64fe237c8151dea80be9fa

I use Unix CLI on a daily basis, so I didn't realize how much of a noob I was until I tried administering my own server.

I can see your comment on the blog now. I haven't gotten around to reviewing it yet.

I haven't encountered your issue before, but If you want me to help you resolve it, please shoot me a PM with your Skype handle, and I'll go over it with you. Alternatively you will want to try to reinstall your server and just follow the steps from the guide again. It seemed to have worked initially, before you killed the server.

i applied filter for only raise/fold on turn with 3bet cap, and it works, but if someone bets, and it comes back to them capped, their only option is fold.

Does anyone have any idea on which configuration Monkersolver will be faster (both of them aren't real, it is just as an example):
-12 cores @ 2.5GHz
-6 cores @ 5GHz

I have read on previous pages that xeons aren't the best choice. Is that true?

It's not as simple as that, a GHZ is not a GHZ. There are other factors as well that are significant, such as whether you are dealing with a multi-socket setup, how much cache the CPUs have, what sort of memory architecture they support etc. That being said, in your hypothetical example, I would go for 6 cores @ 5 GHz. Monker tends to prefer lower core count and high frequency typically over high core count and low frequency.

Can you please explain why the results would not be considered reliable?
Is there a better alternative to buying preflop solves and solving postflop for users with less than supercomputers?

Thanks!

When solving a preflop game tree, you are solving for convergence of the preflop ranges. The convergence point of the preflop ranges is reached much sooner than the convergence point of the postflop ranges within a preflop game tree, since the preflop ranges are not that sensitive to changes that occur on say a single flop for example. Hypothetically you could keep solving until you reached some sort of convergence for postflop ranges, but this would be entirely uneconomical and in any case less accurate than simply resolving for individual flops. I think the lack of accuracy is the main reason not to even attempt that.

Getting coaching, and learning how to effectively solve preflop yourself might be a better alternative, depending on what sort of games you are playing and what your objectives are.

I purchased some preflop solutions through MonkerViewer. None of the decisions are weighted, but I noticed decisions for certain hands change at random. For example, for the CO opening range A5o, K9o, T9o, K3s, and 64s are sometimes 100% raise and sometimes 100% fold. If you click away from the branch then click back those hands will appear and disappear at random.

Is this by design? Is there a setting I can change? If not, is there any plan to show weighted decisions rather than having them change at random to make it much more intuitive and usable?

Hey all,
I've been using monkersolver for a while but just starting to get into running my own sims. I bought a 6max preflop sim for 100bb, but would like to run some sims for 4max/5max on my own and for some different stack sizes. I feel like I would mess up trying to build my own game trees, so I'm wondering if it would be fine to just take the tree from my 6max sim and give UTG/UTG+MP 'Fold' as the only preflop option (for 5max/4max) and run the sim from there? I can't think of a reason why it wouldn't be ok, but also don't want to attempt it if I would be getting useless results in some way. Is this an OK way to go about it? What about stack sizes, if I take the 100bb game tree but run a 150bb sim with it, does that mess it up somehow? Thank guys!

Thanks for the input. The main reason I was targeting Monker is because I wanted to replace Pokersnowie which has questionable abilites (and is at least terrible MW) as a tool but still retain the same abilty to plug in a Hand History for a marked hand and get a feel for how a machine would equilibrate ranges.

Seems like getting coaching is going to be useful given running poor sims could be worse than no sims.

hey, I just wanted to know is it possible to view someone else's calculated monkersolver solutions with monkerviewer if the other person doesn't have a license key for monkersolver, and is only using monkerviewer?