Looking to hire a lab tech of sorts. I am looking to have some custom trees built and/or purchase some trees. Also would be interested in just some coaching since building trees seem so tedious I imagine I have to be going about it the wrong way.
I’m considering renting a huge server so want to have the trees ready to go. Any expierences with 1tb+ servers would be appreciated as well. Seems AWS/Google are the only real options for 1 month, but seem pretty expensive compared to what I pay now through OVH.
How can I join the Skype group? I have only fond requests for such a group itt, no link to the actual group ...?!
This was the original link posted:
Quote:
Originally Posted by ZenFish
It seems many people are interested in MonkerSolver, but Support isn't very active here so I've created an unofficial Skype group where we can discuss technicalities and effective use of the program:
Everybody is welcome to join. I expect it to be a low volume affair where people can get FAQ answers and help to solve various problems. If the link doesn't work, shoot me a PM and I'll add you.
Trying to use monker solver and the settings tab doesn't seem to show me anything useful with regards to the total ram size needed or buckets. This is what it looks like currently:
It's just a blank screen. how can I get the ram estimate and strength buckets to show up?
Also, it seems that my cpu 16 core 32 thread threadripper is not being used very efficiently with the trial version of this program. Will buying the retail version give me better cpu load? Right now, 16 of my 32 threads are only half used.
If I have a preflop tree that requires, let's say 64gb of RAM to run and I buy the solution from the store or someone else solved it for me, could I view it and how many streets at once at maximum on a computer with 16gb? Or do I always need the same full 64gb to view the full solution? Any general rule or guideline for this?
How can I import a preflop solution into monkerviewer?
You need to load the .mkr file in the 'Save' tab. Select 'load one street' to just view preflop.
Quote:
Originally Posted by PlumpSquirrel
If I have a preflop tree that requires, let's say 64gb of RAM to run and I buy the solution from the store or someone else solved it for me, could I view it and how many streets at once at maximum on a computer with 16gb? Or do I always need the same full 64gb to view the full solution? Any general rule or guideline for this?
You only need the full 64gb if you load all the streets. 8gb of RAM should be enough to load one street of a 64gb solution. I loaded up a 150gb solution on my laptop that has 32gb RAM, and it is using about 8gb of memory.
Thank you very much, it seems like one quarter of the required memory to run is well enough to load preflop tree. Another question, can I start running a tree, then stop the run and save it and then later start running the same tree from the saved version? For example I need to turn off computer for the night, can I just save and continue the next morning and the iterations will increase?
Thank you very much, it seems like one quarter of the required memory to run is well enough to load preflop tree. Another question, can I start running a tree, then stop the run and save it and then later start running the same tree from the saved version? For example I need to turn off computer for the night, can I just save and continue the next morning and the iterations will increase?
Yes, it is possible to save a run and load it at a later point to continue solving it. Just make sure to leave the boxes 'Regret' and 'Compress' ticked when saving compressed files, otherwise you will not be able to continue solving from a compressed file.
Quote:
Originally Posted by doctor877
Whats the easiest way to importing PLO ranges from pf solve to PPT?
From my experience the range output from Monker is too large to import into PPTOO. It leads to a java stack overflow error (I confirmed this with the author of PPTOO). Your best bet would be to try to export to PIO format and use PLOCalc.
but I'm getting warning from Norton antivirus. It says there's some "Heur.Advml.C" and it deletes the file.
Is this the right place to download it?
I believe that is the correct url, yes. For piece of mind, you could try uploading the file to virus total (https://www.virustotal.com/#/home/upload). This will run it through multiple different anti virus algorithms and tell you their verdict.
I am building a new PC to use PLO Monkersolver.
I want to run and look thorugh various simulations, mostly the "easier" ones, like HU pots, postflop.
I am planning with an i7 or i9 CPU and 32-64GB RAM at the moment.
I would like to know if RAM speed is important, if yes, in which way?
I would also like to know if more cores or higher clock is prefered on the CPU?
I am building a new PC to use PLO Monkersolver.
I want to run and look thorugh various simulations, mostly the "easier" ones, like HU pots, postflop.
I am planning with an i7 or i9 CPU and 32-64GB RAM at the moment.
I would like to know if RAM speed is important, if yes, in which way?
I would also like to know if more cores or higher clock is prefered on the CPU?
I hope someone can help me!
Best Regards
OMGPhilGalfond
RAM speed is significant since Monkersolver loads the solve into RAM. Faster access times will mean faster solving times.
Monkersolver doesn't scale that well with cores, so usually faster cores are preferable, however it also depends what exactly you are comparing to. In brief, I would expect cores to win out if the difference is substantial (say 28 vs 10 cores for example), but when it's close, I'd expect the lower core count machine with faster/more effective cores to win out.
I am building a new PC to use PLO Monkersolver.
I want to run and look thorugh various simulations, mostly the "easier" ones, like HU pots, postflop.
I am planning with an i7 or i9 CPU and 32-64GB RAM at the moment.
I was just thinking, I'm not sure I would recommend building a machine with Intel after the recent vulnerabilities were discovered around speculative execution and indirect branch prediction (https://security-center.intel.com/ad...nguageid=en-fr). The vulnerability was patched, and this has lead to certain workloads running slower. From what I've heard, some high frequency traders are taking as much as 3-4% perfomance hit on their workload.
I haven't verified how recent patches might have affected the workload in Monkersolver, but intuitively I would see the type of calculations that Monkersolver makes being affected by this, as they seem very predictable.
You might want to consider a Threadripper buld instead, here is an example build that is optimized for work in Monker:
RAM speed is significant since Monkersolver loads the solve into RAM. Faster access times will mean faster solving times.
Monkersolver doesn't scale that well with cores, so usually faster cores are preferable, however it also depends what exactly you are comparing to. In brief, I would expect cores to win out if the difference is substantial (say 28 vs 10 cores for example), but when it's close, I'd expect the lower core count machine with faster/more effective cores to win out.
Thanks for the help first of all. I thought the CPU is mainly responsible for solving time.
Regarding your second post, I have to work through it, since I don't understand it thoroughly.
The threadripper you posted, do you have any experience with it? Do the 16 cores work well with monker? Where did you find the link or how do you know?
What about the AMD Ryzen Threadripper 1900X 8 core (Octa Core) CPU with 3.80 GHz for half the price e.g.?
Hello guys,
i bought Monkersolver as well. Anyone interested to start a studygroup for Monkerbegginers? Its quite challenging to get it going. PM if interested.